Re: TSM altering Windows access time?
On Mon, Oct 30, 2006 at 11:38:17AM -0700, Ben Bullock wrote: > http://as400bks.rochester.ibm.com/tividd/td/TSMC/GC32-0787-02/en_US/HTML > /ans10008.htm > [...] > If only the following items change, they are updated without causing the > entire file to be backed up to the server: This means that the TSM client won't send a new copy of the file to the TSM server, but just tells the TSM server the new attributes of the file. The TSM server then destroys the old attributes, and stores the new ones. This can be a problem when a backup runs after a runaway chmod (or the equivalent on your OS): you cannot restore the old permissions without reverting to older versions of the file (if there are any). -- Jurjen Oskam
Re: TDP Oracle
On Fri, Oct 20, 2006 at 12:10:17PM -0700, Gill, Geoffrey L. wrote: > 3.If this ignores the retention values set in the management class > settings, and I believe it still does, do I need a specific management > class for them? A TDP for Oracle node *only* has active files on the TSM server, as Oracle by default guarantees that only unique filenames are sent. This means that the retention parameters concerning "extra versions" are irrelevant. In addition to that, when Oracle RMAN deletes a backup, it doesn't only delete the file on the TSM server, but also the information it has in the RMAN repository required to restore that file. So, while you can set up a TDP node such that inactive files will be retained on the TSM server, it doesn't make sense because the file is useless without the RMAN repository information, which is already gone. The retention values in the management class are certainly not ignored. If you don't set them correctly (as described above), you're almost certainly storing too much data on the TSM server. > 1.Can we use a domain we have in place already and if so are there > any settings we need to look at? > 4.I am assuming we should use a specific node name for these > backups and not let them use the node name that is now backing up to the > server. Still correct right? You can use an existing domain, but it's much simpler to use a dedicated TDP for Oracle domain. You only need to set one managementclass containing one copygroup in that domain. That way, it's guaranteed that the RMAN backups are bound to the correct management class and you don't have to fiddle with include statements in the API configuration on the client. And you should indeed use a specific node for the TDP backups. The TDP node goes in the TDP domain. > 2.I believe expiring of objects is done from the RMAN side. If > this is correct how do I know they are keeping, or maybe better > described as deleting, what they should? You need access to the RMAN repository ('recovery catalog') for that. The RMAN repository can be located in the database itself, or be a database on its own. Normally RMAN deletes all obsolete objects (as specified by the DBA) on its own, but there are situations where RMAN doesn't know that a certain file is stored on the TSM server. That's where the tdposync tool comes in. It checks the RMAN repository, and then checks what's available on the TSM server. If there are files in TSM which RMAN doesn't know about, it can delete those files on the TSM server. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TDP for Oracle on AIX...configuration
On Fri, Oct 20, 2006 at 02:11:34PM -0400, Lawrence Clark wrote: > tdpoconf password -TDPO_OPTFILE=/home/oracle/ts You should point to the optionsfile itself, not the directory containing it. This means: "tdpoconf password -tdpo_optfile=/home/oracle/ts/tdpo.opt". Regards, -- Jurjen Oskam
Re: Backup Retention Period
On Wed, Sep 27, 2006 at 09:40:30AM +0200, Loon, E.J. van - SPLXM wrote: > For backups: yes. Backup objects are rebound to the new managementclass > during the next backup, so all inactive versions older than 30 days will > be removed. The original poster didn't mention assigning files to other management classes, so another scenario would be to change the retention parameters of the existing management classes. If this is done, those new parameters will be used in the next EXPIRE INVENTORY operation. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TDP for Oracle
