Re: Tape throughput troubleshooting?
Hot Diggety! Richard Sims was rumored to have written: I would start with the netstat inspection outlined in the Backup performance topic of ADSM QuickFacts, which will quickly assess how network flow is doing in the actual event. TCP window size may be involved in what you are seeing, known to be a particular factor over long distance communication. I regularly push up to 980 Mbps sustained over a nearly 3,600 mile (~5800 km) WAN, but usually about 175-200 Mbps for a typical workload. Richard is correct: TCP window size tuning is *vital* for a high bandwidth WAN. Don't leave home without it correctly tuned. :-) For instance, according to Sun support working one of our cases, Solaris is tuned for LAN performance out of the box. It's not a lot more to tune it for a WAN, but it does require some careful notes, theory, measurement, and trial-and-error. We also got a major boost in performance merely by enabling jumbo frames on each end. A smaller boost from using certain tunables to push the NICs to their absolute maximum on Solaris. I haven't done AIX perf tuning in so long that I'm a little rusty now, alas. But for AIX, one tunes sliding windows by adjusting these 'no' options: - tcp_sendspace - tcp_recvspace There's a few more 'no' options... rfc1323 is another key one to enable. http://www.performancewiki.com/aix-tuning.html Regardless of the sliding windows values chosen, strongly suggest testing and retesting with various (power-of-2) values until the optimal value is seen. E.g. 1024, 2048, ..., 65536, etc. Also, the original poster may want to check with the network folks (and/or upstream provider if appropriate) to make sure there's no particular segment along the end-to-end network path that is of small bandwidth. In my case, it's all 'in-house' so that makes it easier. E.g. you're not going to see gigabit performance end-to-end if you have, say, a DS3 (45 Mbps) circuit somewhere in between. Also, before making changes, record before/after values as well as results of measurements. Nothing worse than getting muddled in the perf tuning process due to lack of careful record-keeping. Some food for thought. -Dan
Re: How do I get TDPO to send controlfiles to SBT_TAPE instead of DISK?
Hot Diggety! Dan Foster was rumored to have written: My DBAs and I are completely baffled. How do I get TDPO to send controlfiles to SBT_TAPE instead of DISK? Eventually resolved: needed to get *every* single config file on BOTH hosts to match for a restore to the DR host, AND to disable PASSWordaccess GENERATE. After that, went swimmingly, and the DBAs not familiar with RMAN/TDPO were impressed. We then set up a RMAN catalog db, which worked great, too. -Dan
Re: How do I get TDPO to send controlfiles to SBT_TAPE instead of DISK?
Hot Diggety! Steve Stackwick was rumored to have written: On Thu, Apr 10, 2008 at 3:26 PM, Dan Foster [EMAIL PROTECTED] wrote: My setup: TSM server (Enterprise) 5.5 on AIX 5.2. TDPO 5.4.1 on Solaris 10/SPARC and Oracle 10gR2. Do you really have TSM 5.5 running on AIX 5.2? I wondered if that would work, and kind of thought it would, but Tivoli sez u need AIX 5.3. Just wondering. There's two kinds of 'work': - 'probably technically works but you're on your own' vs. - '20 foot wall of flames will erupt if you attempt it' IBM's statement in this particular case is more of the former, presumably partly to reduce costs in regression testing (amongst many other valid reasons). And, of course, if you hit a bug with such a setup, they're unlikely to fix it or even give the time of the day. The $65,536 question: Why did I put 5.5 on AIX 5.2? Simple: I was a little too hasty in reading the min requirements, a rare departure for my character. :) I always, with that one glaring exception, stick to supported revs for everything. Why not backout? A lengthy maint moratorium amongst other things in my environment made it more practical to just stay put. With that said: # lslpp -l tivoli.tsm.server.aix5.rte64 Fileset Level State Description Path: /usr/lib/objrepos tivoli.tsm.server.aix5.rte64 5.5.0.0 COMMITTED IBM Tivoli Storage Manager 64 bit Server Runtime Path: /etc/objrepos tivoli.tsm.server.aix5.rte64 5.5.0.0 COMMITTED IBM Tivoli Storage Manager 64 bit Server Runtime # oslevel 5.2.0.0 (It's actually at 5.2 ML9 + some patches) I haven't seen any issues so far, though I would not recommend doing this to anyone except the truly desperate. It's by far better to stay supported for the server side. You can usually get away with downrev clients past the officially supported revs, but server [and support] is too important to risk for most folks. Then there's that TSM server revs may have some relation to device support of the underlying AIX OS, particularly for newer devices. In my case, I hadn't changed anything else, so got by with that. I'm looking to bring the server up to a newer AIX version soon to resolve this uncomfortable position. Last, but not the least, I've also been duly thwacked by colleagues with multiple copies of the old ADSM 3.1 printed manuals to serve as a reminder to be more careful and diligent in my reading. :) Sure deserved that. -Dan
How do I get TDPO to send controlfiles to SBT_TAPE instead of DISK?
My DBAs and I are completely baffled. How do I get TDPO to send controlfiles to SBT_TAPE instead of DISK? My setup: TSM server (Enterprise) 5.5 on AIX 5.2. TDPO 5.4.1 on Solaris 10/SPARC and Oracle 10gR2. RMAN settings: RMAN show all; RMAN configuration parameters are: CONFIGURE RETENTION POLICY TO RECOVERY WINDOW OF 7 DAYS; CONFIGURE BACKUP OPTIMIZATION OFF; CONFIGURE DEFAULT DEVICE TYPE TO 'SBT_TAPE'; CONFIGURE CONTROLFILE AUTOBACKUP ON; CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE SBT_TAPE TO '%F'; # default CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '%F'; # default CONFIGURE DEVICE TYPE 'SBT_TAPE' PARALLELISM 3 BACKUP TYPE TO BACKUPSET; CONFIGURE DEVICE TYPE DISK PARALLELISM 1 BACKUP TYPE TO BACKUPSET; # default CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE SBT_TAPE TO 1; # default CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE DISK TO 1; # default CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE SBT_TAPE TO 1; # default CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE DISK TO 1; # default CONFIGURE CHANNEL DEVICE TYPE 'SBT_TAPE' PARMS ''; CONFIGURE MAXSETSIZE TO UNLIMITED; # default CONFIGURE ENCRYPTION FOR DATABASE OFF; # default CONFIGURE ENCRYPTION ALGORITHM 'AES128'; # default CONFIGURE ARCHIVELOG DELETION POLICY TO NONE; # default CONFIGURE SNAPSHOT CONTROLFILE NAME TO '/opt/oracle10/product/10.2.0/db_1/dbs/snapcf_mydb.f'; # default How we backup -- RMAN commands: allocate channel t1 type 'SBT_TAPE' parms 'ENV=(TDPO_OPTFILE=/opt/tivoli/tsm/client/oracle/bin64/tdpo.opt)'; allocate channel t2 type 'SBT_TAPE' parms 'ENV=(TDPO_OPTFILE=/opt/tivoli/tsm/client/oracle/bin64/tdpo.opt)'; allocate channel t3 type 'SBT_TAPE' parms 'ENV=(TDPO_OPTFILE=/opt/tivoli/tsm/client/oracle/bin64/tdpo.opt)'; sql 'alter system archive log current'; backup incremental level = ${LEVEL} filesperset 5 format df_%d_%s_%p_%t.lv${LEVEL} (database include current controlfile); backup filesperset 20 format ar_%d_%s_%p_%t.lv${LEVEL} (archivelog all delete all input); release channel t1; release channel t2; release channel t3; tdpo.conf for the restore on a test box: DSMI_ORC_CONFIG /opt/tivoli/tsm/client/oracle/bin64/dsm.opt DSMI_LOG/var/log/tsm TDPO_FS prod-oracle TDPO_MGMT_CLASS_2 tsmoracle-mgmt2 TDPO_MGMT_CLASS_3 tsmoracle-mgmt3 TDPO_MGMT_CLASS_4 tsmoracle-mgmt4 Not a fancy setup; pretty much 'stock' (standard). Thoughts/ideas? -Dan
Re: Is there a simple way to test a tape drive?
Hot Diggety! Richard Rhodes was rumored to have written: Sometimes I want to have TSM test out a tape drive. Something like . . . - pick a drive - pick a tape - have tsm mount the tape and read the label (or something) - tell me this succeeded or not Like yesterday . . . IBM repaired a tape drive. After fixing it, IBM tells us it's fixed. We put the drive online . . .and wait for TSM to use it. For this particular drive, it was several hours before TSM used it. We also get this after an upgrade, or adding a new drive. To get TSm to used any drive, let alone a particular drive, requires starting a migration, update stgpool, backup to tape, etc. All are large processes. Is there a simple way to get TSM to exercise a particular tape drive? Or, what do you do in these situations? Well, we run a r/w test with a known good scratch tape before letting the IBM CE leave the site. :) It's not perfect but it's a quick test to test the major functionality. You didn't mention what OS and version the server runs nor the TSM server version. So this will vary a bit, but for AIX and 3584/LTO: # tapeutil -f /dev/rmtX rwtest -b 262144 -c 10 -r 2 is good with our LTO-1 drives in our 3584 library. Block size: 256KB, 10 blocks (of 256K each), 2 runs, read and write test. Obviously, make sure it's a scratch tape, since this test will scribble some data over it. You may want to have this tape temporarily checked out of TSM so that TSM won't try to use it for any operation. Some other platforms may have tapeutil; I'm not sure what platforms has it and if it's named differently on some platforms or not. Other things you can do: # tapeutil -f /dev/smc0 inventory | grep -p ^Drive Address num ...where num corresponds to the element number of the drive. (AIX supports grep -p; on other platforms, just use 'more' instead.) You can also test mount/dismount of tape by moving the scratch tape from either a library slot or from a I/O slot. # tapeutil -f /dev/rmtX move elem ID #1 elem ID #2 ...where elem ID #1 is the element number of your tape source location ...and elem ID #2 is the element number of your tape drive And, of course, rmtX is the drive in question (rmt0, rmt1, etc). When done, can do 'tapeutil -f /dev/rmtX unload' and then 'tapeutil -f /dev/rmtX move elem ID #2 elem ID #1'. The tapeutil rwtest is best done with the drive disabled in TSM *FIRST*, so that TSM doesn't try to use the drive at the same time. Once you're finished with the work, bring the drive back online in TSM ('upd drive ...'). Then you can do 'sh libr' and look at the output; should look sane. Doing all this only takes a minute or two, so I don't feel bad about asking the CE to wait. I believe it's mtlib for 3494 libraries. tapeutil AND mtlib usage can be found here: ftp://ftp.software.ibm.com/storage/devdrvr/Doc/IBM_Tape_Driver_IUG.pdf -Dan
Re: Nervous Nellies and TDPO -- feasible?
Hi ladies and gentlemen, I *really* appreciate the feedback including one by private e-mail! It's helped a lot in resolving our DBAs' concerns. Yesterday, we configured RMAN/TDPO (our first TDPO setup) and I must say the performance flies with use of multiple parallelized streams! I'd only previously set up TDP-MSSQL (which is pretty good, too.) The point regarding continuing to use some form of additional backups outside of TDPO for 'insurance' purposes is a good one, and we will continue to do so. The Redbooks tip was a good one. I've worked with IBM gear for over 10 years but for some reason, hadn't occurred to me to look there. I guess I'm too used to knowing the stuff that I normally need to know. :-) This database isn't on storage that allows for BCV/FlashCopy type snapshots, unfortunately, or we'd leverage that. A pretty good idea, though. (We have various EMC, IBM, and Hitachi enterprise disk storage but not for this particular Oracle database -- historical reasons partly tied in to fairly messy politics.) I think we're on the right track now, and thanks again for your assistance! Cheers, -Dan
Re: HELP!! TIPS on freeing up library slots
Hot Diggety! Jim Young was rumored to have written: This will give you a list of tapes under twenty percent utilized from all pools select volume_name, stgpool_name,pct_utilized, status from volumes - where pct_utilized 20 and status='FILLING' - order by pct_utilized, stgpool_name, volume_name After that, do a MOVE DATA VOLUMENAME to help it along a bit. Does anyone know whay TSM has this issue with keeping lots of 'filling' volumes? Could be a collocated setup? That's one way you get lots of individual tapes, each filling (relatively) slowly? -Dan
Nervous Nellies and TDPO -- feasible?
Hi - My DBAs has some serious concerns about running TDPO and I don't know enough about it to be able to answer their concerns authoritatively. Any information would be much appreciated. The subject line was _NOT_ meant to denigrate them -- not at all (they're clued in DBAs) but meant to get your attention, as I'm sufficiently desperate. Sorry. :) Setup: - TDPO 5.4 (Solaris 10 Update 3 + patches) - Oracle 10gR2 + patches - TSM server 5.4 (AIX 5.2 TL 10 + patches) - Plans to go to TDPO 5.5 + TSM server 5.5 shortly Their concerns: 1) RMAN doesn't deal with non-transactional DB data loads well. (i.e. data loaded not through the usual INSERT/UPDATE methods, but done via sqlldr. This doesn't generate redo logs.) True or false? 2) RMAN requires a recovery database be created to do recoveries to (instead of recovering to original database/tablespaces in-place). True or false? I thought RMAN didn't need an intermediate temp database and could restore directly 'in-place'. 3) Will RMAN guarantee a good point-in-time view of DB data? 4) Same as #3 above, but what if the DB is currently processing a large series of INSERTs or UPDATEs or sqlldr run, will RMAN only send data that was present at the moment of backup start? 5) Does RMAN/TDPO use scale to really large DB setups? Say, tens or hundreds of terabytes? Or even sub-10 TB? Anybody using TDPO well with terabytes of Oracle data? (I think they're hoping for assurances that it really does work in the real world at that scale.) 6) How does Oracle RMAN know what rows to send for an incremental backup? Does Oracle maintain a bitmap of some kind or time-based logging of changed blocks or something? Any pointers to whitepapers or information on how it works behind the scenes? They would prefer to utilize Veritas DB Edition (DBED) checkpoint feature which is where a script runs to put all tablespaces in backup mode, brings DB-based filesystems to a consistent state, generate a checkpoint (snapshot-like) of the filesystem, then brings all tablespaces back to online mode. The TSM BA client then mounts the most recent checkpoint R/O then backs it up at the file level. The problem with the above approach is this essentially results in what amounts to a full backup every single time it kicks off. That's just not practical in terms of tape cost as well as the length of backups. I see TDPO as the only practical choice for much faster and more frequent backups (and restores) and is guaranteed for consistent point-in-time views of DB data. The BA client was also not meant to back up databases (though it *could* do so if properly quiescied). My issue? I don't have enough meaty information yet to put concerns to rest, and I'd really like to make use of TDPO after paying out $16,000. Any info would be much welcome. Thanks, -Dan
How to register TDP Oracle 5.4.1 licenses?
Setup: - TSM 5.4.1.1 server on IBM 7026-6H1 running AIX 5.2 TL10 SP3 - TDPO 5.4.1 client on Sun T2000/SPARC running Solaris 10 Update 3 - Oracle 10gR2 database - version 10.2.0.3 I've only previously set up TDP MS SQL 5.2? on a TSM 5.1.9.6 server. We recently went to TSM and TDP 5.4 and had not previously registered an Oracle TDP client. I'm now finding out that you can't use REGISTER LICENSE FILE=oracle.lic with TSM/TDP 5.4. It's not completely clear how I get this registered. Do I just simply go through a normal TDPO installation on the client, and let tdpoconf or REGISTER NODE register the client on the TSM server? Is there something special I need to do to make TDPO licenses show up on the TSM server? Do I need to fetch license enrollment certificate files from somewhere? Any insight leading to enlightenment would be most welcome. :-) Yours in bafflement; may your tapes never oxidize, -Dan
Re: Minimum TSM server version for Solaris/x86 client?