On Tue, Sep 12, 2006 at 11:09:27AM +0700, Wira Chinwong wrote: >As your suggestion, currently instead rman database, I'm using > controlfile to keep the catalog. Can I still follow by your recommendation? I'm assuming that you mean that you're not using an RMAN recovery catalog database, but instead using the controlfile? Yes, you can (and should) still use the RMAN retention mechanism, but you should be aware that some uses *require* a recovery catalog database (such as KEEP FOREVER, for example). This is all documented in the RMAN documentation, which is a required read when you're doing this sort of thing. Also, make sure you *test* your procedures. Good luck, -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TDP for Oracle
On Sun, Sep 10, 2006 at 11:47:57PM +0700, Wira Chinwong wrote: > perform daily backup. But I don't have an idea to setup monthly & yearly > backup policy. The expiration on daily & monthly & yearly are conflicted. > Anybody who have an experience please advise. You shouldn't use TSM expiration parameters like Versions Exists, Versions Deleted, etc. Those settings should be configured according to the TDP installation manual. You must use RMAN to configure your database backup retention policy. You can do this using the RMAN CONFIGURE command, and you can always use the KEEP option for individual backups. See the RMAN documentation for this, it's not complicated. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: errno: 2
On Fri, Sep 08, 2006 at 09:36:57AM -0700, Gill, Geoffrey L. wrote: > I received a couple of responses on this and both point to the file having > been deleted or did not get installed. Doesnt TSM report this sort of > issue if it saw the file was there when it originally scanned the system but > when it finally went to back it up it was gone? The weird thing is unless > this files comes and goes the administrator of the box showed me the file is > there and its not like the date has todays date so it needs to be backed > up again. > The other odd thing is there are a ton of these errors in the dsmerror.log, > one after the other on the same day, then it will go away for either days or > months and return. > > What am I missing? > > -r-xr-xr-x 1 root bin 385024 Jun 25 2003 libPiIMG.sl First, to be absolutely sure what errno 2 means *on HPUX*, run the Perl-snippet. If it indead means: "No such file or directory", continue. If not, give the output of the Perl-snippet. Second, show the output of: ls -l /opt/tivoli/tsm/client/ba/bin/plugins/libPiIMG.sl and (as root): dsmc query backup -inact /opt/tivoli/tsm/client/ba/bin/plugins/libPiIMG.sl If you have another HPUX machine, show the output of the above commands on that machine as well. Does the error also occur on that machine? Does this error only occur when attempting to do image backups? If yes, do those backups succeed or fail? -- Jurjen Oskam
Re: errno: 2
On Wed, Sep 06, 2006 at 03:54:48PM -0700, Gill, Geoffrey L. wrote: > 09/06/06 11:48:29 shl_load() of > /opt/tivoli/tsm/client/ba/bin/plugins/libPiIMG.sl failed, errno: 2 It isn't familiar, but if you want to know what errno 2 means you van do this: perl -e '$!=2 ; print "$!\n";' On AIX, errno 2 means: "No such file or directory". -- Jurjen Oskam
Re: Making TSM twin-center compliant
On Fri, Sep 01, 2006 at 04:31:30PM +0200, Loon, E.J. van - SPLXM wrote: > Correct me if I'm wrong, but as far as I know, you cannot backup to a > copypool. Am I wrong here? No, but that's not a problem. Suppose you have a disaster. At the DR site, you have all your copypool data, but all your primary pool volumes are marked DESTROYED. At that point, you just insert some new empty tapes in your library, and assign them to the primary pool. Bingo, you can backup your clients (if any). Be sure to think this through, since you'll also want to restore your clients ASAP (presumably) and bring your primary pools back using RESTORE STGPOOL. A lot of activity going on there. Regards, -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM Database Reorg
On Wed, Aug 30, 2006 at 03:32:44PM -0500, Whitlock, Brett wrote: [TSM database reorganization] > recommended. Problem is there is another dba up there, who has some TSM > experience, who says she used to do it painlessly. Anyone want to weigh > in? Again: *why* did the DBA reorg the database? The fact she used to do it frequently is not really relevant. In our case, I could run EXPIRE INVENTORY three times a day rather painlessly, but that doesn't make it *useful*. As someone once said: "Profile, don't speculate." (It's starting to remind me of the "paging space needs to be 2 times as large as your RAM"-argument. You hear it all the time, and when someone asks why, all you get is: "eeehm, because everybody does it") -- Jurjen Oskam