Hot Diggety! David McClelland was rumored to have written: Do you mean client or server here? TSM *client* on x86 Solaris has been available since 5.3: Ah, right; duly corrected, thanks! Given that TSM 5.2 Server goes out of support next month, you'll only really want to be connecting it to a 5.3 or 5.4 server anyway for which I wouldn't see there being any problems. I eventually managed to stay connected with the FTP server long enough to fetch the 5.3 and 5.4 clients and found out they do appear to work with the older TSM server (5.1.9.6). Needs are very basic; nothing fancy so I'm not too concerned for deploying this on a small subset of our kits on an as-needed basis. Buys me some more time while I continue to see if we can either get an upgraded TSM server or if we can continue to run TSM instead of the preferred organisational standard of EMC [Legato] Networker. I don't disagree about the Apple II performance of the ftp.software.ibm.com site, although I prefer to liken it to my Acorn BBC Micro Model B (I don't think they ever made it over to the US) reading at 1200 baud from cassette tape :o) :-) Heard much about the Acorn BBC kits over the years. :) Cheers, -Dan
Minimum TSM server version for Solaris/x86 client?
TSM 5.4 added support for Solaris/x86... but what is the minimum server version it will connect to? 5.1.5? 5.2? 5.3? 5.4? I looked in the documentation and couldn't find that info. The FTP server keeps disconnecting me while emulating the performance of an Apple II running off floppies so I haven't been able to get the client to test with just yet, either. Any pointers or information would be MUCH appreciated. Thanks!! -Dan
Re: TSM Client on GENTOO
Hot Diggety! Christian Svensson was rumored to have written: Has anyone try to install ITSM BA/API on Gentoo? Yes. The 5.1.5 client works well for me. I have not tried 5.2 or later. I can not even install ITSM on it because I don't have any RPM software. # emerge rpm # rpm -ivh --force Linux/x86 TSM client file.rpm or something like that. Then after that, just do normal dsmc client setup. (dsm.opt, dsm.sys, add to inittab, pre-seed the password, etc.) -Dan
Re: 3584 Library
Hot Diggety! Gill, Geoffrey L. was rumored to have written: I'm bringing up a remote AIX system with a 3584. Seems as though IBM left it in an unusable state and I'm trying to figure out why. What I originally saw was the below output minus the 3584 info. I ran cfgmgr and now have the below output. What is odd to me is that there are 3 3584's showing up. Can anyone shed some light on the library or tape drive issue? # lsdev -Cc tape rmt0 Available 40-60-00-5,0 SCSI 4mm Tape Drive rmt1 Available 14-08-02 Other FC SCSI Tape Drive rmt2 Available 17-08-01 Other FC SCSI Tape Drive rmt3 Available 54-08-01 Other FC SCSI Tape Drive rmt4 Defined 1A-08-02 Other FC SCSI Tape Drive rmt5 Defined 21-08-01 Other FC SCSI Tape Drive rmt6 Defined 5A-08-01 Other FC SCSI Tape Drive smc0 Available 14-08-02 IBM 3584 Library Medium Changer (FCP) smc1 Available 17-08-01 IBM 3584 Library Medium Changer (FCP) smc2 Available 54-08-01 IBM 3584 Library Medium Changer (FCP) A possibility is that there were multiple control paths defined in the 3584 library. This can be adjusted via the 3584 front panel LCD, or I think, also the web interface (though I never used it much due to the infamous 3584 ethernet issues). The 'Other FC SCSI...' looks odd, though. Does the system have Atape installed? And the drives are IBM LTO drives? Could be that rmt4-6 was seen before additional control paths were defined, because once you have additional CPs, your view of the drives contained in them can be blocked. Ultimately, I think you want to determine if the 3 CPs is what you want. If not, needs to be set back to the default one CP for entire library. You'll also want to figure out why the LTO tape drives aren't showing up right if they're IBM LTO tape drives. Another thing to triple check with the 3584, in my experience, is to ensure the drives, library, and FC adapter microcode are all at the latest available and stable revision. If you're still stuck, even after adjusting all these as desired, then please describe the 3584 setup (1 L32 and how many D32s, how many drives, what type of drives and their mfr, microcode revs for library, drives, and FC adapter, what FC adapter, what host type/model, what host OS and version is running, level of Atape installed, etc. -Dan
Re: adm login is locked!
Hot Diggety! Alexandr was rumored to have written: I`ve installed TSM sever and client (full pack) on the same server(AIX). Have registred client node on the server TSM.(input correct pwd for registred client node). All it was working well for long time. But after long off-working time was tring connect to tivoli-server under admin-login and have got message: ..ANS8056E Your administrator ID is locked. ANS8023E Unable to establish session with server. ANS8002I Highest return code was 61. How unlocked administrator ID? May be exist any method of restore access to server without re-install server? You do not need to reinstall the TSM server. What TSM server version are you using? Do you have any other administrator IDs? If yes: login to another admin ID then do in dsmadmc: tsm unlock administrator your locked admin ID If no: login to TSM server, and use UNIX shell to do: # kill -HUP dsmserv pid sleep 5 \ kill -KILL dsmserv pid # DSMSERV_DIR=/usr/tivoli/tsm/server/bin # DSMSERV_CONFIG=/usr/tivoli/tsm/server/bin/dsmserv.opt # cd /usr/tivoli/tsm/server/bin # ./dsmserv tsm disable sessions tsm unlock admin your locked admin ID tsm halt # nohup /etc/rc.adsmserv Then wait for the TSM server to come back up. You will be able to login in dsmadmc when it is back up. Then investigate to see why admin ID was locked. Did it expire? tsm query admin your admin ID f=d Example: tsm query admin alex f=d Your account was locked due to either: 1. Exceeding maximum number of invalid admin sign-ons or 2. Your password expired If your password was expired, then either set a longer password expiry period or change your password more often. -Dan
Re: Trying to use TSM API
Hot Diggety! Mike was rumored to have written: I have a program that compiles properly, and connects, but when querying (I'm testing managemt classes first) it gets an error: ANS0245E (RC2065) The caller's structure version is different than the TSM library version. The version of the tivoli.tsm.samples is the same as the tivoli.tsm.client. Is there a way to check what is wrong? Is there a way around the problem? What version is the API, client, and what platform? -Dan
Re: Trying to use TSM API
Hot Diggety! Mike was rumored to have written: tivoli.tsm.client.api.32bit 5.3.2.2 COMMITTED TSM Client - Application Programming Interface 32bit That looks correct. Do you have VisualAge C/C++ installed? It apparently needs the C++ compiler, and specifically, VisualAge's. Did you copy /usr/tivoli/tsm/client/api/bin/libApiDS.a to /usr/lib? Do you have a copy of /usr/tivoli/tsm/client/api/bin/sample/*.h in the same directory as your app? Or have you copied and adjusted /usr/tivoli/tsm/client/api/bin/sample/Makefile to directory of your app and adjusted CFLAGS's -I. to -I/usr/tivoli/tsm/client/api/bin/sample ? -Dan
Re: Trying to use TSM API
Hot Diggety! Mike was rumored to have written: The file (tsm.so) compiles great and connects to the tsm server just fine. Only when I issue the management class query does it complain about the structure version number. I am using vac/cc. No need to copy the library to /usr/lib since the extension compiles and connects without error. Well, the reason why I mention it is because this version error indicates a mismatch against library and header files. Meaning, it's possible to compile fine but fail at the run-time check. Besides, I think the makefile is specifically looking for libApiDS.a in /usr/lib; take a look at the makefile. Makefile is also looking for header files in current directory by default, too, I think. Next step would be to verify that /usr/lib/libApiDS.a and /usr/tivoli/tsm/client/api/bin/libApiDS.a are identical, possibly by running md5, md5sum, or some such utility. Then compare time/datestamps, and so forth, for both libApiDS.a *and* the header files. You could be picking up an older version of either from somewhere unexpected or forgotten about. You've got the right package installed, but the error is indicating something is out of sync between the library and header files. So you need to identify and find where the older/incorrect stuff is, then correct it, then recompile. -Dan
Re: ./domdsmc: Error while loading shared libraries
Hot Diggety! TSM User was rumored to have written: ./domdsmc: Error while loading shared libraries: libnotes.so: cannot open shared object file: No such file or direcotry. I made sure that libnotes.so exists in the correct direcroty that I specified while running dominstall. I am having this same problem with SUSE 9. May I suggest a much shorter subject line capped to 60 characters or less? I've shortened it for you. Shared libraries in Linux is not loaded from the current directory. It is loaded from any directory listed in /etc/ld.so.conf. You will need to research to find out how your Linux distribution handles updating of library paths because it differs amongst each distribution. I'm not familiar with SuSE 9 or RHEL 3, but it should be Googleable, in distro vendor's docs, or someone here likely knows. In Gentoo Linux, updating ld.so.conf is done by adding a file in /etc/env.d/ with a path to add, then doing 'env-update'. This procedure is different with various Linux distributions. In a hurry, you could always directly edit /etc/ld.so.conf to add the directory that holds these shared libraries for Domino, then do 'ldconfig' as root. Then retry using the utility. -Dan
Re: SQL Select commands
Hot Diggety! Richard Sims was rumored to have written: Go to http://www.rz.uni-karlsruhe.de/rz/docs/TSM/WORKSHOPS/3rd/handouts/ Handouts I of Andrew Raibeck is the online version. Handouts II of Andrew Raibeck is the PostScript version, Very nice stuff, that's for sure (as are the others' presentations). which is trivially converted to PDF by opening on a Mac, where the Preview application takes care of it. Or on a UNIX machine, can be converted via the ps2pdf utility. E.g.: $ ps2pdf raibeck.ps raibeck.pdf $ acroread raibeck.pdf (or xpdf raibeck.pdf, or transported to a Windows/Mac machine to view.) A nice side effect of this conversion is that it shrinks the file size for PDF by about 5.5 times without any apparent visual degradation. :-) Can be fetched from: http://www.cups.org/espgs/software.php I got my copy from my Linux distribution's package repository, so it was a mindlessly quick fetch and install. -Dan
Re: backup performance
Hot Diggety! Mario Behring was rumored to have written: I´ve started a backup operation at a Linux client using CENT OS (similar to RHat). The operation took 1 hour and 59 minutes to finish, and backed up 5.45GB of data.I think this is kind of slow..considering that the LTO3 tape unit is supposed to be very fast Can you post the end of job statistics? For example, in my dsmsched.log, I have: 05/18/06 22:32:33 --- SCHEDULEREC STATUS BEGIN 05/18/06 22:32:33 Total number of objects inspected: 845,061 05/18/06 22:32:33 Total number of objects backed up: 26,665 05/18/06 22:32:33 Total number of objects updated: 0 05/18/06 22:32:33 Total number of objects rebound: 0 05/18/06 22:32:33 Total number of objects deleted: 0 05/18/06 22:32:33 Total number of objects expired: 1,260 05/18/06 22:32:33 Total number of objects failed: 0 05/18/06 22:32:33 Total number of bytes transferred: 3.11 GB 05/18/06 22:32:33 Data transfer time: 464.32 sec 05/18/06 22:32:33 Network data transfer rate:7,040.43 KB/sec 05/18/06 22:32:33 Aggregate data transfer rate: 2,600.08 KB/sec 05/18/06 22:32:33 Objects compressed by:0% 05/18/06 22:32:33 Elapsed processing time: 00:20:57 05/18/06 22:32:33 --- SCHEDULEREC STATUS END If you can post yours for that backup run, it would help give some insight. Also, do you use a disk pool on the TSM server? For example: TSM client - TSM server (diskpool) - TSM server (tapes) If you do not use a diskpool, you will not be able to feed data to the LTO-3 tape drives fast enough. If that happens, it will do start-and-stop which dramatically slows down performance. -Dan
TDP oddity with an extra registered client?
I've got: TSM 5.1.9.6 server on AIX 5.2-ML7 TDP 5.2 for MSSQL (Windows 2003 Server running MS SQL Server 2000) The strange thing is: tsm: SERVERq lic [...] Number of TDP for MS SQL Server in use: 3 [...] But I only have two TDP 5.2 clients. How can I determine a list of all 3 registered TDP clients? And how can I delete the offending client? Or does the TSM server count as a TDP client for licensing purposes? Any suggestions or insight welcome. Thanks! -Dan
Re: TDP oddity with an extra registered client?
Hot Diggety! Stef Coene was rumored to have written: tsm: SERVERq lic [...] Number of TDP for MS SQL Server in use: 3 [...] But I only have two TDP 5.2 clients. How can I determine a list of all 3 registered TDP clients? select LICENSE_NAME, NODE_NAME from LICENSE_DETAILS Ah! Thanks. As I suspected, the problem was due to me making a silly configuration file error without thinking when I first installed it on a TDP client. (It registered under the wrong name.) I changed config file to use the proper name immediately after I realized my error... but too late, wrong name was already registered in TSM. Does anybody know how I can delete the TDP registration for the incorrectly registered client? tsm select LICENSE_NAME, NODE_NAME from LICENSE_DETAILS MSSQLVN01-DB MSSQLVN02-DB MSSQLW3USPHX1 [...] VN01 and VN02 is correct. W3USPHX1 was the mistake. tsm: GBLX-PHXselect * from license_details where node_name='W3USPHX1' LICENSE_NAME NODE_NAME LAST_USED TRYBUY -- -- -- MSSQLW3USPHX1 2006-04-06 FALSE 01:31:02.00 MGSYSLAN W3USPHX1 2006-05-13 FALSE 19:41:46.00 So I need to delete MSSQL license registration for W3USPHX1 only, but leave the MGSYSLAN license registration alone. Is there a way to do that without having to delete W3USPHX1 entirely then recreate and re-backup its TSM data? -Dan
Re: dsmadmc using non-root id access
Hot Diggety! Ameerul Mazli was rumored to have written: Has anyone got any idea whether it is possible to execute dsmadmc with non-root id? Sure, no problem. You only need root privileges to start TSM and possibly for a few things like creating/defining db/log volumes? (May not even need root for these if the filesystem has appropriate permissions?) It is possible to set up sudo so that an unprivileged TSM user account can only start the TSM server with sudo-granted root privileges, and you get the benefits of logging, too. Can we use a e.g.tsmadm UNIX id and runs dsmadmc? Will there be any 'side effects'? No side effects that I've seen. That's how I run my TSM jobs, via an unprivileged UNIX user called 'tapeadm' and it uses a privileged TSM user/pass so I can tell the difference between human administrators and TSM jobs in the activity logs. Works great. -Dan
Re: Cancel all processes
Hot Diggety! MC Matt Cooper (2838) was rumored to have written: I am not aware of a cancel all processes. However, you can write a script to do it. It would be based on the fact that TSM will tell you all the processes that are running.select process_num from processes will give you the process_num numbers that need to be canceled One caveat, from the 5.1 help information: : Some processes, such as reclamation, will generate mount requests in : order to complete processing. If a process has a pending mount request, : the process may not respond to a CANCEL PROCESS command until the mount : request has been answered or cancelled by using either the REPLY or : CANCEL REQUEST command, or by timing out. : Use the QUERY REQUEST command to list open requests, or query the : activity log to determine if a given process has a pending mount : request. So a simple q proc (or equivalent) + canc proc num operation may usually work most of the time, but may fail at certain times if you don't also check for this, too. -Dan
Re: A quick question
Hot Diggety! David Longo was rumored to have written: I have used only IBM LTO1 tapes in my 3584 Fibre drives. I did have a few (3 or 4 in a year or so.) As I slowly gathered over this time, there were two problems. 1. The cases on early LTO 1 tapes didn't have the halves welded together. Therefore if when loading, the leader pin was pushed in slightly, it could partially open the case and the leader ping would then get stuck and IBM would have to open the drive to get tape out. Later LTO1 tapes (as off about a year ago or so) are supposed to be welded and LTO 2 tapes from the start. 2. There also was a mechanical mod to the LTO1 drives, that would reduce or elimimate the problem. It involved installing a clip in the LTO1 drive. I had this done on mine and have not had the problem since. Don't know offhand the EC number on this, I think you have to ask about this one, and press a little bit to get it. I understand it originated in Europe. If you can't get info on it from your IBM CE, let me know and I'll dig up info. Talked with the IBM CE here, and he looked it up. The clip is put over the tape retension spring inside the drive. It's ECA 009, which is not a mandatory EC. Should be applied only if the customer sees frequent B881 errors in the library. If so, the CE needs to order part #18P7835 to get the parts and tool for the EC. Also says the work takes an half hour per drive. We're seeing a lot of these B881 (tape unload from drive issue) errors! Scheduling installation of these clips now. And, yes, we do have the original generation of IBM LTO1 tapes. :) B881 errors certainly makes sense now since there's only a few common major culprits for such errors; leader pin being out of alignment certainly does it. We haven't seen these errors with the other 3584 at another site, but will keep this EC in mind. -Dan
Re: TSM 5.2 on AIX 5.1
Hot Diggety! Lawrence Clark was rumored to have written: We originally decided to put TSM 5.2 on the AIX 5.2 system for the migration because of a notice in the 5.2 install doc that said TSM 5.1 would cause 5.2 to crash (at least the version that was on the Bonus disk.) If my recollections are correct, that combination was an extremely early version of both and the issues were fixed long ago. I don't have TSM 5.2, but I believe quite a few people are running TSM 5.2 on AIX 5.2 just fine. Are there any people on this list running TSM 5.1 on AIX 5.2? I'm running TSM 5.1 on AIX 5.2, works fine, for what it's worth. -Dan
Re: Open VMS and TSM
Hot Diggety! Muhammad Sadat was rumored to have written: I wonder if TSM supports Open VMS as server or client?? There's no TSM server software for OpenVMS BUT yes, OpenVMS can be a TSM client if you purchase an additional commercial third party software called ABC which uses the TSM API. I do not have any VMS servers left but I know a VMS system manager who uses TSM + ABC and is very pleased with both software. For more information, please see: http://www.rdperf.com/RDHTML/ABC.HTML The license pricing information is at: http://www.rdperf.com/RDDOC/ABC_PRICE_LIST.DOC I do not have any business relationships with the company, and have never used it myself... only heard about it from someone who has it. Also, keep in mind that this does not provide any bare metal restore (BMR) capabilities... you still have to use the standalone BACKUP utility to disk or tape to have a supported bootable restore from scratch for that. -Dan (I still have at home: a VAX, two Alphas, and an emulated VAX via SIMH, all running 7.2 and 7.3, so I am familiar with OpenVMS.)