Re: TSM Database Reorg
On Wed, Aug 30, 2006 at 03:12:51PM -0500, Whitlock, Brett wrote: > I have a dba who is recommending I do a regularly scheduled database > reorg (ie weekly). Anyone doing this? Can you share the good, the > bad, and the ugly about doing this? The general consensus here is that it almost always isn't worth doing. What reasons does the DBA provide for his recommendation? Regards, -- Jurjen Oskam
Re: To DirMC, or not to DirMC
On Sat, Jul 22, 2006 at 01:10:19AM +0200, Remco Post wrote: > This was discussed at Oxford last year. I don't recall the outcome. I do > recall that because TSM allocates 256k chunks minimum for a transaction, > you do not really want to use this feature since TSM v5.3. Isn't this only for sequential access volumes? -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
AIX Client 5.3.4.3 packaging error: webclient doesn't work
Hi there, I've just upgraded an AIX 5.3 box from 5.3.4.0 to the 5.3.4.3 client. >From that moment, the Web-client wouldn't start, with an error message that it was unable to start dsmagent. Indeed, dsmagent wouldn't even start from the command line: $ dsmagent Could not load program dsmagent: Dependent module libxmlutil-5.3.4.0.a could not be loaded. Could not load module libxmlutil-5.3.4.0.a. System error: No such file or directory In /usr/lib, there is a symlink to libxmlutil-5.3.3.0.a (pointing to a non-existant file), which is the wrong version. Manually creating a symlink to libxmlutil-5.3.4.0.a fixes the problem. Manually inspecting the install package reveals that this is a packaging error (see the file tivoli.tsm.client.ba.32bit.web.inventory and search for SYMLINK). So, when you want to install 5.3.4.3 on AIX and want to use the Web-client, prepare to make a symlink yourself. (I already reported this to IBM) -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TapeAlerts on LTO2 drives, tapes
On Wed, Jun 28, 2006 at 02:23:33PM -0400, Richard Sims wrote: > You might experiment with one of your drives, boosting it to the > latest microcode level (if not already there) and see if amelioration > occurs. Back in 2005 we also had problems with our IBM LTO2 drives in a 3584 library. In the end we found out that these were all microcode-related. After upgrading to a fixed microcode level, we still kept seeing problems with some tapes. What happened was that the chip in the cartridge contained 'bad' information (due to the drive microcode problems earlier on) and this causes weird behaviour (TapeAlerts) even after we installed correct microcode. The fix for this was to use 'Space to End of Data' in tapeutil: this apparently clears or reinitializes the chip in the tapes. (Even now, we sometimes get a TapeAlert "This tape is not data-grade" on a rarely used volume.) -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: META-topic: USENET mirror/archive of ADSM-L ?
On Wed, Jun 21, 2006 at 02:46:58PM -0400, Allen S. Rout wrote: > If, like me, you find that the search engine on adsm.org is less > useful than it might be; if you are less than enamored with the > fragmentation of discussions into forums and posts, then this could > help. Just as an FYI, this list is also archived at http://marc.theaimsgroup.com/?l=adsm-l. The search capabilities of MARC are (IMHO) quite good. (Of course, I don't mean to say that gmane is a bad idea (I don't even know it), but there is an alternative for adsm.org) Regards, -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: Metadata
On Fri, May 19, 2006 at 01:57:44PM -0400, David E Ehresman wrote: > I have a customer who is saying they backed up files on a linux box, > changed ownership of the files, backed them up again. Then he restored > the original versions of the file but got the changed ownership > assoicated with the later backup. Your customer is right. This behaviour is by design (for the Unix clients) and documented: "If /only/ the following items change, they are updated without causing the entire file to be backed up to the server: * File owner * File permissions * Last access time * Inode * Group ID" Yes, this means you've potentially got big problems when a backup runs after you've issued an erroneous chmod -R. You cannot restore the correct permissions, because TSM simply doesn't have them anymore. (Perhaps on inactive versions, but then you'd be restoring old data.) Regards, -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: 3584 checkin/checkout