Re: TSM 32Bit to 64bit Upgrade
Hot Diggety! Hart, Charles was rumored to have written: We are currently are running TSM 5.1.7.3 - 32bit on a p630 AIX 5.2.2. We would like to change the TSM from 32bit to 64bit during this upgrade. We ran a preview of the install choosing the 64bit rte and server code and lic filesets but the preview failed because it stated that the tsm.server.rte was already there. We decided to just go ahead and upgrade the 32bit code. Has anyone upgraded TSM 5.x from 32 to 64bit code? The Quick Start and Admin guide only make reference to AIX 4.3 to 5.2, nothing about 32Bit to 64bit TSM. Haven't done the exact situation you described but my guess is you'd have to upgrade the OS to 5.2, verify bos.64bit is installed, bootinfo -K says 64 (if it doesn't, you'll need to adjust the two symlinks, do bosboot, and reboot), then uninstall tsm.server.rte and install tivoli.tsm.server.aix5.rte64, and make sure TSM and the OS is patched/upgraded. You'd be well advised to take a mksysb backup of the system prior to the upgrade and perhaps do the 4.3.3-5.2 upgrade via alt_disk_migration. -Dan
Re: 3584/lto2 code
Hot Diggety! Stef Coene was rumored to have written: Then use the website. Why don't you visit the homepage of your library: http://www-1.ibm.com/servers/storage/support/lto/3584.html There is a nice Firmware link and after 2 clicks you can download any firmware you want for your library. Yes, but the firmware there is relatively outdated, too. (At least, it is for LTO-1, and presumably also true of LTO-2 judging from the comments seen earlier in this thread.) We got newer firmware by asking our main IBM CE very nicely, so he fetched it off the latest of his firmware CDs (don't recall the name, but it's available only to CEs and other IBM service personnel). -Dan
Re: Any experience with Sepaton VTL
Hot Diggety! Johnson, Milton was rumored to have written: I got a call from a rep asking if I was interested in a Sepaton S2100 VTL (Virtual Tape Library) (www.sepaton.com). It's billed as: My questions include where's the down side? What's the catch? If your choice is between expanding by purchasing a second 3494 frame or a S2100 VTL, why choose a 3494 frame? Keep this rep honest -- ask him/her what the down sides (negatives) are, what the typical failure modes are, and so forth. You'll soon know if you're about to drink sugary Kool-Aid or not :-) Well, simply put, I haven't heard of VTLs as a single source replacement for tape drives per se. They are pretty good when you can't wait a single second longer (or 2-3 minutes) for a restore to begin -- compare disk vs tape restore load-to-ready time. Lots of places has this kind of rapid restore requirements -- financial firms (banks, Wall St, etc), hospitals, nuclear power plants, utilities, and other places where any sort of downtime is extraordinarily bad. However, I don't think disks yet have the long-term reliability that tape drives do... well, server class SCSI drives *can* usually last 5 years in brutal 24x7 operation, but drives in general aren't too tolerant of being underutilized or if it's the cheaper engineered hard drives (e.g. typical IDE drives), overly utilized. So the way I see VTLs as being most useful is if you can't wait the 2-3 minutes it takes to load+spool a tape to 'ready to peel data off'; it's still no replacement for any serious archiving past perhaps 12 months or so, and still needs another backup source to restore data from in case a drive goes south and loses the data. Modern tape drives are pretty zippy, too, at 70 MB/sec, don't forget. Tapes (not tape drives) also contains far fewer moving parts that can fail than hard drives which has a motor, PSU, depending on the sub-1mm air gap (Bernoulli effect), etc. VTLs has a place, in my honest opinion, but only if you've got the need and only if it isn't the sole source for backup data. I don't think most folks has this need; so it just seems like a big push for companies to make money at your expense, unnecessarily (unless you actually do have a need and a well-engineered overall setup). After coming off rough economic times, you can expect lots of these pitches. :-) I've already gotten two of these so far. :) -Dan
Directory storage pools
As per the TSM redbook at: http://www.redbooks.ibm.com/redbooks/pdfs/sg245416.pdf It suggests use of DISKDIRS and OFFDIRS to store directory information to avoid restore-time directory reconstruction hits. Everything makes sense, including the need to explicitly bind all directories on a client to the DISKDIRS management class with DIRMC. The next question is... *how* do you select ONLY directories for this management class without selecting any files within them, in dsm.opt? (Server is 5.1.8.0 on AIX and clients are 5.1.5.14/15 on Windows, Linux, Solaris, and AIX.) I had also checked the TSM 5.1.5 UNIX client reference from: http://publib.boulder.ibm.com/tividd/td/TSMC/GC32-0789-02/en_US/PDF/GC32-0789-02.pdf ...and read through everything referencing include, exclude, dirmc, but it's still not obvious to me how to select only directory data for use with DIRMC DISKDIRS. I see there's a dirsonly keyword, but how to use it within context of dsm.opt? include / -dirsonly -dirmc diskdirs ? -Dan
Re: Sizing an AIX platform and tape libraries
Hot Diggety! Nancy R. Brizuela was rumored to have written: 1) Right now we are backing up about 150 GB/ night, but we need to add Exchange and a new student information system (Banner) to this. We are estimating that we will soon grow to at least 500 -600 GB/night. 2) Workload consists of 95 clients, consisting of from small 20-30gb Unix and Windows systems, to DB (Oracle) and file servers with half to one terabyte of storage. This will grow to 120 or so clients soon. Sounds like you know your current needs and have some idea of growth -- that's great! Helps to size potential candidate setups so much better and more accurately. 3) We are currently storing 10.7 TB total data in two libraries, one local tape library and a second copy of everything at a remote tape library (3494 ATLs w/3590 E drives). Looks like this storage could triple, given how much we will be backing up each night (15 TB in each location). My suggestion might be to consider something like the IBM 3584 LTO library with LTO-2 tape drives. These drives are pretty darned fast at up to 70 MB/sec (with hardware compression enabled), don't suffer from the same terrible stop/start issue that LTO-1 had, and holds 200 GB per tape natively (no compression) -- for *our* library, we're seeing 2.24:1 compression with LTO-1 (so no reason not to expect that to be similar for our LTO-2 setup) which suggests perhaps 400-450 GB per tape with hardware compression enabled, depending on data mix. The 3584 can go up to *16* library frames, with up to 12 drives per frame (ie, 192 tape drives total) and up to 6,881 tapes in a 16 frame setup. That's a lot of future expansion potential, but the nice thing is that you can buy a frame or two now, and add more on later as your needs grows. In a two frame setup with 12 drives (our 3584 setup), has 610 tape slots... which, in a LTO-2 setup, yields 122 TB of uncompressed data. A single base frame 3584 LTO library has about 250 tape slots if you have a 30 slot I/O station... so, 250*200 = only 50 TB. If you say you expect to eventually triple... 15 TB in each location * 3 = 45 TB, and you'd still have room for as much future expansion as you want. The 3584 has been a very nice ATL library for us. Footprint isn't bad; the two frame setup is about 5'x5'. 4) Network is Gigabit Ethernet. No problem. Get as many IBM FC5700 (Gig-E/fiber) or 5701 (Gig-E/RJ-45) adapters you want/need, plug in, determine how you want to handle the networking (EtherChannel trunking, subnetted for each adapter, or whatever). 5) We would like to use one large server vs. multiple servers. Not a problem. We use the pSeries 660 model 6H1 server, which has been a perfect fit -- it's modularized with the CPU and I/O units separate so it hasn't been a physical hassle to rack mount it, and these 6H1s has so many slots like you wouldn't believe. Up to 2 RIO drawers per system; each drawer has *14* PCI slots! So we don't have slot pressure -- we can throw in as many adapters as we want! FC cards, SCSI HBAs, Gig-E, SSA adapters, whatever you want, throw it in there. It also has multiple busses to spread out the bus traffic as well as a great memory and CPU interconnection, internally, so performance is not an issue at all. It also has 2 GB of RAM but can easily go to much larger configurations as our load grows further. At today's memory prices and systems configurations, probably wouldn't think twice about 4 GB of RAM... but if stretching every dollar, 2 GB would be an adequate config for your current needs (you mentioned number of clients, size of library, etc) and short term needs. Also has 4 processors in it due to the high I/O load -- CPUs are kept busy putting data on / off I/O adapters, processing the data, then offloading to yet some more I/O adapters (network, disk, or tape). 4's an healthy number of CPUs that has worked out really well. Load is nearly nonexistent! (We could probably do fine with 2 CPUs, but decided it was easier to get 4 and overengineer slightly, to grow into it over time, than to go back later and beg for expansion in a bad situation with loading issues. Besides, it'd have meant downtime to add more CPUs in the future amongst other logistical issues...) The 6H1 also has dynamic hot-plug slots (you have to run a command to turn slot power off, do whatever, turn slot power on, cfgadm, done), a service processor (good for powering on/off a machine remotely as well as forcing an halt even if OS is unresponsive... really rare but have used that once), monitoring/management features, and many other things. IBM isn't selling the 6H1 any more; they're selling its successors which are even better -- the POWER4 based systems. (The 6H1 is basically a rebadged and modularized H80 server, and one of the last machines released before the POWER4 systems came out in force.) So my point is simply that you want to look at a mid-level server that has remote management, reliability and availability features,
Re: 3583 Library Firmware upgrades
Hot Diggety! Prather, Wanda was rumored to have written: You can go to www.storage.ibm.com www.storage.ibm.com and download the latest 3583 or LTO firmware any time. But it doesn't tell me what's CHANGED in each level of firmware. The local IBM CE usually is able to get something for me if I make a request for it explaining the rationale behind it (change management, planning upgrade, need to verify it's fixed known existing issues we've seen, etc) and asking really nicely. ;) He gets it via internal request to the tape storage people (IBM Tucson?) and gets actual info via either internal email or internal web site, prints it out, and sends it to me. I sure do wish they'd post the change info on the web, too! -Dan
Re: 5.2 upgrade path
Hot Diggety! Jolliff, Dale was rumored to have written: We are looking at an upgrade from 4.2.1.15 to some version that supports LTO Gen 2 drives, which I hear is 5.2 or later... Nope, LTO-2 support arrived in 5.1.6.1. There are *serious* problems with some of the 5.1.6 tree to the point where IBM actually yanked some code after release; started to stabilize around .3 or .4; .5 is last in that tree. You can see LTO-2 support arrived in 5.1.6.1 by looking at: ftp://ftp.software.ibm.com/storage/tivoli-storage-management/patches/server/AIX/5.1.6.1/TSMSRVAIX5161.README.SRV Do note that the initial release (5.1.6.1) with LTO-2 support didn't cover the MVS or PASE platforms, if you run either of these two platforms. I have the 5.0 media. Are all the versions between 5.0 and 5.2.x downloadable, or is there another step that requires a media set? Don't know about 5.0... but we got 5.1, and all 5.1.x.x versions are freely downloadable. 5.2.x.x stuff requires a 5.2 CD for the base filesets. By 5.0, I think you mean 5.1? I don't think 5.0 was ever released. If you do indeed mean 5.1, then you're in luck -- just upgrade to anything at or between 5.1.7.0 and 5.1.8.2 (latest 5.1 release). To get the fixes of any TSM version for any client or server on any platform: ftp://ftp.software.ibm.com/storage/tivoli-storage-management/patches/ TSM server patches would be in the 'server' subdirectory of the above URL. Then pick your server OS platform, then desired server version, and then download the actual file containing patches. What is the oldest/most stable version that understands LTO Gen 2? I'd say 5.1.7.0 or so... but we've had no problems with 5.1.8.0 and in fact, somewhere in 5.1.7.x or 5.1.8.0 fixes an annoying thread message that used to appear in the actlog for every single backup session. 5.1.8.0 fixes a serious security hole; IBM **strongly** advises all 5.1 customers to run at least this rev for that reason. (There are also patched versions of 4.2.x and 5.2.x for this hole as well.) Is anyone else out there running TSM/ACSLS/STK library/IBM Gen2 drives? Nope, we use the 3584 with Gen 1 LTO drives, for what it's worth. -Dan
Re: AIX startup of TSM scheduler
Hot Diggety! Prather, Wanda was rumored to have written: This always works for me: nohup dsmc sched 21 /dev/null /dev/null Alternatively, drop this one liner in /etc/inittab on an AIX system: dsmc::respawn:/usr/bin/dsmc sched -password=quiet /dev/null 21 (It is usually preferrable to put this at or near the very end of the file) Then do: # kill -HUP 1 (to make inittab aware) Subsequently, if you want to force the scheduler to update, it's just a matter of kill -9'ing the 'dsmc sched' process ID and inittab would automatically respawn a fresh copy of the dsmc scheduler. For instance: # kill -9 `ps -ef|grep '/usr/bin/dsmc sched'|grep -v grep|awk '{print $2}'` This general approach works if you've got the client password stored as an encrypted hash in a file locally... if you have a setup that requires someone to manually enter the password every time, then this may not be a suitable approach (use of inittab to respawn). -Dan
Re: List all tapes, highlighting those in library.