On Thu, May 04, 2006 at 02:10:09PM -0400, Jim Zajkowski wrote: > People leave their IO doors open? After exchanging tapes, it *did* happen once or twice here. When that happens, people automatically get informed Really Quickly and come to the library to close the door. Which is no problem, because those people are in the room next to the library. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: 3584 checkin/checkout
On Thu, May 04, 2006 at 12:01:49PM -0500, John Monahan wrote: > The whole point of using scripts is so you don't have any manual tasks > *unless* something is wrong. I would consider an open library door as > something being wrong. I meant that it probably would be wise to not use the WAITTIME=0 parameter and use the REPLY mechanism. The REPLY would then only be sent when the operator acknowledges (via a menu or something) that he closed the door. The handling of the REPLY can be scripted of course, i.e.: the operator doesn't need to be in dsmadmc and issue the REPLY himself. > Automation scripts also won't work if someone powers off the library. > Seriously, you can't expect a TSM admin to write automation scripts to > cover every possible instance of human error. Technology can't fix > stupid. True, but if you can foresee errors which are easy to make that have large(ish) consequences, it's silly not to think about how to prevent either the error or the consequences. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: 3584 checkin/checkout
On Thu, May 04, 2006 at 10:24:14AM -0400, Jim Zajkowski wrote: [scsi library] > In 5.3, you can say checkin libv waitt=0 and it will not bother to > generate a prompt, it will just do it. Handy. Yeah, but when you enter this command and the library door is NOT closed, all you library operations will hang until the door is closed. So you might want to require a manual action when kicking off this command. (The way I solved this is to check for the "library requires manual attention" error. If it occurs, scream loudly. :) ) -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: eject tapes from 3584
On Fri, Apr 28, 2006 at 10:24:32PM -0700, Gill, Geoffrey L. wrote: > I am assuming that changing remove=yes to remove=bulk will fix the issue of > the system waiting for a reply and if the I/O port becomes full it will > continue to wait till they are removed and the port made available again. I > am also assuming that if the 1 hour time set elapses without the port > clearing will also result in an incomplete process. It's all quite clearly explained when you type "HELP MOVE DRM". The 3584 is a SCSI library, but I assume you know that. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: 3584 library
On Wed, Apr 19, 2006 at 03:12:32PM -0700, Vats.Ashok wrote: > I don't think I understand how inventory & audit works. When you do Audit > library 3584lib checkl=barcode should I see the robotics move and read each > and every tape slot and resysnc with TSM data base ? how do I get these back > in sync A 3584 library remembers which cartridges are where. After you open and then close the door, the library scans every slot and builds up an inventory. When you issue the AUDIT LIBRARY command in TSM, the library already knows which cartridge is where so it can tell TSM without having to scan anything. In your situation, after an AUDIT LIBRARY you'll see that TSM noticed that the tape was missing, and was removed from the TSM library inventory. (The library itself already knew it was missing, so it wasn't in the *library* inventory at all.) After this, you can check in the tape normally (with status PRIVATE if there's any data on it). Note that during an AUDIT LIBRARY you can get the following error: "ANR8788W Unable to read the barcode of cartridge in slot-id 1044 in library 3584; loading in drive to read label." This happens when TSM expected a tape in a particular slot, but the library told TSM that slot is now empty. This is harmless; if you don't receive any errors about missing volumes everything is OK. It means that you (or someone/something else) rearranged the tapes in the library without TSM knowing about it. If this rearrangement was expected, all is fine. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM for VMWare ESX Question - Where can I find the storage agent for TSM?