Hot Diggety! Deon George was rumored to have written: Has anybody got an SQL, that can list all VOLUMES, and highlight (either showing the element number or something) those that are in the library? This report would be useful to see quickly if a list of tapes are already in the library - or which tapes in the list are not and need to be checked in. I got a list of tables by doing: tsm select tabnames,remarks from tables Then I decided to look at the table called LIBVOLUMES because your request is essentially the SQL equivalent of 'query libvolume'. So then I decided to see what fields (columns) were present in the table called 'LIBVOLUMES' with: tsm: MYSERVERselect colname from columns where tabname='LIBVOLUMES' COLNAME -- LIBRARY_NAME VOLUME_NAME STATUS OWNER LAST_USE HOME_ELEMENT CLEANINGS_LEFT Based on that, I figured only three columns might be useful. So: tsm select library_name,volume_name,home_element from libvolumes ...which would produce an output like: tsm: MYSERVERselect library_name,volume_name,home_element from libvolumes LIBRARY_NAME VOLUME_NAMEHOME_ELEMENT -- -- 3584LIB1 MYS000 1026 3584LIB1 MYS001 1027 3584LIB1 MYS002 1028 3584LIB1 MYS003 1029 3584LIB1 MYS004 1030 3584LIB1 MYS005 1031 [...snip...] If you have only one library, you are free to leave the library_name off the select query. Parse the results as you like, regardless of if it's called via an internal TSM server-side script or if it's called via a script that parses the output of calling dsmadmc in batch mode. It is trivial in either case. -Dan
Re: 3584 LTO1 new firmware installed : 36UC
Hot Diggety! Lambelet,Rene,VEVEY,GL-CSC was rumored to have written: we just upgraded the firmware for our LTO1 drives in the 3584. Version in now 36UC instead of 36U7. The bad bug (ask your CE about it please, if you are using 36U7) should be fixed now!! (Library is still at firmware level 3480). More details from the CE an half hour ago: LTO gen 1 drives with microcode version 36U6 through 36UB is affected by this *really* serious bug. IBM is recommending customers put 36U5 on now if they can't/won't load 36UC right away. The bug: if certain types of permanent write errors is encountered during operation AND a subsequent rewind command is issued, then an end of data mark may be written at the BOT (beginning of tape). That effectively loses existing data on the tape. The data can be recovered but they must be sent to IBM for that, and to be arranged through customer's CE if this pops up. Looks like I get to load 36UC on Friday afternoon. :) Otherwise, you don't want to find out at disaster recovery time that some tapes are not currently usable for restores, with a wait of days or weeks between repair and getting it back. -Dan
Re: anyone had network problems with 3584?
Hot Diggety! i love tsm was rumored to have written: The two 3584s that I run have both recently gone off the air,i.e I can't ping them. They both have green link lights on the NIC. They are set to 100/Full although I have tried Auto and 100/half. 3584 Operator Guide recommends they be set to 10/half (if hub) or 10/full (if switch). They do not work real well at 100 Mbps, and they haven't quite ironed out the culprit despite multiple revisions of library firmware, from my understanding. We have given up on ever getting it to work reliably at 100 Mbps. Seems fine at 10 Mbps, especially if the switch port is hardcoded to match the 3584 enet settings exactly. Don't really trust autonegotiation. IBM CE was here recently for a firmware upgrade and he had some internal IBM documentation for himself (recently issued) that still advised if Ethernet was hooked up, to set it to 10 Mbps only. -Dan
Re: SV: Query from z/OS-platform to intel-platform
Hot Diggety! Richard Sims was rumored to have written: 3. What does the rumor say - when can we expect a real DB2 ;-) Imagine what the licensing fee for TSM would be then!! ;-)) Might jack up the price quite a bit... but would likely still be several orders of magnitude cheaper than licensing Oracle ;) /tongue-in-cheek We've got various of the major databases in-house, and the licensing costs for each is just simply... *amazing*. On some days, I get the distinct impression we're singlehandedly paying for Oracle CEO Larry Ellison's private high-performance jet. -Dan
Re: TSM 5.1 server on AIX 5.2 64-bit?
Thanks to everyone who helped out! We successfully migrated the AIX 5.1 server (6H1) to AIX 5.2 using alt_disk_migration cloning and then doing a migration upgrade install on hdisk1 (while preserving AIX 5.1 on hdisk0). We had also done a bootable mksysb to DVD-RAM beforehand with mkcd. Was trivial -- only a few commands required from start to finish. TSM 5.1 64-bit started up just fine under AIX 5.2 64-bit. Once we were satisfied everything worked, after a while, we eventually removed the old AIX 5.1 rootvg on hdisk0 and re-mirrored with AIX 5.2. Had there been a serious problem with the AIX 5.2 environment, we'd have simply booted off hdisk0 to gain the pre-upgrade AIX 5.1 environment and junked AIX 5.2 on hdisk1. (This approach is similar to how you can use Solaris' flash archives for OS installation on an alternate disk.) As an unrelated note -- we had dsmc core dump with internal program error while doing an incremental backup. Turns out it was because the Solaris snapshot filesystem was corrupted internally with some bad metadata, so it wasn't dsmc's fault after all. We unmounted, redid the snapshot, and re-mounted it; dsmc ran fine that time. (We also figured out the cause of the Veritas Volume Replication for the Oracle DB snapshot filesystem corruption and applied a small adjustment to prevent it in the future.) -Dan
TSM 5.1 server on AIX 5.2 64-bit?
1. Will that combination work? 2. Is that a supported combination? Or is only TSM 5.2 server on AIX 5.2 supported? The docs I've seen to date hasn't directly addressed TSM 5.1 server on AIX 5.2 64 bit, and I don't have a spare 64-bit 5.2 machine to test with at the moment. (Got plenty of 32-bit 5.2 machines; verified the client works fine) Thanks! -Dan
Re: Mkcd-AIX help
Hot Diggety! PINNI, BALANAND (SBCSI) was rumored to have written: Please excuse me for asking aix issue .But I think this was pressing issue for me right now. I use mkcd to backup AIX os to DVD . My question is what is equivalent to /etc/exclude.rootvg for mkcd compared to mksysb. I want to exclude file system (application) from backing up on DVD. Well, you can do a mksysb if rootvg or savevg if non-rootvg volume group, with your own exclude file. Then you supply that image to mkcd. Ie: # mkcd -m /path/mksysb.image any other flags or # mkcd -s /path/savevg.image any other flags -Dan
Re: Non-IBM LTO tapes in a 3584
Hot Diggety! Shawn Price was rumored to have written: For what it's worth, my boss cheaped out and got some non IBM tapes for our 3584, and I've had nothing but problems with them. Mostly Imation and a few Emtecs. We are running the first generation LTO 1 drives, so there might be an issue with those as well. I have now gone in and just started pulling the tapes that have gone read-only more than twice and replaced them with IBMs. Make sure your library and drives are at the latest firmware revision levels -- this is especially important since more recent firmware revs fixes bugs that causes problems with non-IBM tapes from what I hear. Latest 3584 library firmware you can get from IBM's web/ftp site: 3300 if no D42 drives in library; else, 3060. To check firmware rev for library w/AIX: # lscfg -vl smc0 | grep FW The library firmware rev # also appears on the main LCD panel screen. Latest 3584 drive firmware you can get from IBM's web/ftp site: 36U3 for LTO-1 drives and 38D0 for LTO-2 drives. To check firmware rev for drive w/AIX: # lscfg -vl rmt0 | grep FW (...and for each drive... rmt1, rmt2, etc.) -Dan
Re: New 3584 library and drive microcode
Hot Diggety! Lambelet,Rene,VEVEY,GL-CSC was rumored to have written: Eric, following some unmount failures, IBM recommended to install 36U3 (36U5 was chosen because it appeared on the IBM ftp site the day before). I noticed that several people installed 36U5 but it does not appear to be on either the web or ftp site... 36U3 is present for both. So they may have yanked 36U5 -- possible serious bugs? Just mentioning as an advisory note / heads-up. -Dan
Re: audit library fails
Hot Diggety! Karel Bos was rumored to have written: On both sites we had some problems with tapes stuck in drives. On both sites we had to do audit libraries to get TSM in synch with the libraries. While the tapes were stuck in the drive, the audit library failes and now one of the slots is empty (tape was in the drive and is completely gone) and still the audit fails. Has anyone seen this type of behavour? Is it a TSM thing (don't think so because of the version difference between sites) or has something been altered in the microcode of the library what changes things? The 3584's a great unit. TSM's audit library requests data from the 3584's inventory data... so even though you may have removed a stuck tape, the 3584 may still not be aware of it for its scanned inventory data. What works for us in this situation is to go to the front panel (or the unit's web interface if you hooked up Ethernet to it) and have it do an inventory. 45-50 seconds later, it's done. Now go to TSM and do an AUDIT LIBRARY. It should now work as expected. -Dan
Re: Wouldn't it be nice...
You bring up a good point - these two folks are posting here because they genuinely want to help out and willing to tolerate the flak and other divergent opinions at times. For having gone so far beyond the official duty requirements, I think it'd be nice to give them due recognition and a big round of thanks. They are not official spokespeople; they are not expected or required to participate here, to forward reports, to come up with ideas, to defend the organization, amongst many other things, but they still hang out, anyway, and help out a lot. I would *never* jump on them for not doing more - they have already willingly gone so far beyond what is expected of their duties; I appreciate it and accept it, and allow them to be as human as they wish - that means maybe they aren't always interested in responding to every single report concerning their speciality at all times. That's all right; that's why we have support contracts to guarantee we will always get the appropriate support that we need at the right times. If we wish to have greater participation and representation by IBM/Tivoli experts here in adsm-l, I believe the appropriate folks to discuss it with or pressure would be their management, directly. Oftentimes, some reasons why more people from a vendor don't participate in their product's public forums -- fear of legal issues, not enjoying being singled out and flamed (we're all human) for decisions made by others, an environment that's perceived to be demanding, amongst many other reasons. So, thank you very much, Mr. Raibeck and Mr. Hoobler. (amongst numerous others whom also posts here, of course.) -Dan
Re: LTO throughput - real world experiences
Another comment... LTO numbers should probably be reported as being LTO-1 or LTO-2. LTO-1 max uncompressed is 15 MB/sec, max compressed is 30 MB/sec; LTO-2 max uncompressed is 35 MB/sec, max compressed is 70 MB/sec. This assumes a typical average compression ratio of 2:1, of course. We use tapeutil under AIX (part of the Atape driver) to test the raw throughput to tape with the rwtest option. May be possible to do similar tests with other tools under other OSes. We did this with the drive in question disabled in TSM then put an unused blank tape (unlabelled and unknown to TSM) in the upper I/O station, closed the door, then did: # tapeutil -f /dev/smc0 move 769 257 (move from first I/O station slot to first drive in first frame; AIX knows this drive as rmt0 in our case. **MAKE SURE YOU VERIFY THE ELEMENT ID FOR DRIVE TO OS'S DRIVE NAME MAPPING OR YOU COULD DESTROY A TAPE WITH VALID DATA ON IT!!!**) # tapeutil -f /dev/rmt0 rwtest -b 262144 -c 20 -r 3 (do a destructive read/write test with a 256K block size factor, run for 20 blocks for each read test, 20 blocks for each write test, and do both set of tests 3 times) # tapeutil -f /dev/smc0 move 257 769 (put the tape back in the upper I/O station slot when done) This destroys the ANSI tape label so it's best to do this on a previously unused tape or you'll have to re-label (dsmlabel or 'LABEL VOLUME') it afterwards. And then, of course, re-enable the drive within TSM. That gives us some idea of the raw potential. Then within TSM, I like to see how fast it can push large files through to tape in a diskpool-tape pool migration. Of course, this is also another 'test best case' scenario because tape drives are more optimized for large files than for a number of small ones, but it still gives you an idea about the upper bound of real world performance. Between the tests and TSM tape writes, it appears we get 22.5 MB/sec sustained in compressed mode for LTO-1 drives. That's pretty decent; it's halfway between the theoretical min and max for compressed mode, and we're happy by its performance so we've got no complaints :-) I should note that we don't really do database backups (well, we usually export them to a SQL dump file in ASCII and back that up because the TDP type tools are priced *outrageously* and our needs aren't _that_ pressing) and that our data generally compresses well to tape -- we get an average of 2.24:1 hardware compression across the entire tape pool with a few bursts of 3:1 and once, almost 4:1 compression for a LTO-1 tape. The only way we get this kind of performance is to put a disk pool in front of the tape drives... and we also limit it to two drives per SCSI bus (because they're capable of peaking at 30 MB/sec or so in compressed mode... so 2x30 = 60 MB/sec which means you need at least a Ultra2 SCSI controller) to avoid contention or bottlenecks at the bus level. The earlier tips about testing the FC setup was good, too. -Dan
Re: TSM Staffing level question for Enterprise Shop
Hot Diggety! Gordillo, Silvio was rumored to have written: TSM Staffing Level Question: I'm searching for documentation, industry best practised, bench marks, etc... on recommended staffing levels for our TSM Environment. Need to get pointed in the right direction to provide management with levels of support for enterprise sized TSM shops. Maybe some TCO (total cost of ownership) validation docs out there to support staffing levels. Below is what we manage. 4 TSM Servers ver 4.2.2.12 (they will be replaced upgraded soon) AIX 4.3.3 780 TSM Client Nodes various flavors databases mostly oracle, SQL, Exchange 6 LAN-Free Clients. (Gresham EDT Knowledge) 4 STK 9310 Silos running ACS Server 6.0.1 Direct off-site copy pool stg process between data centers via Optical Carrier Well, my take on it is that for an enterprise sized backup system... needs to have a bare minimum of at least two skilled administrators - mostly for coverage reasons (vacation, illness, unavailability in crisis, etc). We're a small shop here - about 150 TSM clients, and we've found that one administrator is just fine (although we really do want more... but interesting other co-workers in learning can be a bit of a daunting task sometimes...). Reason? Once you've configured TSM to your satisfaction, it pretty much runs itself if you've set it up properly -- including scripts to manage routine events, to check for abnormal events, etc. The high cost of TSM setup for us was purely in the time it took to read/re-read the manual and experiment with various setups, do testing, iron out stupid SCSI issues with the tape library, and then to write about 40 TSM management scripts for the first pass of the updated TSM platform. After that, it pretty much ran itself trouble-free after we'd worked out the initial kinks during the first week of production for the updated TSM setup (5.1, new library, new server, etc). Only change we had to do after introducing the tested setup into production was to increase the tsm db log size. That's it. So for us, the expense is mostly in the time spent planning and configuring the setup... post-setup, costs us just about nothing for manpower - two seconds to scan the daily reports and that's it. However, with more complex setups, definitely do need more people - especially useful if you had a serious crisis such as a large scale site breakage or even just for coverage reasons. How to calculate how many people required... that's honestly a good question. I'm of the school of thinking that says with a properly designed setup, it should be able to pretty much scale up to any size. I know I could go to 500-1000+ clients with the existing tools without modification because I designed it to scale up easily and without any additional work. So in theory, I could continue to run this setup ok just by myself... although not most recommended idea for a large setup. So... 'appropriate number'? May vary with actual production workload - for instance, are TSM administrators expected to do tape swaps? If a *lot* of tape swaps each week (or day) is required, this can take hours at really large sites. Also, if you have older tape technologies or a really complex setup that breaks down often, then that's also very labor intensive in dealing with it, too... and would require more administrators. Also, if you have significant tape infrastructure at multiple sites that are far apart, you would most likely want to have at least one local TSM administrator per site - for tape swaps, for debugging tape library issues, for disaster recovery, etc. If it was a couple of sites in the same town or nearby, could get by with fewer people... but harder to do that if you had sites 500+ miles apart. I don't mean a couple drives and 50+ tapes at a site, I mean the more significant ones with multiple robots and hundreds or thousands of tapes. A single admin that is very good with churning out decent scripts to manage the platform in 'automatic' mode (self-running) is worth his/her weight in gold and is significantly cheaper than 5 or 10 less experienced people whom insists on doing everything manually (very labor and time intensive). So admin experience and skill level also has to be factored into the TCO, too. Because experienced admins making good tools will have setups that pretty much run themselves well and effectively which saves you heaps of money in the long term and hence, reduces the overall TCO if averaged and amortized on a yearly basis. So, the answer you seek... will depend on a variety of factors... age of infrastructure, complexity of infrastructure, size of installation, number of sites with sizable tape infrastructure, skill of personnel, coverage needs, job expectations/responsibilities, growth rate for tape infrastructure over 6-36 months, and so forth. Training is also a more significant cost with more complex and larger enterprise setups -- training classes, test server and associated hardware to
Re: oops forgot the error message.
Hot Diggety! Scott Figgins was rumored to have written: I'm getting this error in my Server log on AIX 5.1 with TSM 5.1.6.5 after defining a 7336-205 4mm Tape library. Any ideas? IBM says this library is supported in the BASE codeset. 1. What is the output of: # lsdev -Cc tape 2. What Atape version do you have? 3. What were the commands you used to define the library? -Dan
Re: oops forgot the error message.