On Tue, Apr 18, 2006 at 02:10:45PM -0500, Troy Frank wrote: > You "might" be able to run the linux tsm client directly on the esx > server, to get guest os backups. However even if that did work, it > would be whole image backups (not file level), and you'd need to bring > down the virtual machine before running it's backup. I think the more > typical solution is to just run the tsm backup client within the os of > each virtual machine. You can run the TSM client on ESX Server, it's just Red Hat 7.2. You can use the add redo feature to back up running virtual machines. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: multiple dsmcad procs running
On Thu, Apr 06, 2006 at 03:25:58PM +0200, Vlijmen, M. van wrote: > Am I right in assuming that correct authentication only takes place between > the server and the dsmcad process that initiated the first > contact with the server? So, in this case the TSM server didn't recognize > the dsmcad process? What probably happened (no way to be sure without the logs): the first dsmcad process you started began listening to the default port number, probably 1581. The processes you started after that couldn't use that port, since it was in use. Then you killed some dsmcad processes, including the one that was listening on 1581. The dsmcad that you left running was listening on an unknown (by the TSM server) port. The TSM server then tried to contact the dsmcad process on its default port, but on that port nothing was listening so communication failed. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM server automation products
On Mon, Apr 03, 2006 at 01:09:35PM -0400, Allen S. Rout wrote: > I'll allude to a bit of logical pedantry by suggesting a google of > 'beg the question' :) Darn, I /thought/ I got that right. From a Google search: "If you're not comfortable with formal terms of logic, it's best to stay away from this phrase, or risk embarrassing yourself." Serves me right trying to look smart. :) > And -that- is the function folks hope to get out of the third-party > tool, for which they're willing to sustain complexity and render up > many dollars. I've tried the demo version of TSMManager, and it was indeed a nice experience to be able to fire up a report here and a graph there without much trouble. Our TSM environment is too small to justify that niceness, since the Perl-code keeps everything running virtually maintenance-free. But yes, I think there are many installations where these off-the-shelf products really are quite benificial to have. (I guess the vendors are quite happy with the ISC/AC-behemoth :) ) -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM server automation products
On Mon, Apr 03, 2006 at 10:48:13AM -0400, Thomas Denier wrote: > 2.I am spending significant amounts of time on enhancements to the > script. Yes, that would be a reason to consider an off-the-shell solution. The Perl-automation here runs more or less autonomic and doesn't require much maintenance. A nice side note is that in a former job, the TSM admin used Rexx to automate his Windows (then) ADSM server. Since I come from an Amiga background I had experience with ARexx, but all the other Windows admins had no clue about Rexx. :) I guess a lot depends on the specifics of each installation. -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM server automation products
On Sat, Apr 01, 2006 at 09:46:00PM -0500, Allen S. Rout wrote: > Speaking as someone buried in my own PERL up to my nose: [snip: quite a good argument] > I don't think I could be as effective with a third-party product as I > am with my own stuff. I do think that the person who gets my job > after I get hit by a truck will curse me for years. Thanks, those are good points. But it does beg the question: how bad is your current situation? :) I mean, is it such a spaghetti that nobody, except you as the developer, can *really* get it? Isn't *that* something you could change, so that your successor can as effective as you are? (Implementing and migrating to a third party product costs time and money, wouldn't that better be spent cleaning up the custom code?) Third-party programs are indeed spread more widely, but this also means they must be more complex because they must support configurations which you don't even use. Homegrown code has only the complexity you need. As a side note: it would really *really* be nice to have a documented API to enter dsmadmc commands -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: TSM server automation products
On Fri, Mar 31, 2006 at 01:23:11PM -0500, Thomas Denier wrote: > error conditions and reports them to our operations staff. My > management is convinced that we can buy an off the shelf product > that will replace 80 percent of the Perl code. Does such a > product exist? Uhm, may I ask *why* you'd want an off-the-shelf product instead of your own code? Is the Perl-code not functioning correctly? What problem are you trying to solve? -- Jurjen Oskam Savage's Law of Expediency: You want it bad, you'll get it bad.
Re: 3584 library: problems with some trays
On Wed, Mar 29, 2006 at 10:26:52AM +0100, Leigh Reed wrote: > I raised a call with IBM and they have come back with the fact that it > may be to do with a barcode scanner that has been installed upside down. My call is still open. When a solution is available, I'll report back. > - > CONFIDENTIALITY NOTICE: This message is confidential, intended only for > the named recipient(s) and may contain information that is privileged, > or exempt from disclosure under applicable law. If you are not the > intended recipient(s), you are notified that the dissemination, > distribution or copying of this message is strictly prohibited. If you > receive this message in error, or are not the named recipient(s), please > notify the sender by reply e-mail, delete this e-mail from your > computer, and destroy any copies in any form immediately. Receipt by > anyone other than the named recipient(s) is not a waiver of any > attorney-client, work product, or other applicable privilege. This > message and all contents may be reviewed by authorized parties of the > Catholic Health System other than those named in the message header. > > The contents of this message do not bind the Catholic Health System to > any contract, position, or course of action unless the sender is > specifically authorized to enter into contracts on behalf of the > Catholic Health System. The contents of this message do not necessarily > constitute an official representation of the Catholic Health System. Oh dear, it looks like someone from your Legal Department escaped and came too close to a computer again. Please be more careful with them in the future. -- Jurjen Oskam
Re: tsm admin question !