Hot Diggety! Dan Foster was rumored to have written: Hot Diggety! Scott Figgins was rumored to have written: I'm getting this error in my Server log on AIX 5.1 with TSM 5.1.6.5 after defining a 7336-205 4mm Tape library. Any ideas? IBM says this library is supported in the BASE codeset. 2. What Atape version do you have? Er, I just realized that I'm not sure you need the Atape driver for a 7336, and most likely you don't need it. I've got an IBM 4mm library but last time we used it was about seven years ago when we did backups with Sysback. Do you have patches applied to AIX 5.1? Latest maintenance level is either ML3 or ML4. -Dan
Re: TSM LTO 3580 Performance
Also, is there a diskpool? ie: client-server[diskpool]-server[tapes] If there's no diskpool in between the client and the tape drives, will be a lot of stop/go writes to tape, resulting in about 1 MB/sec vs 10-25 MB/sec. At least, that's true for LTO-1 drives. I've heard that LTO-2 drives better handles this kind of situation. -Dan
Re: TSM LTO 3580 Performance
Hot Diggety! Colby Morgan was rumored to have written: The SCSI adapter is a 2940UW. It does run great with large files, so if we had an adapter hardware bottleneck file size shouldn't make a difference. We are running the v5.0.2183.1 of the Microsoft drivers. I also opened a call with IBM, and they didn't make any recommendations to update the adapter drivers. How many drives per SCSI bus? We limit it to two for a 80 MB/sec SCSI bus because a single drive is capable of pushing up to about 30 MB/sec in compressed mode over the SCSI bus. As a side note - with a diskpool, we achieve about 70-80 GB/sec with our LTO-1 drives, so the fact you're getting 1/5 to 1/8 the performance does sound pretty terrible, indeed. Could it be cabling quality issues - ie, reflection on the bus causing excessive retries or other related SCSI errors? That's another place where performance could be killed. -Dan
Re: Small TSM deployment
Hot Diggety! Levinson, Donald A. was rumored to have written: has anyone deployed TSM from a B50 or a 43p 140 ? I need to back up about 4 GB a night on a local network from 2 clients and maintain a total tape pool of about 2.5 TB on about 100,000 objects or less. A B50 is just a repackaged 43P-150 in a rack mount case. It's (either one - the B50 or 150) is a very capable machine. Only limitation to the B50 is that it is not slots-happy ;) Has only two expansion slots so you have to choose *VERY* carefully with what and how you expand it... and it also can not support gigabit ethernet adapters -- presumably due to the fact it's a 32 bit machine. Doing some quick calculations shows 8 GB/night to be about 8.3 MB/sec for about 8 hours which means about 70 Mbps on the network -- which would just barely saturate a fast ethernet network. You would be *highly* advised to add a second FE network adapter to the B50 and split the clients -- point half of the clients towards one B50 interface and the other half towards the second FE interface. Ie: backup1.foobarbazz.com points to 10.1.1.1 backup2.foobarbazz.com points to 10.1.2.1 (and both IPs would be individual IPs on each ethernet interface... would put both on separate IP subnets such as a class C or /24 because this would prevent return traffic from hogging one primary outgoing interface.) On half of your clients, point them to backup1 in dsm.sys, on other half, point to backup2 in dsm.sys, etc. How do you decide which clients to split up? Do it based on data volume for backed up data. If one client does 50% of the data traffic and 99 clients does 50% of the data traffic, then that's how you'd split it up. I would not recommend the use of a 140 because IIRC, it's a PReP based machine and I think AIX 5.2 dropped support for these? The current generation RS6Ks and pSeries servers are CHRP based, including the 150 and B50. Also, if you can daisy chain your tape drive(s) off the onboard SCSI adapter for the B50, that might work... but you would most likely want to have a diskpool and split the SCSI bus between disks and tapes; which implies adding a second SCSI adapter for the B50, using up its second expansion slot (first would be for another FE adapter). Otherwise you could end up with a serious performance bottleneck that could totally negate whatever the B50 is capable of handling. It's also uniprocessor which could potentially mean some CPU contention for various resources (tapes, disks, network, cpu) when it's busy. In short, the B50 should work out ok, but you appear to be quite pushing the very limits of its packaging for your _current_ needs, nevermind long-term future growth needs. You would be highly advised to get a slightly beefier machine, or at least a machine with more PCI slots than the B50. Right now, the next step up above the B50 would be the pSeries 6C1 or 6C4, at about double to triple the cost of the B50. Unfortunately, we have not found any other intermediate range machines between these. I do use a B50 for some TSM testing so I know it works fine as long as you don't try to push more data through it than what its adapters can handle. -Dan
Re: urgent! server is down!
Hot Diggety! Richard Sims was rumored to have written: After a reboot yesterday tsm doesnt start. ... ... ANR0900I Processing options file dsmserv.opt. ANR000W Unable to open default locale message catalog, /usr/lib/nls/msg/C/. ANR0990I Server restart-recovery in progress. ANRD lvminit.c(1872): The capacity of disk '/dev/rtsmvglv11' has changed; old capacity 983040 - new capacity 999424. ... I would take a deep breath and stand back and think about that situation first... There's no good reason for a server to be running fine and [...] I agree with what Richard had to say. Taking a deep breath is always step #1 for handling a crisis without making it worse. 999424 - 983040 = 16384, which is exactly 16 MB and sounds suspiciously like the PP size. 'rtsm...' sounds like a raw LV rather than a filesystem. Perhaps someone with root access had done this at some point earlier: # extendlv tsmvglv11 1 [or had done the equivalent in SMIT.] (DO NOT EXECUTE THE ABOVE COMMAND! I am only theorizing what may have happened) As for eng_US vs C, do this: # grep LANG /etc/environment If it says LANG=C then try: 1. Changing it to LANG=en_US in /etc/environment 2. At the root prompt: # export LANG=en_US 3. Try starting up TSM now And you will probably want to ask your operations staff if anyone had increased the LV's allocation, perhaps by one physical partition with extendlv or similar. If someone had done it, I'd have made them put Humpty Dumpty back together as a great learning experience ;) Tell people to *NOT* mess around with the TSM server if they do not know what they're doing. I did a quick test with TSM 5.1 by creating a small 16 MB DB logical volume (1 PP), started up server OK. Then I did 'extendlv tsmdblv 1', halted server, and tried to start it up again. I got the exact same errors you got. I suspect you may have to remove that LV, recreate it with the expected size that TSM wants, then do a DB restore from your most recent full db backup tape. But before you do that, you'll want to save a copy of your current device config and volume history file if you have these, as well as your dsmserv.opt file. Then look in the TSM 5.1 Server for AIX Administrator's guide at: http://publibfp.boulder.ibm.com/epubs/pdf/c3207680.pdf (This is assuming you use TSM 5.1 for AIX; if you use another version, you'll want to consult that guide instead, but the steps will probably be similar or still exactly the same.) DB restore is covered in Chapter 22. 'Restoring a Database to its Most Current State' at bottom of page 524 is probably your easiest option since it sounds like you have everything else intact -- volume history info, logvols, stgpool vols, etc. Then you'll have to delete (with 'rmlv -y tsmvglv11') the offending LV, and recreate it (with 'mklv -y vg tsmvglv11 number of PPs'). Then... Find out which tape has the most recent full DB backup, then do: # cd /usr/tivoli/tsm/server/bin # ./dsmserv restore db devclass=whatever vol=tape volser If that command worked (it's a preview, basically), then do: # ./dsmserv restore db devclass=whatever vol=tape volser commit=yes ...which will make the restore actually happen, for real. The actual restore operation is no big deal if you have a good and recent db backup tape, and know which tape it is. I did this as part of testing recently, and it worked right off the bat with no problems at all. If you don't know which tape volser has the latest full db backup, then you could look into your volume history file. For example, with my setup: backup1:/usr/tivoli/tsm/server/bin# grep BACKUPFULL volhist.cfg 2003/02/21 14:41:24 BACKUPFULL 5 0 1 3584_DEVCLASS1 ROC010 2003/03/01 21:06:51 BACKUPFULL 6 0 1 3584_DEVCLASS1 ROC012 (The file might be called 'volhistory.cfg'; I had explicitly defined mine to be 'volhist.cfg' at server installation time.) I have two full db backup tapes... one was done on 2/21, is version 5. The more recent was was done on 3/1, version 6. So I'd restore ROC012, for example. I should warn you that any backups done after the date/time of the most recent DB backup will effectively be lost, so be sure you really do want to restore the DB before committing to it. Doesn't sound like you have too much of a choice in this particular case. Note for the discerning ADSM-L reader: my production server has daily db backups! The above was from a test box. -Dan
Re: Automatic checkin on a 3584?
Thanks for all the comments, suggestions, ideas, and code examples. They were very invaluable! I also found expect to be an interesting approach that I hadn't considered, but makes sense. Looks like I'm on the right track now; thanks again! -Dan
Automatic checkin on a 3584?
Howdy - I seem to be having some sort of timing issues with the checkin process via an automated script. The procedure: 1. Fill the I/O station(s) with brought back tapes to be checkin'd 2. Issue a 'checkin libvol library search=bulk status=scratch checklabel=barcode' 3. Wait a bit for a request to appear in the request queue 4. Issue a reply corresponding to what appears in 'q req' 5. Wait a bit for the whole checkin process to finish in search mode 6. Issue a 'checkin libvol library search=bulk status=private checklabel=barcode' 7. Wait a bit for a request to appear in the request queue 8. Issue a reply corresponding to what appears in 'q req' Manually, works great. My script parses all data just fine using dsmadmc in batch mode (with -id= -pass=...) and the command syntax is correct -- I know because it works fine by hand. I'm a little stuck in getting the exact timing *just right* for the delays as well as the best way to construct a while loop logic to get the timing right. It just runs through it so fast that it exits before it's got a chance to answer both requests. Detecting certain corner cases seems a little interesting when you throw in vagaries of timing, also. Alas, cheating by direct loading and using search=yes to avoid the requests isn't an option for the weekly tape swaps. ;) Does anyone happen to have a similiar script fragment or even just suggestions on how to construct the while loop 'just so'? :) Anything would be much appreciated! (I'm trying to make the limited human interaction with tapes reasonably bulletproof without requiring them to manually wade through TSM - reduces chances of errors and not all personnel are skilled.) -Dan
Re: Automatic checkin on a 3584?
Hot Diggety! GUILLAUMONT Etienne was rumored to have written: You didn't say in what type of OS you where. If it is unix, no problem, you Oops! I'm normally good about that. Server is AIX 5.1 on pSeries 660 model 6H1 server. I do have some sleeps, but it's still running too fast or not exactly right. -Dan
Re: Linux 5.1.5.14-client
Hot Diggety! Tom Tann{s was rumored to have written: Hello *SM'ers! Is there a problem/bug in the libPiSNAP.so and/or libPiIMG.so? We don't use the default installation-path, but DSM_DIR, DSMI_DIR etc are set to the right paths. dsmc generated these entries in dsmerror.log: 02/19/2003 20:18:45 dlopen() of /local/tsm/bin/plugins/libPiSNAP.so failed. 02/19/2003 20:18:45 libApiDS.so: cannot open shared object file: No such file or directory. You have two options: 1. set up symlinks to /usr/lib then do 'ldconfig' or... 2. add /local/tsm/bin/plugins to /etc/ld.so.conf then do 'ldconfig' -Dan
Where to get 5.1.6.0 server for AIX?
I have the 5.1.6.1 server code but that apparently requires 5.1.6.0 code be installed...and I can only find 5.1.5.4 and 5.1.6.1, but no 5.1.6.0 on the FTP site...? Is 5.1.6.0 supposed to be for-pay software for existing v5 customers? I know they yanked 5.1.6.0 due to serious QA issues, but it seems unbelievable they'd issue 5.1.6.1 if it requires a yanked version that is no longer available? Are there any workarounds for going from 5.1.5.4 to 5.1.6.1 without having to deal with 5.1.6.0? Or is this going to be a case of calling TSM support and requesting a special cut of the package for 5.1.6.1? -Dan
Re: Where to get 5.1.6.0 server for AIX?
Hot Diggety! [EMAIL PROTECTED] was rumored to have written: [1]ftp://service.boulder.ibm.com/storage/tivoli-storage-management/mai ntenance/server/v5r1/AIX/LATEST/ Ahh! No wonder -- thanks! (I was looking at patches rather than maintenance tree.) Got it now. -Dan
Re: LTO Cleaning Cartridge Mount Points
Hot Diggety! Joshua Bassi was rumored to have written: Does anybody know offhand the estimated number of cleanings an IBM LTO cleaning cartridge supports before it should be disposed of? Thanks in advance. Exactly 50. ;) It's got a built-in counter on the tape, and the IBM libraries will report this via front panel LCD (at least, on the 3584, it does). I believe it returns an error if you try to use it a 51st time, from my recollection of the manual, but don't hold me to that. -Dan
Re: 3584 library loading
Hot Diggety! Matthew Glanville was rumored to have written: But, I have found bulk loading tapes with the door open on the 3584 is a problem when tapes are in the drives. I had to figure out which slots NOT to put tapes into because those are the slots that the tapes in the drives use. When TSM tries to unload a tape after you closed the door it knew which slot it was in and tries to put it there, thus failing if you have manually put a tape there. A simple audit library check=barcode don't work as it wont run until all drives are unmounted... Is that called a paradox or a conundrum? With TSM v5, you can parse the output of an undocumented (I think?) command: tsm show library libraryname It will have something like this for the drive that has a tape loaded: loaded volume home slot=1033 loaded volume name=ROC007 That tells you a) what the tape label name is that's loaded in drive right now, and b) which element number it was originally fetched from. So you just basically parse the output of that command and find out all the home slot numbers, and *NOT* load tapes in them. Volume name may be useful if you're going to be physically eyeballing the library, but otherwise not programmatically useful. I'm not sure if there's a better and documented command to do the same sort of thing. Also, don't recall if this existed in previous versions, but pretty sure I remember seeing it in TSM v4. Alternatively, if you're using AIX, you could parse the output of: # tapeutil -f /dev/smc0 inventory Something along the lines of: # tapeutil -f /dev/smc0 inventory|grep -p ^Drive Address | \ grep Source Element Address|grep -v Valid | \ awk '{print $NF}' -Dan
Re: TSM SERVER 5.1.5.2
Hot Diggety! David Longo was rumored to have written: 5.1.5.2 has been out about a month or so now. Also that something else isn't broke? There was a major performance bug fixed in 5.1.5.3; we had a really nasty situation where a west coast client was sending data at only 2 Mbps to the east coast server; raw network between the two is up to 32 Mbps as evidenced by ttcp tests between west coast client and other east coast servers on the same subnet. Turns out we needed an efix for the pre-release version of fixed netinet driver in AIX 5.1 (to be bos.net.tcp.client 5.1.0.37 and released next week) to fix the OS perf bug *AND* to also upgrade to 5.1.5.3 from 5.1.5.2. Doing so improved the ITSM network perf from 2 Mbps to 8 Mbps, and raw OS network perf to 32 Mbps. Merely fixing the OS perf bug wasn't sufficient for us. (We tested and retested between every single step to rule out multiple variables.) From the 5.1.5.3 README: PQ68076 PERFORMANCE DEGRADATION AFTER UPGRADE TO 5.1.5.0, 5.1.5.1 OR -Dan
How do you back up 2 PB of data?
2 PB is 2,048 TB, or 2,097,152 GB. A fun thought exercise: http://www.cnn.com/2002/TECH/biztech/11/19/ibm.supercomputerr.ap/index.html Well, assuming several things: 1. Using LTO (just because I know the numbers for this best off the top of my head) -- a 3584 library 2. LTO delivers maximum of 30 MB/sec in compressed mode, but 22-23 MB/sec is usually realistic. Let's use 22.5 MB/sec. 3. Typically 1.7:1 to 1.8:1 ratio for hardware compression Let's use 1.75, or 175 GB for a 100 GB uncompressed tape. 4. 72 drives per maxed out LTO setup (1 base frame + 5 expansion frames) for about 2000 tapes in all frames? 5. A single 3584 complex therefore delivers (using hardware compression) a grand total of 175 GB * 72 = 12.6 TB of compressed data *within* the library at any one time, and assuming the client is constantly streaming data to the ITSM server at peak efficiency, can back up 81 GB per hour at max write-to-tape speeds. 6. Assuming a 16 hour window for all backups to complete per day (so that you have time for other ITSM server processing), that's 81 * 16, or 1.3 TB per 3584 _drive_ per day. 72 * 1.3 means a single 3584 complex can do about 94 TB per day. 7. For a single full backup of 2 PB, that's 2048 TB, or 2,097,152 GB... or about 12,000 maxed out LTO tapes. Since a single fully fleshed out 3584 library is about 2,000 tapes... that would mean 6 3584 libraries for tape capacity alone. 8. 2048 TB divided by 94 TB yields about 22 3584 libraries. 9. Then you've got the small problem of having to come up with an appropriate ITSM server design... for starters, number of slots required would be incredible. You'd put max of 2 3580 drives on a single Ultra HVD SCSI adapter... so 72 drives per complex would be 36 slots alone! 36 slots multiplied by 22 complexes would be 792 slots! 10. Not sure about a p690 but think it's got a couple hundred slots? 11. Then you need more adapters for disk and network controllers. To support 22 MB/sec over 1,584 drives concurrently would be... 465 gigabit ethernet adapters assuming a perfectly tuned setup that can push 600 Mbps per adapter through. 12. You'd probably kill the bus with so much data zipping around long before you max out the slots... more likely you would need multiple (6-10?) p690 Regatta systems *just* to deal with ITSM backups for 2 PB of data alone. 13. The HVAC requirements for all these disks must be interesting ;) For the disks -- data, diskpool, db... total BTUs/hr would possibly be in neighborhood of about 3 million BTUs/hr which demands *seriously* beefy HVAC units for the disks alone, and nevermind for the servers, routers, etc...! 14. Probably has their own electrical substation for the computer room(s) alone. Run on an UPS? If they went to the extent of having own electrical substation, they might as well... The disks alone are probably going to eat about 15,300 amps at the bare minimum... total for entire room could be in neighborhood of 30-40,000 amps when you consider the large network equipment, servers, and other supporting infrastructure. I listed LTO and pSeries here just simply because I know the numbers and hardware the best, but feel free to offer other possible approaches. Keep in mind, all that is only a small part of the big picture... this one is *just* for a single full backup, and doesn't take into account the long-term needs such as ITSM db sizing or I/O loading of db or diskpool disks; each hard drive has a finite amount of I/Os it can do at any given time. Then you've got other issues such as performance vs reliability, which becomes even more tricky with the extremely large scale setups because use of RAID-5 could become a *very* real serious bottleneck that gums up the entire works. I actually wonder if ITSM on zSeries hardware would actually be better in this particular scenario because mainframes typically have superior I/O management, far beyond simple tricks like I/O pacing that exists on commercial UNIX OSes. Mainframes also have incredible I/O capabilities. Saw a zSeries box, had about 500 I/O controllers, and was still humming along just fine even under varying workloads. But I think that's balanced somewhat by the extensive training and support requirements, along with licensing and support contract costs. I do imagine that if I was the data center manager for that site, I'd be hiring an entire team of senior ITSM administrators with 20 years of experience ;) Teams of operators to deal with
Re: How do you back up 2 PB of data?
Hot Diggety! Orville Lantto was rumored to have written: A 72 drive, 10 I/O slot 3584 library will hold 2207 cartridges. with 175 GB/cartridge that works out to 6 libraries. Aye, in terms of tape capacity. However, if you have a requirement that it finish an entire full backup in a single day -- say, 20 hours...and you get between 15-30 MB/sec for HW compressed writes with the LTO drives. So we pick a common number that's actually sustainable in real life: 22.5 MB/sec. 22.5 MB/sec * 72,000 seconds (20 hrs) * 72 drives * 6 libraries = 683,438 TB, or about 32.6% of 2 PB. Therefore you need 3 times the original number of 6 libraries, and that's assuming you can keep up this rate constantly for every single second of 20 hours. 18 fully decked out 3584 libraries would be something to behold, I think ;) That'd span 106 frames and 1,296 LTO drives. A single full backup would run about $18M (list price) worth of tapes in that kind of configuration. I'd like to work for a place that can afford it! Even after deep discounting, that's still $8-10M worth of tapes. -Dan
Re: Help requested !!!
Hot Diggety! Murthy V Gongala was rumored to have written: I have a 3583 LTO library with 2 drives. TSM Server v4.2 on AIX. I have been running Scheduled backups without any problems so far. Yesterday the server started reclamation process for one of the Storage pools as shown below. 784 Space ReclamationVolume 053ABS (storage pool TECH_TAPE_POOL), Moved Files: 0, Moved Bytes: 0, Unreadable Files: 0, Unreadable Bytes: 0. Current Physical File (bytes): 10,487,404 Waiting for mount of input volume 053ABS (68221 seconds). Some of the activity log entries are : ANR1040I Space reclamation started for volume 053ABS, storage pool TECH_TAPE_POOL (process number 723). 10/06/02 14:06:19 ANR1044I Removable volume 053ABS is required for space reclamation. Even though both the tape Drives are free, and a scratch volume is made available, all the migration processes are Held up. Which AIX version? 4.3.3 or 5.1? And with which maintenance level patch set? (Ie, 4.3.3 ML10, 5.1 ML02 are the latest) Or if you don't know the ML set number, what version is listed in 'lslpp -l bos.rte.libc'? What does the output look like for: # lsdev -Cc tape And within TSM, what does TSM think: tsm q libv library name specifically, for the tapes that were mentioned in the actlog. You may want to try a library audit: tsm audit library library name checklabel=barcode Finally, how is the 3583 attached to the server? SCSI or FC-AL? -Dan
Re: Help requested !!!
Hot Diggety! Murthy V Gongala was rumored to have written: Level indicated by lslpp -l bos.rte.libc - 5.1.0.25 That sounds like 5.1 ML02 -- latest major patch set collection, very good. TSM shows the volumes as - Private. Ouch. Ok, the basic problem you're running in is that the TSM server *wants* at least one more scratch tape, but all the tapes you have right now is marked as Private. I think you haven't ruled out a tape communication problem yet, because TSM marks all unreadable or inaccessible tapes as Private. Most likely they are in the Private state because they already had existing backups, but I can't rule out a tape drive communication issue of some sort. Not running a FC-AL setup rules out a FC-AL switch failure issue, and since you say the setup's worked fine daily until now AND the 'lsdev -Cc tape' output shows all devices as 'Available' state, unlikely to be a drive communication issue, I think. I'm far from a TSM expert (many folks here are!) but I'm guessing you'd want to compact the data on the existing tapes to create enough free space for at least one or more scratch tape. May want to check tape utilization of each volume with: tsm q vol Are they at 100% of about 190 GB per tape? Or are there some tapes that have like 50%, 60%, something below 100%? You may need to look into several things: 1. 'expire inventory' TSM command that will mark expired files as being 'free' (reclaimable space). Run it daily after all of your backups has completed - can be done via a server-scheduled command that you enter only once. Then do another 'q vol' to see if that helped clear up more free space, enough for a 'move data' command to work. 2. look to keep a tight control on retention of data through adjusting management classes, policy domains, storage pools, etc... or by limiting what data you back up through inclexcl include/excludes in client's dsm.opt file. 3. do you have an off-site copygroup? (It can be an on-site copygroup or off-site copygroup or whatever you want to call it, but it has to be something in addition to your primary tape storage pool) If you do, then you will want to make a copy of your existing tape data into the offsite copygroup regularly -- once a week or whatever frequency works for you. But you need to do it regularly because it requires scratch tapes to do that, so you can't do that right now without clearing up some room first. Other folks will probably have better suggestions on how to get you out of this dug hole. My comments may help more with maintaining the setup to avoid this in future, once you're back to normal. I'm afraid I'm not experienced enough to give you the best answer for how to recover, but we've got a lot of good folks on this list who can give pointers. -Dan
Re: TSM 5.1 license registration failed?
Hot Diggety! Zlatko Krastev/ACIT was rumored to have written: look out carefully in 'q lic' output: Number of Managed System for LAN in use: 1 Number of Managed System for LAN licensed: 0 This ought to explain everything. Try 'reg lic file=mgsyslan.lic number=# of nodes you expect' D'oh. Thanks. ;) (I was wondering about where clients went, and looks like I confused mgsyslan with library sharing.) All good now! -Dan
Re: TDP for SAP R/3 on Solaris
Hot Diggety! Seay, Paul was rumored to have written: Is anyone using the TDP for SAP R/3 3.2.0.11 on Solaris 2.8 with Oracle 8.1.7 at SAP 4.6c2? Not I, unfortunately. No matter what we do we seem to have serious problems with restores and difficulty implementing the TDP in production. The restores hang about half way through. Backups usually work. We are just at wits end trying to figure out what is going on. Well, one place to start might be doing something like this: # truss -f -p dsmc pid -o /tmp/dsmc-truss.log Then get an UNIX expert to look at the output of that logfile, to see if there's any really obvious thing such as blocking on a particular resource or action. (I don't mean to offend -- I don't know how much you know about UNIX, and you did mention a mainframe background. If you're used to UNIX internals, then please accept my apologies. :) ) truss is the Solaris utility used to trace system calls performed by an application; sometimes it tells you exactly why something broke, but you need someone with an understanding of UNIX system calls, what looks normal and what isn't, and some experience in decoding truss output. It may generate voluminous logs, so it's good to run it for a few minutes or however long it takes to capture a good 'snapshot' of dsmc's activity. Can be started once the restore stalls. It's not guaranteed to give you the answer, but for the situation you describe (hanging), it's certainly what I'd have started with, myself. It might point to waiting on a response from the TSM server -- in which, it might be a TSM server-specific issue (threading, blocking on I/O, etc), or it might point to some other local oddity and culprit. Finally, Sun's constantly fixing a lot of all sorts of OS-specific things 'under the hoods' that may have interesting impact on applications... so there's always a chance that a recent OS patch set may fix it. -Dan
Re: TSM 5.1 client error
Hot Diggety! Mavis Jenkins was rumored to have written: So far, we've installed the server client on an RS6000 running AIX 4.3.3 The server is working fine if I use a 4.2 client from another server but if I use the 5.1 client installed on the same box as the TSM server, I get the following error when I try to run dsmc or dsmadmc: ANS0101E NLInit: Unable to open message repository 'dsmclientV3.cat' The file does exist and the NLSPATH points to the correct location. LANG is set to en_US What does your NLSPATH list? Should say something like this: NLSPATH=/usr/lib/nls/msg/%L/%N:/usr/lib/nls/msg/%L/%N.cat Then, next question: Do you have a 'dsmclientV3.cat' file here: /usr/lib/nls/msg/en_US/dsmclientV3.cat or /usr/tivoli/tsm/client/ba/bin/en_US/dsmclientV3.cat ? If it's a 32 bit TSM 5.1 client, you may want to see if this package is installed with: # lslpp -l tivoli.tsm.client.ba.aix43.32bit.common Fileset Level State Description Path: /usr/lib/objrepos tivoli.tsm.client.ba.aix43.32bit.common 5.1.1.5 COMMITTED TSM Client - Backup/Archive Common Files -Dan
Re: Does TSM use all DB Volumes for I/O?
Hot Diggety! Kilchenmann Timo was rumored to have written: I would vary much appreciate an answer to the question: Does TSM use all DB volumes for I/O (like round-robin) or does it fill a volume and then goes to the next one? I do not know for sure, because I don't know of a TSM way to report utilization statistics for db or log volumes on a per-volume basis. But what I can tell you that from my observations, it _DOES_ do some sort of round-robin on the diskpool volumes because it was almost perfectly even when I filled up two diskpool volumes for the first time, throughout the whole time. That does not answer your question about dbvols and logvols, I know, but it suggests that it might, since it does seem to do that for diskpool vols. I'm sure that someone here will know the definite answer. :) -Dan
Re: bad tapes
Hot Diggety! Alexander Lazarevich was rumored to have written: 1) is there any scenareo where a tape is still good, yet the server sets it to READ ONLY? Yes, there is. It can happen if you have a drive whose element ID mapping to rmt device name has gotten out of sync with each other. More common with new installations, but has happened to at least one other person on this mailing list when his FC switch broke and was replaced. It's pretty easy to distinguish from other cases because this will usually affect all or many tapes, rather than just a single tape. But I imagine the sheer majority of these incidents will normally be either a tape going bad or its label becoming unreadable. -Dan
Re: TSM 5.1 license registration failed?
Hot Diggety! Zlatko Krastev/ACIT was rumored to have written: - instead of grep-ping why you do not use `lslpp -L tivoli.tsm\*` or `tivoli.tsm\*license\*`. In this way you can see the version of filesets and ensure there is no one left from v4.2. You may also check reverse lookup - `lslpp -w /usr/tivoli/tsm/server/bin/library.lic` There's no TSM 4.2 filesets at all, as this was a fresh OS install (no preservation, only overwrite) and TSM 5.1 install straight from the 5.1 CDs. I did, however, double check as you suggested, just to make sure. Looks clean. - number of libraries licensed is one. Either you have two libraries or 3584 is partitioned and look like two libs or some other stanza is having Only one physical and logical library, no partitioning. insufficient number of registered licenses. Can you provide full output of `q lic`? Sure thing. tsm: GBLX-ROCq lic Last License Audit: 09/29/02 16:33:59 Number of space management clients in use: 0 Number of space management clients licensed: 0 Is Tivoli Disaster Recovery Manager in use ?: No Is Tivoli Disaster Recovery Manager licensed ?: No Number of TDP for Oracle in use: 0 Number of TDP for Oracle licensed: 0 Number of TDP for Oracle in try buy mode: 0 Number of TDP for MS SQL Server in use: 0 Number of TDP for MS SQL Server licensed: 0 Number of TDP for MS SQL Server in try buy mode: 0 Number of TDP for MS Exchange in use: 0 Number of TDP for MS Exchange licensed: 0 Number of TDP for MS Exchange in try buy mode: 0 Number of TDP for Lotus Notes in use: 0 Number of TDP for Lotus Notes licensed: 0 Number of TDP for Lotus Notes in try buy mode: 0 Number of TDP for Lotus Domino in use: 0 Number of TDP for Lotus Domino licensed: 0 Number of TDP for Lotus Domino in try buy mode: 0 Number of TDP for Informix in use: 0 Number of TDP for Informix licensed: 0 Number of TDP for Informix in try buy mode: 0 Number of TDP for SAP R/3 in use: 0 Number of TDP for SAP R/3 licensed: 0 Number of TDP for SAP R/3 in try buy mode: 0 Number of TDP for ESS in use: 0 Number of TDP for ESS licensed: 0 Number of TDP for ESS in try buy mode: 0 Number of TDP for ESS R/3 in use: 0 Number of TDP for ESS R/3 licensed: 0 Number of TDP for ESS R/3 in try buy mode: 0 Number of TDP for EMC Symmetrix in use: 0 Number of TDP for EMC Symmetrix licensed: 0 Number of TDP for EMC Symmetrix in try buy mode: 0 Number of TDP for EMC Symmetrix R/3 in use: 0 Number of TDP for EMC Symmetrix R/3 licensed: 0 Number of TDP for EMC Symmetrix R/3 in try buy mode: 0 Is Library Sharing in use: No Is Library Sharing licensed: No Number of Managed System for LAN in use: 1 Number of Managed System for LAN licensed: 0 Number of Managed System for SAN in use: 0 Number of Managed System for SAN licensed: 0 Number of Managed Libraries in use: 0 Number of Managed Libraries licensed: 0 Tivoli Data Protection for NDMP in use ?: No Tivoli Data Protection for NDMP licensed ?: No Server License Compliance: FAILED I haven't been in the computer room yet because that's where the TSM 5.1 CDs are (but will, shortly) and then I can try Paul's suggestion, also. Thanks, much appreciated all ideas on this so far. :) -Dan
Re: Withdrawal of 3570 Magstar
Hot Diggety! Emil S. Hansen was rumored to have written: How about setting the mount retention for the devclass to something like 10 to 30 mins? That will keep the tape mounted for at least 10 mins after the last access, so that if the tape is needed by the client it will likely still be mounted. That's just the catch; mount retention was indeed set to 10 minutes. Yet, according to q actlog, TSM 4.2 was in a big hurry to dismount it for no apparent reason. I'm still rebuilding the 5.1 setup, and will retry the same tests once ready. -Dan
Re: Withdrawal of 3570 Magstar
Hot Diggety! Dan Foster was rumored to have written: Hot Diggety! Emil S. Hansen was rumored to have written: How about setting the mount retention for the devclass to something like 10 to 30 mins? That will keep the tape mounted for at least 10 mins after the last access, so that if the tape is needed by the client it will likely still be mounted. That's just the catch; mount retention was indeed set to 10 minutes. Yet, according to q actlog, TSM 4.2 was in a big hurry to dismount it for no apparent reason. I'm still rebuilding the 5.1 setup, and will retry the same tests once ready. Well, appears that it now works correctly for me in TSM 5.1 -- it does not unmount tape after the write completes... keeps it available in the tape drive for any subsequent writes. No need to set KEEPMP=YES in the client node definition. Don't know what was wrong with TSM 4.2; I set up 5.1 with the identical settings (I'd documented the step by step install and followed it) except with some updated commands such as for library, drive, and path configs. Oh well. Works great now. :) I'm liking what I see of TSM 5.1 so far. -Dan
TSM 5.1 license registration failed?