On Tue, Mar 21, 2006 at 11:43:54AM +0100, goc wrote: > just wondering if you when creating a new node do create an administrator > for that > node or use some/your admin account for every node ? Nope, most nodes here are created with USERID=NONE. Whats's best for you depends on your individual sitation though... -- Jurjen Oskam
Re: 3584 library: problems with some trays
On Mon, Mar 20, 2006 at 02:33:00PM -0500, Kathleen M Hallahan wrote: > Is this something you folks are seeing in all frames, or in just one? Both. Our 3584 is just one frame. :) -- Jurjen Oskam
Re: 3584 library: problems with some trays
On Mon, Mar 20, 2006 at 12:25:07PM -0700, Ben Bullock wrote: > I have a guess. I have a 3494, so I might be wrong, but I will > take a shot. [snip explanation] Thanks for your explanation. I don't think the situation you describe is the case though (but it could certainly have something to do with it), but it did remind me that I forgot to mention something explicitly: the library only thinks the mentioned slots are completely empty after an initialization that ran due to a powerup or the opening/closing of the entire front door. The weird thing is that when I perform a library inventory using the front panel, *all* tapes are detected, *including* the ones that the library didn't see just seconds earlier. This is consistent behaviour: inventory triggered by powerup/front door --> tapes invisible, inventory triggered manually --> all tapes visible. Weird. -- Jurjen Oskam
3584 library: problems with some trays
Hi there, Some time ago we experienced a power failure that lasted long enough to make us shut down our 3584 library. This was done gracefully, and when the power came back we turned the library back on. The TSM server was also gracefully shut down and turned back on. After TSM came back, I did an AUDIT LIBRARY which exited with a failure; two tapes which TSM thought were in the library, weren't (according to the library) so TSM took them out of its inventory. The strange thing was that the tapes were indeed in the library. Some investigation revealed that the library was somehow "blind" to tapes in the last row of the first and second column, but this only happened after a library initialization due to a powerup or open/close of the entire front door. This happens with every tape I tried, so the labels aren't the problem. A tape which is perfectly well recognized in any slot doesn't show up in the lowest slot in column 1 and 2. The library doesn't detect anything there, it'll also let you move a tape to a full slot using the control panel. Of course, this results in an error (94 80: slot unexpectedly full). IBM is already on this, and it's no big problem because I can easily work around it by leaving the first row almost completely empty. This results in the problematic slot not to be used, because our tape rotation isn't that high. I was interested if this symptom perhaps looks familiar to someone here. We're on library code 5770, and the TSM symptoms are errors about slots being unexpectedly full and TapeAlerts about the library inventory being inconsistent. Thanks, -- Jurjen Oskam
Re: linux dsmcad problem
On Fri, Mar 10, 2006 at 09:08:36AM +0100, Dirk Kastens wrote: > After running the backups with "dsmc sched" and schedmode set to > prompted for years, I didn't know that I had to set the schedmode to > polling when I use dsmcad. You don't have to set the schedmode to polling when using dsmcad. Something was causing schedmode prompted to not work, and that something is still there. It's probably wise to look into this, because that something might also cause other things to fail. -- Jurjen Oskam
Re: linux dsmcad problem
On Thu, Mar 09, 2006 at 09:07:54AM +0100, Dirk Kastens wrote: > Using lsof (thanks to Richard for the hint) I saw that the dsmcad is > listening on port 47358. Could this be the problem? Do I have to specify > the webports option when I use the dsmcad? Normally, you don't have to specify the port. However, when the default port is already used by another application when dsmcad starts up, it will pick another one and the TSM server can't reach the client. -- Jurjen Oskam
Re: restore file permissions