I'm stumped by something (simple?) relating to TSM licensing. We've got a pretty plain vanilla TSM 5.1 setup with a 3584 tape library. No additional features purchased or in use (SAN, TDP, NMDP, etc). The 3584 requires a managed library license, as I understand it (has 12 drives and 610 tapes spanning the L32 and D32), so we paid for that license, along with the number of client nodes we wanted to use. Didn't receive any paperwork or anything special -- just TSM 5.1 CDs with all the filesets that were appropriate. So then, once I've got the 5.1 server set up... went to do: tsm register license file=library.lic tsm query license Now says: [...snip...] Number of Managed Libraries licensed: 1 Server License Compliance: FAILED What did I do wrong? Why did it fail? How do I fix so that I can have both 1 managed library AND server compliance = SUCCESS? The documentation is extremely sparse in this area, to put it charitably. ;) Did I install the wrong filesets for AIX 5.1 server? For the license, I have these filesets installed: # lslpp -l|grep tivoli|grep -i license tivoli.tsm.license.aix5.rte64 tivoli.tsm.license.cert5.1.0.0 COMMITTED Tivoli Storage Manager License This is for a brand new TSM 5.1 install on a brand new AIX 5.1 install. No existing nodelock file was present, nor were anything restored. The only thing I wonder about, after looking at the library.lic file is a line that says this: ProductVersion=4.2 Does that mean the license certificate file is specific only to TSM 4.2? Finally, in previous versions (ADSM 3.1, TSM 4.2, etc), I had to register an appropriate license for the number of nodes I'd paid for. But there's no client license certificate files nor anything mentioning it in the TSM 5.1 'q license', nor in the 5.1 docs...? Does that mean I continue to pay for client node licenses (I'm not here to rip anyone off!) but just simply don't need to register them with the server...? -Dan
Re: Withdrawal of 3570 Magstar
Hot Diggety! Coats, Jack was rumored to have written: Try 3 minute :( to get a LTO loaded and started spinning ... great for bulk store, but not up to 'interactive' response needs :( ... Using LTO for HSM would seem counterproductive IMHO. 3 minutes?! Something sounds wrong there. I've got a 3584 with 12 LTO drives attached to host system via SCSI, and it takes 3-9 seconds to have the robot fetch the tape from storage slot, move it to drive, load it in the drive, and about 46 seconds to read the volser on the tape and other mount-related processing, for a total of about 55 seconds from issuing command to load to start using a tape. We've got a L32 and D32 frame; guess it could conceivably be a little longer in worst case scenarios if had a decked out setup (one L32 and five D32s) but I can't see it being much more than 1m to 1m10s or so. However, that 55 seconds is a little painful in certain situations. For example: client-disk stgpool (10MB max)-tape stgpool Client sends data to TSM server As long as client data is 10MB, streams to disk pool Soon as client sends a 11 MB or 500 MB file, then... TSM fetches a tape, mounts it (55 sec wait) TSM starts writing file to tape Once done, TSM dismounts the tape (!) Client continues sending data If it hits another 10MB file, the whole mount-write-dismount process repeats. This results in a significant performance hit from all the mounts/dismounts. To alleviate this, I've set the node's KEEPMP option to YES, so it ends up mounting the tape once on first access, then keeps it mounted throughout the entire client run so that we get no more subsequent 55 sec mount delays. When does it dismount the tape? After the tape retention in drive period expires, but usually the next client session grabs the same tape if it's got free space. The above was with a TSM 4.2 setup; I just installed TSM 5.1, so I've got to retest to see if this is any different without the KEEPMP option being enabled. Anyway, I'd strongly urge you to get that 3 minute wait looked at! -Dan
Re: tapes getting marked PRIVATE
Hot Diggety! Seay, Paul was rumored to have written: Suppose when the tapes were labeled the drive that labeled them was bad and now none of the drives can read them. I had that issue pop up while setting up the 3584, and the culprit was when the /dev/rmtX and element ID mappings had fallen out of sync. End result? TSM loaded tape by element ID, then tried to access drive via its incorrect rmtX mapping -- so it hit on a drive that didn't have the tape loaded, got a read/open error, moved the tape back to its storage slot (via element ID), marked it as Private in the TSM db, and repeat for the next tape endlessly until TSM finally timed out this entire process. Fixing the mappings instantly fixed that particular problem. So I'd be more inclined to think that or some issue with DEVCLASS definition was the culprit since one of the original poster's key words were '...all tapes' -- if it was a bad tape, would have been an one-off. -Dan
3584 tape questions
I've finally got the new 3584 library up and worked through all hardware and TSM issues, and just finished with the first round of successful tests. Looks sharp! Currently tuning the setup (disk, memory, network, TSM, etc). Environment: AIX 4.3.3 ML10, pSeries 660-6H1, TSM 4.2 (5.1 next week) to 4.2.2.12, SCSI attachment to host, 3584-L32 and 3584-D32 frames. Two questions about 3584 tapes -- 1. Is it normal for it to take 45 seconds to read/verify the label on the tape? The tape load itself is pretty quick, but takes forever-and-an-half to read/verify the label. 2. There's two label sources -- one is the barcode scanner, which works great and *very* quick, and it seems the other label source is on the tape somewhere. Where exactly on the tape is the label preserved, other than the volser bar code label? Is it in the LTO-CM area of the tape? Or is it on the first part of the tape, or something? I'm curious about this because the implications is that if I do destructive read-write tests with tapes involving overwrite, I might potentially overwrite the label, and have to relabel it? The documentation suggests that 'LABEL LIBVOL' for a 3584 would update label in the LTO-CM area of the tape cartridge, but it isn't real definite or clear on that point. -Dan
Re: Help for 2108 failure
The other possibility is that since the 2108 was replaced, there may be a small chance the /dev/rmtX mapping no longer matches what the previously assigned rmtX-to-element ID mapping was. Which conceivably could produce this sort of error, if TSM and the library disagreed about what logical/physical drive matched up? What I'd suggest is: tsm q drive f=d Then note each rmtX device name and what element ID it corresponds to. (in TSM) Then do some tracing to figure out which element ID each rmtX device actually physically corresponds to. Then you'll soon see if the TSM setup and the probed devices in AIX matches or not. If they don't match, update the TSM drive entries for rmtX and element ID ASAP. Should normally take only about 5-10 minutes of work to figure out the mapping from scratch. Keep good notes and issue a whole bunch of commands to assist in the logical-to-physical tracing. I just finished fixing all this after no end of fun with the 3584 setup and mismatched rmtX/element IDs. -Dan
Re: baffling tape, 3584 HW compression, and stgpool design
Hot Diggety! David Longo was rumored to have written: My experience with tapes in this condition is that the write protect tab was set on by accident. AS Dwighty check the tape out. Then check the Write Protect. If it is on, then set it off and check the tape back in as scratch. Looks like two bad tapes in the 3570 library. Thanks! Going to store them in the 'rejects pile' until I can legally have them degaussed and then disposed of. -Dan
Re: baffling tape, 3584 HW compression, and stgpool design
Hot Diggety! Koen Willems was rumored to have written: Maybe that the access state is unavaileble try q vol f=d look at the access state if it is not in a readw state use update vol access=readwrite Then do a move data on the volume to see if the remaining data moves Do a del vol and or a checkout if you want to lose the volume. Apparently, it was bad tapes. Compression is automaticaly used when using format is drives. Interesting. Guess that makes sense. (Do not use client compression) Nah, wasn't planning to. The backups are fast enough that we wouldn't have any real benefit to doing client compression. Storage pools give a beter restore performance on file data if the contain less volumes ( this cuts on mount and search times ) Ahh, makes sense. Make a pool for priority restore nodes ( if possible with collocation ) Make a pool foor non priority nodes without collocation Make a pool for every type of TDP you install. ( colloction will cost you tapespace but will speed up restores ) ( the less volumes an uncollocated storage pool will have the faster the restores ) ( TDP pools have a beter restore performance because op the large sequentiall backups that kan stream back when restoring) Makes sense. Good tip about TDP pools, thanks. Definitely will keep these separate. Do not leave data on your diskpool longer then necessary. Further more one does not want to have backup data on a diskpool for longer than one backup cycle. after backup sessions do a: update stg diskpool hi=0 lo=0 and move data to tape and then offsite. Sounds good. ;) I'm looking to concentrate all client backups in a specific window, then have the scheduler fire off a daily job to do the migrations (disk-tape, tape-offsite copypool) with some basic error checking, and also do other things such as expire inventory, dbbackup, etc. -Dan
Re: baffling tape, 3584 HW compression, and stgpool design
Hot Diggety! Zlatko Krastev/ACIT was rumored to have written: David already answered to question 2 but as an additional remark - when TSM started to write a tape volume uncompressed the tape has to become back scratch to start write on it with compression. Very interesting tidbit. I'll note that and file away for future reference. Thanks! -Dan
baffling tape, 3584 HW compression, and stgpool design
Couple unrelated questions: 1. When I query a 3570 library, there's one tape that baffles me: adsm q libvol 3570lib1 Library NameVolume NameStatus Last Use Home Element ----- 3570LIB1133060 Private 34 I can't do anything with it -- ie, I can't discard the data on it because ADSM says it doesn't belong to a storage pool, and yet, no indication that it's a DbBackup. All other tapes in the library are sane and belongs to either scratch, a storage pool, or a dbbackup. So what could it be, and how do I wipe that lone offending tape and change its status to scratch? (This is with ADSM 3.1+patches) 2. How do I enable hardware compression for LTO tapes? Some sort of device class or drive definition parameter? (This is with TSM 4.2.2+patches) I've looked through docs set and not finding anything that addresses this. I'm currently using: tsm DEFINE DEVCLASS 3584_DEVCLASS1 DEVTYPE=LTO FORMAT=DRIVE- MOUNTLIMIT=10 MOUNTWAIT=60 MOUNTRETENTION=0 LIBRARY=3584LIB1 tsm DEFINE DEVCLASS 3584_DEVCLASS2 DEVTYPE=LTO FORMAT=DRIVE- MOUNTLIMIT=2 MOUNTWAIT=60 MOUNTRETENTION=0 LIBRARY=3584LIB1 The mountlimit of 10 and 2 is to put an hard upper bound on number of drives that the backups (10) and offsite copying (2) can use. The mountwait of 60 minutes is to avoid having jobs fail if they're sitting there, waiting for a particularly huge 300GB job to finish and free up a drive for it to use. Mountretention=0 is because this is a fast automated library (3584 with 12 drives); I can see setting it to 10 or so if it was a smaller human operated library or requiring operator tape swaps like the 3570 library. 3584LIB1 refers to the library, obviously. ;) (device /dev/smc0, etc) As I understand it, FORMAT=DRIVE means it uses the drive's compression settings. I haven't seen any explicit mention of what this defaults to nor how to adjust it in either the 3584 or TSM docs. Or would I use something like DEVTYPE=LTOC...? 3. Is there really a reason to use multiple storage pools for the same tape repository? Ie: the TSM 4.2 docs suggests an out-of-box default setup such as DISKDIRS, DISKDATA that then migrates to TAPEDATA, which then copies to OFFDIRS and OFFDATA (offsite copy pool tapes). This makes sense. (I understand the purpose for each and every one of them, and how they're organized.) Is there a good reason why someone might want to split up diskdata and tapedata into multiple stgpools? I ask because I've heard references to other folks having multiple stgpools, and wondering what I'm missing here. That's the only thing I think that eludes my understanding in preparing the new design, at this point. Any insight or comments much appreciated. Thanks! -Dan
SNMP MIB for TSM 4.2?
Does anyone know where I can get a SNMP MIB definition file for TSM 4.2? I know it's capable of SNMP support, but in order to meaningfully use it, need a MIB to plug into the monitoring system and I don't seem to be able to find a MIB anywhere. -Dan
Multiple logical libraries for 3584
I've got a question. One 3584-L32 with 6 drives and one 3584-D32 with 6 drives. Is it possible to have a logical library that covers 4 drives in the L32, and a second logical library that covers last 2 drives in the L32 and all 6 drives in the D32? Or is that not a valid configuration -- ie, do I need to keep logical libraries from spanning multiple drives in multiple 3584 units? -Dan
Re: SNMP MIB for TSM 4.2?
Hot Diggety! Mark Stapleton was rumored to have written: From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED]]On Behalf Of Dan Foster Does anyone know where I can get a SNMP MIB definition file for TSM 4.2? I know it's capable of SNMP support, but in order to meaningfully use it, need a MIB to plug into the monitoring system and I don't seem to be able to find a MIB anywhere. How about adsmserv.mib, in your server subdirectory? D'oh! Don't know how I didn't see that. Thanks ;) -Dan
Re: Multiple logical libraries for 3584
Hot Diggety! Don France was rumored to have written: Yep (to the last question); you cannot span multiple physical libraries to make a single logical. You can define multiple logicals within a single physical; that is a common thing I've done, for various reasons. Darn. ;) But makes sense, all considering. Thanks for confirming -- couldn't find explicit restrictions except for a few other items in the manuals. Much appreciated! -Dan
Re: Multiple logical libraries for 3584
Hot Diggety! Mark D. Rodriguez was rumored to have written: Judging by your configuration I beleive that you have just one physical library. I beleive the L32 is the base unit and the D32 is an expansion cabinet, i.e. the 2 of them are physicaly attached to one another and share the same robotics. Is this correct? If so then yes you can partition the library as you described. Aye, that is correct, from what I understand of the setup. But the real question is whatare you trying to accomplish? If this is to be connected to 2 different servers there are a few more things that have to be in place. If both of these logical libraries are to be used by the same TSM server I am not sure I understand the rational for doing so. You could just as easily manage the drive utilization through other TSM server options. Please explain a little bit more what you are trying to accomplish. Basically I'm trying to split on-site vs off-site (copypool) tapes, with each group having its own drives and pile of in-use (private) vs spare tapes. Currently we do it with a 3575 for on-site tapes and a 3570 library for the off-site tapes, and it avoids significant potential handling issues. Not wanting to give Murphy more chances to possibly break things anywhere due to improper or mistaken tape handling, basically. It also protects availability for doing client restores as well as gives us a guaranteed number of drives/tapes that can be used at any given time for doing daily backups. Gives me some hard guarantees about available resources that I can then plan the entire operation around. Connection is to only one server. I just didn't think it was possible to safely handle multiple TSM libraries via a single media changer interface (/dev/smc0) and from reading the literature, was led to believe that multiple logical libraries was required (each one providing a smc device) in order to control the drives in each partitioned group of tape drives, as well as for determining what tapes were in a pool and available for spare use. -Dan
Re: Performance of TSM 4.1.4 with Solaris
Hot Diggety! Charles Anderson was rumored to have written: 500% increase?! We are using UFS right now and not RAW. I have been a little hesitant to migrate to RAW though. Any suggestions for keeping it as painless as possible ? What did you do, just set your migration threshold to 0, backup the log, backup the db, repartion the disks, restore above from the backups? What did you do about getting the data back onto the disks? I believe Veritas has a product that allows you to continue to use ufs filesystems, but has a special driver that completely bypasses the double-buffering, which gives you the same performance as raw logical volumes, but with all the integrity protections of ufs. May be worth investigating if you'd like the best of both worlds. -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: Keeping an handle on client systems' large drives
Hot Diggety! Seay, Paul was rumored to have written: Ask them where they were on 9-11-2001. Are they totally brain dead? Ahhh, so that's what you referred to in passing in the other post. That's all right, and understandable. I have a first rate appreciation of this. If you'll allow me to indulge briefly on a tangentially related (but not completely) issue on this list, just once... I used to be a VMS admin. Best, most robust OS that I ever worked with - probably true for the IBM mainframes but didn't work much with them, alas. (A little OS/400, DOS/VSE, and one or two other related OSes) Anyway, come post-9/11, a *lot* of financial firms were in a world of hurt. The ones who planned and re-tested over and over again, each year, for an alternate site a good distance away from NYC, was able to reopen for business only a few days later. Many were based in NJ or about an hour west/north of NYC... one was even based not too far from home, their DR site being about 4-5 hours northwest of NYC. Around this time, I heard that Compaq (company that bought out DEC) was making a lot of frantic calls all around the country seeking out high end machines such as the AlphaServer 8400s and VAX 7000s...that had been discontinued for perhaps 10 years since, because a lot of customers were suddenly calling in for warranty replacements (under their expensive support contracts) in NYC and DC -- you can guess what kind of customer it was in DC. How desperate was Compaq? They were calling up even third level resellers of used equipment that they would normally never ever think of talking to. Compaq was in a nasty hole, because they had run out of set-aside reserve spares. Fab plants *long* since shut down...they can't just take the original plans and re-fab, since the engineers no longer there... I'm not sure how they eventually resolved that... probably offered newer machines to customers and provided migration assistance at Compaq's cost, is my guess. But what the bean counters don't realize is that it doesn't take a catastrophic national event to mean a bad effect on the business bottom line, which I find unfortunate. Can be all sorts of more 'mundane' (albeit not very common) events such as that train which burned in a Baltimore tunnel and closed a part of downtown near Oriole Park at Camden Yards. My company (used to also own a telco) was personally affected by an homeless man burning something in a former abandoned railroad tunnel that melted fiber optics and took out OC-12 to the area for 12+ hours, with a nice number of servers based out of here. It doesn't have to be a corporation for a nasty disaster to mean bad things for their bottom line. I am very well reminded of a colossal failure at an academic institution almost a decade ago that was a chain of events ultimately resulting in failure of a critical drive in a RAID-5 array, and the tapes weren't really usable for recovery...which they found out the hard way. An entire semester of classwork was effectively disrupted, with much data lost, before they were finally able to convince DEC to send out the very best people to recover about 80% off the RAID-5 array through some custom work. So many classes, projects, research papers, etc. were affected that it just simply isn't funny. Same place where if the IBM mainframe ever went down, school was closed for the day. (Happened only once ever, to best of my knowledge.) ...and that is truly unfortunate, that the people who are actually tasked to make things happen, like us, understand and appreciate, whereas others higher up may not share the same view, knowledge, and experience. In a D/R scenario, it also behooves you to know your power sources, how they kick in, at what levels, how fast/when, evacuation plans, how to config PBXes, have emergency equipment handy (eg flashlights), and a million other details. Hardware that can be quickly hooked up/activated, written step by step plan nearby, software CDs handy if needed, dry runs done, backups/restores/app operation verified, and all of this tested once or twice a year depending on level of need and impact, etc. Still, I resolve to do my best to do whatever I can realistically do. :) With that said, I now return you to the normal *SM discussions. ;) (with the reason for copy stgpools driven home ;) ) -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: TSM 4.2/AIX setup questions
Hot Diggety! Miles Purdy was rumored to have written: 1) Is there any particular reason to set a max file size for a disk stgpool? (Assuming a setup where disk stgpool will migrate to tape stgpool) Generally yes, if you will be backing up a file larger than the stgpool. If you will be backing up a file larger than the stgpool it may be more efficient to send it right to tape, this is the general rule. Ahhh, efficiency...makes sense. Also explains why there might be a situation where client-disk [but too big or unavailable]-direct to tape instead, so the copy stgpool might not have every single file in the disk stgpool, hence this being the reason for backing up tape stgpool to the copy stgpool in order to fill in any gaps (in addition to backing up disk stgpool, also). 2) Should the TSM server have its own stgpool for backing up itself? A. I don't think so. Do you mean the filesytems on the TSM/AIX server? No, it doesn't _need_ its own. Ok. Must've been an original design decision at this site for the existing setup. Can't say I know the rationale (and can't ask since the folks in question are long since gone now), but I'll keep that in mind. We set up the existing scheme about 5-6 years ago, and in this high-turnover industry (ISP), people moved on every few years when the grass was still green. ;) 3) I've heard mixed things about 358x firmware version 22UD... I think we have 18N2 (but not near it right now to confirm), although what I've heard about 22UD is generally (but not 100% in agreement) positive. Stable? A. I'm using 22U0, things seem good. Hmmm. Duly noted, thanks. 4) Whom is supposed/allowed to upgrade firmware? IBM CE only? It depends how comfortable you are performing the work. I'm good with any sort of firmware updates from SP nodes to disks to disk arrays to tape drives, etc. But for this one time, I think I'll let the CE do it the next time he's on-site for the 3584. :) No reason to throw caution to the wind. 5) The only docs for firmware upgrade references a NT/2000 box and the NTUtil application, whereas I'm in an all-UNIX (AIX and Solaris, although I do have a laptop with Linux and Windows XP if need be) environment, so wonder how to upgrade the firmware without Windows if it's even possible. A. You can upgrade the firmware from a drive (drive to drive) or from a tape cartridge. Check the docs for the main panel. Ahhh, yes. See it now, thanks. 6) To *SM, all backups are incrementals (except for the first backup of a new client), is my general understanding. Is there a way to force a full backup of a particular client as an one-time operation? I'm guessing maybe not, but thought I might try asking, anyway. :) A. This is called an 'archive'. There is plenty of docs on this. Ahhh, sure is! Archive, being TSM's term for that... okay, can deal with that. Back to reading up more on the archive section. 7) The biggest single question... I don't have a real good understanding of the purpose of copy stgpools. I've read a lot of documentation -- hundreds of pages of multiple docs, re-read, read old adsm-l mail, Google searches, etc... but still just don't quite 'get it'. I can set up HACMP clusters, debug really obscure things, but this eludes me. ;) A. Copy pools are for offsite. Copy pools are what is in your vault. A copy pool is a complete copy, usually offsite. Ahh, that makes much more sense, thanks. (And rest assured, we *do* have an off-site storage vendor and bring tapes off-site. :) ) Almost, you do send the (primary) data to another (copy) pool, but all the time. I would do something like this, everyday: 1. backup your clients to disk (disk storage pool) 2. make a copy to a copy storage pool (OFFSITE) 3. backup from the disk storage pool to your primary storage pool 4. backup your database, devconfig, volhist 5. send the OFFSITE tapes and the database backup offsite 6. during the day run reclamation on each storage pool Makes sense. Hmm, I saw a suggestion in the TSM 4.2 admin guide to disable reclamation for the copy stgpool (specifically) to avoid a situation where tapes gets sent off site and then it wants them for the reclamation. But of course, for the other stgpools, reclamation does make sense. Much appreciated the pointers...big help, thanks! -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Keeping an handle on client systems' large drives
I've always been curious about something. How do you keep an handle on the fact that commodity PC storage is growing at a far faster rate than tape capacity/system is? For example, if I had a small LAN of about 300 PCs -- let's say, an academic or corporate departmental LAN environment... each has at least a 40 GB HD, and probably a fair amount of apps and files on them. In the stores, I see drives up to 160 GB, with even larger ones on the way! So let's say, an average of 25 GB utilization per system... a single full backup would be about 7.5 TB, which is quite a few tapes ;) Not everybody is using LTO or higher capacity. So do those sites rely purely on the incrementals to save you? Or some site specific policy such as tailoring backups to exclude (let's say) C:\Program Files, or some such...? Just wondering. Not every site is lucky enough to be able to convince the beancounters the merits of having a backup system that keeps up with the needs of the end users, even if it means one has to explain doomsday predictions on the business bottom line -- they invariably hear that then say Oh, pshaw, you're just exaggerating because you want money It sucks to be the one that's right ;) And the ones who warns well before a nasty event occurs may also be the first one to be fired out of spite after something happens and gets the blame for not having prevented it. -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: TSM 4.2/AIX setup questions
Hot Diggety! Miles Purdy was rumored to have written: NO! Remember that a copypool is a complete copy. So everything that is offsite is onsite. TSM does not need the offsite tapes to reclaim them! Ah, that is a good point, when put that way. ;) But you are correct: You could send a tape out Monday, 10% used (90% percent reclaimable). Reclaim it during the day (Monday), and ask for it back Tuesday. With LTO tapes this can be mitigate by using a high reclaim, like 90%, or 10-to-1. Or run reclamation less often, like every second day. If you don't run reclamation you could have 100GB tape with 1KB used, stuck in your vault. Ick, that'd be really nasty and an huge waste. Duly noted, thanks. Also make sure that you don't use collation on a copy pool, unless this is what you REALLY want. In general I don't think you want to you. Not using collation will keep fewer tapes, more full. Preventing tape thrashing, if you like. Indeed. I read about that...sounded great at first, but then soon realized that potential for tape thrashing was probably going to be too great given how we manage the tape environment here. So it's definitely not going to be enabled. (But I appreciate the TSM folks having made that an available option. Options are always great to have, even if not used.) -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications Internet: [EMAIL PROTECTED]
Re: Busy log files
Hot Diggety! Thomas Denier was rumored to have written: It depends on the purpose of the log files. Some applications append When put in that way, does make a lot of sense why the TSM behavior would be have options to control that. Fortunately, the log files in question is non-database, and used for accounting; works by continually appending entries. planning, or accounting. This type of log can usually be dealt with by using an include statement in the include/exclude file to bind the log to a management class whose backup copy group is defined with 'serialization=dynamic'. This will cause TSM to read the file once Groovy, thanks. (Also appreciated the other poster whom also suggested looking at the serialization options.) -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: Keeping an handle on client systems' large drives
Hot Diggety! Seay, Paul was rumored to have written: What you have to do is revisit what you are saving and put in exclude.dirs for all directories that contain software that can be rebuilt from a common desktop image (hard drive replacment). Have your users save their documents in specific folders and only back them up. Then they just have to customize their desktop configure their node name in the dsm.opt and restore the stuff that is backed up. This is the trade-off. Makes sense. Basic education + cost saving vs expense from a brute force approach. The trick is to have education that works well for a wide range of users, with differing expertise, and to also clearly communicate expectations (if you save anywhere else, you won't get it back!). Now that sounds like I also have to train them to not just blindly click whenever an application offers them a default directory (often within app area) to store documents in. Perhaps a small data area carved out on the hard drive, like say, 5 GB partition for user documents as Z: or whatever, and similiarly for other platforms (/userdocs/user as a symlink from ~user/docs or whatever), to provide a consistent and easy-to-use area for end user, yet predictable area for mass-deployed *SM configurations to use. I'm sure that the IT shop can help out significantly if they're able to preconfigure these settings within each application before users gets their hands on the machine. Hard part is when not every place has that luxury, especially at smaller places where end users may be configuring everything on their own. Anyway, the overall education/training approach is definitely cheaper than having to save everything on the HD, I do agree. ;) -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
TSM 4.2/AIX setup questions
Environment: TSM 4.2.x on an AIX 4.3.3 ML10 server (660-6H1) with a 3584-L32 and a 3584-D32 expansion frame. In the process of setting it up, which is a luxury that I'll have only once to get it right. :) So, some questions (since I am still coming up to speed on *SM). 1) Is there any particular reason to set a max file size for a disk stgpool? (Assuming a setup where disk stgpool will migrate to tape stgpool) 2) Should the TSM server have its own stgpool for backing up itself? 3) I've heard mixed things about 358x firmware version 22UD... I think we have 18N2 (but not near it right now to confirm), although what I've heard about 22UD is generally (but not 100% in agreement) positive. Stable? 4) Whom is supposed/allowed to upgrade firmware? IBM CE only? 5) The only docs for firmware upgrade references a NT/2000 box and the NTUtil application, whereas I'm in an all-UNIX (AIX and Solaris, although I do have a laptop with Linux and Windows XP if need be) environment, so wonder how to upgrade the firmware without Windows if it's even possible. 6) To *SM, all backups are incrementals (except for the first backup of a new client), is my general understanding. Is there a way to force a full backup of a particular client as an one-time operation? I'm guessing maybe not, but thought I might try asking, anyway. :) 7) The biggest single question... I don't have a real good understanding of the purpose of copy stgpools. I've read a lot of documentation -- hundreds of pages of multiple docs, re-read, read old adsm-l mail, Google searches, etc... but still just don't quite 'get it'. I can set up HACMP clusters, debug really obscure things, but this eludes me. ;) What I want to do is: client - TSM server - disk stgpool - (automatically migrate to tape based on space utilization of disk stgpool) tape stgpool That's the general concept of what I want to achieve. Is a copy stgpool really needed, to be attached to either one of the primary stgpools? I was under the impression that a copy stgpool was something you wanted when you wanted to copy a primary stgpool so that you could send it to another stgpool when ready (based on whatever trigger...space, date), such as in a disaster recovery scenario? -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: Recovering SP node to Standalone RS6000
Hot Diggety! Jolley, Bill was rumored to have written: I have a TSM server located on an IBM SP Node (winterhawk) and would like to recover to a standalone RS/6000 (H80). Do anyone have/know of a procedure or have suggestions? Offhand, best bet is probably to do a mksysb backup (making sure that device drivers for both boxes exists, among other key gotchas) and then a mksysb install from that image. You can't just do a normal file backup and then restore because there are a bunch of system specific stuff such as the ODM entries, installed filesets, etc. Even more critical to get it right when SP and non-SP is concerned. I haven't personally done that with our SP and non-SP nodes as you describe, so that's about all I can suggest from what I've heard from others who has done similar things in the past. It's usually done as a way to recover from a total SP node system failure, rather than migrating to or from the SP, though. -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: UNABLE TO DELETE FILESPACE
Hot Diggety! Stephen Pole was rumored to have written: Hello felloe ADSM/TSM'rs, We are trying DELETE FILESPACES belong to a NODE so that we can REMOVE the NODE. One way to delete filespaces indiscriminately if you're just going to remove the whole node, is to do: DELETE FILESPACES nodename * TYPE=ANY -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications
Re: mailbox.pst
Hot Diggety! Mark Stapleton was rumored to have written: As discussed back a month or two ago: 1. Load kill.exe from the appropriate Windows resource kit. 2. Create a preschedcommand line that runs drive:\path\kill.exe outlook.exe 3. Run the backup. The .pst file will be available for backup. Those pesky users. How dare they keep Outlook open when they lock the workstation for the night... :) Perhaps it would also be prudent to, between steps 2 and 3, to insert some way of notifying the user about the pending kill 10, 5, 1 minute beforehand (and tool should be intelligent enough to auto-dismiss the dialog box if no response after a timeout period). I'm just afraid that the proposed solution, as is, may really anger someone who was working on something important during a late night work session, and seeing it all disappear. It's especially not fun getting realy angry calls from execs. :) And then you've got some sites that runs call centers, and so forth. (24x7 operations, essentially.) -Dan Foster IP Systems Engineering (IPSE) Global Crossing Telecommunications