On Fri, Jan 27, 2006 at 07:20:17AM -0500, Richard Sims wrote: > You seem to be under the false impression that permissions are stored > in TSM independent of the object data. Not. No, that is not the impression I'm under. I'm under the impression that this: > This is why when file > system object attributes are changed, the object itself gets backed > up again. ... is not the case. > (See backup candidate criteria details in the client > manual.) The manual specifically states: "If /only/ the following items change, they are updated without causing the entire file to be backed up to the server: * File owner * File permissions * Last access time * Inode * Group ID" > Object restoral gets the attributes of the object. True, but you get the *updated* attributes. Which are the ones you *don't* want, in case a backup happened after a runaway chmod. I've included a transcript of a terminal session to demonstrate (edited for legibility, un-edited version is at http://www.stupendous.org/tsm.txt). I create a file "demo" with permissions rw-r--r--, and store it in TSM using "dsmc i". Then, I change the permissions to rw---, and again perform "dsmc i". Note that the file is *updated* in TSM, and still only has 1 version in TSM (no inactive version). It is now impossible to restore the file with the original permissions rw-r--r--, as you can see near the end of the transcript. $ ls -l demo demo not found $ dsmc query backup -inact demo ANS1092W No files matching search criteria were found $ echo foo > demo $ ls -l demo -rw-r--r-- 1 joskam staff 4 Jan 27 14:44 demo $ dsmc i demo Incremental backup of volume 'demo' Normal File--> 4 /home/joskam/demo [Sent] Successful incremental backup of '/home/joskam/demo' Total number of objects inspected: 199 Total number of objects backed up:1 Total number of objects updated: 0 $ dsmc query backup -inact demo Size Backup DateMgmt Class A/I File ----- --- 4 B 01/27/06 14:44:43MC_TRIAS_D A /home/joskam/demo $ chmod go-r demo $ ls -l demo -rw--- 1 joskam staff 4 Jan 27 14:44 demo $ dsmc i demo Incremental backup of volume 'demo' Updating-->4 /home/joskam/demo [Sent] Successful incremental backup of '/home/joskam/demo' Total number of objects inspected: 199 Total number of objects backed up:0 Total number of objects updated: 1 $ dsmc query backup -inact demo Size Backup DateMgmt Class A/I File ----- --- 4 B 01/27/06 14:44:43MC_TRIAS_D A /home/joskam/demo $ echo bar >>demo $ chmod go+x demo $ ls -l demo -rw---x--x 1 joskam staff 8 Jan 27 14:47 demo $ dsmc i demo Incremental backup of volume 'demo' Normal File--> 8 /home/joskam/demo [Sent] Successful incremental backup of '/home/joskam/demo' Total number of objects inspected: 199 Total number of objects backed up:1 Total number of objects updated: 0 $ dsmc query backup -inact demo Size Backup DateMgmt Class A/I File ----- --- 8 B 01/27/06 14:47:24MC_TRIAS_D A /home/joskam/demo 4 B 01/27/06 14:44:43MC_TRIAS_D I /home/joskam/demo $ rm demo $ dsmc restore -inact -pitdate=01/27/2006 -pittime=14:45 demo Restoring 4 /home/joskam/demo [Done] Total number of objects restored: 1 Total number of objects failed: 0 $ ls -l demo -rw--- 1 joskam staff 4 Jan 27 14:44 demo $ rm demo $ dsmc restore demo Restoring 8 /home/joskam/demo [Done] Total number of objects restored: 1 Total number of objects failed: 0 $ ls -l demo -rw---x--x 1 joskam staff 8 Jan 27 14:47 demo > Furthermore, in Unix, such attributes are in the inode, not in the > directory. Take a look at your operating system's dirent.h header > file to realize how paltry Unix directory information is. Yes, I know this. (This can also be seen by performing "cat" on a directory.) -- Jurjen Oskam
Re: restore file permissions
On Thu, Jan 26, 2006 at 08:42:01AM -0700, Andrew Raibeck wrote: > > Is there also a way to use tsm to restore the file permissions without > > doing a restore of all files? > > No. Suppose I recursively mess up the permissions of a directory, and that directory is then backed up as part of e.g. "dsmc i". When I then rm -rf the directory on the host and restore it using a point in time restore from *before* the chmod, which permissions will the restored files have? (This is assuming a large enough "retver" c.s. parameter) -- Jurjen Oskam
Re: dsmadmc keyboard input problem
On Sat, Jan 21, 2006 at 02:16:07PM -0600, Mike wrote: > Have you tried changing your terminal emulation and the TERM variable? Yes, I have, as I explicitly mentioned. The problem is not that Backspace does or doesn't work, the problem is that with the *same* settings, it works *differently* depending on what kind of input dsmadmc expects. And this wouldn't even be a problem if dsmadmc rejected input like "nyes" or "y^Hn". -- Jurjen Oskam
dsmadmc keyboard input problem
Hi there, Some time ago, while using dsmadmc (on AIX 5.3) to do some filespace-maintenance, I noticed that dsmadmc does something interesting when it asks you if you really want to delete a particular filespace. When you enter a DELETE FILESPACE command, you're normally asked if you really want to delete the filespace, with a prompt indicating that dsmadmc want a Y or a N for an answer. I noticed that dsmadmc doesn't care what you type, as long as the first character is either a Y or a N (or the lower case variant). For example, typing "nyes" will not delete the filespace, typing "yno" *will* delete the filespace. While this is debatable (IMHO), this gets nasty when you combine this with how dsmadmc handles other input, like, say, Backspace. I have tried on two types of Unix terminal emulators: PuTTY (xterm, send Backspace as Ctrl-?), and the HMC supplied virtual terminal (vt320). When entering commands, keys like the cursor keys and backspace work and do what you'd expect. Backspace deletes the character left of the cursor. If I type: "q pgr", I get the output of the QUERY PROCESS command, as expected. However, when I enter a DELETE FILESPACE command and dsmadmc asks me if I'm really sure, and I type: "yn", the deletion *will* occur, i.e.: the filespace is *gone*. Depending on the terminal I've seen "y^Hn" or just plain "n" (which is particularly nasty), but once you've typed the initial "y" there's no way except Ctrl-C to prevent the deletion. I then changed (on the PuTTY) session my TERM environment variable to "dumb". This caused the Backspace-key to work as generally expected when I was asked if I was sure to delete the given filespace. However, it *also* caused the Backspace key to not work at all during normal command entering. I am aware that Unix and terminal emulation can be a tricky issue, but I believe there *is* a problem in dsmadmc in this case, because it treats the same keypress (Backspace) differently in different parts of the program. Also, the fact that dsmadmc only looks at the first character of input when asking a Y/N-question, is IMHO not quite correct. What do you think? -- Jurjen Oskam
Re: TDPO and different Management Class
On Fri, Dec 16, 2005 at 10:44:38AM +, David McClelland wrote: >As the other guys have pointed out, there should be a single >management class with the well-known set-in-stone backup copygroup >retention policies as a destination for DP for Oracle backups. Take a I'd like to repeat a tip I got right here on the list: multiple management classes for TDP for Oracle can be quite handy. Every management class should have the same retention policy, but you can vary the storagepool destination. That way, you can send e.g. your archivelogs to a diskpool and regular backups directly to tape by using the RMAN FORMAT parameter and the INCLUDE-option in the TSM API configuration. -- Jurjen Oskam
Re: TDPO and different Management Class
On Thu, Dec 15, 2005 at 05:46:11PM -0500, Muthukumar Kannaiyan wrote: > Separately, how do we specify management class while backing up from > RMAN? As Kurt Beyers already stated: you don't. Point your Oracle DBA's to the KEEP option in RMAN. -- Jurjen Oskam
Re: Unload a specific dsmcad question
On Sun, Dec 11, 2005 at 12:14:05PM +0200, Robert Ouzen wrote: > Sorry for the missed information > 1. Netware OS > 2. Tsm client version 5.3.0 > 3.Want to stop the process by running unload dsmcad (but my problem I have > some dsmcad running and want to kill only specific one ) Ah, in that case: sorry, I can't help you. I have exactly zero experience with Netware. :) Good luck, -- Jurjen Oskam
Re: Unload a specific dsmcad question
On Sun, Dec 11, 2005 at 08:44:48AM +0200, Robert Ouzen wrote: > Quick question about unload dsmcad if I have several dsmcad loaded ? When > I run an unload dsmcad it kill all the processes. Did is a way to unload just > a specific one 1. Which OS? 2. Which TSM version? 3. What do you mean by "unloading" dsmcad? If, by pure chance, you're running dsmcad on Linux, and you see several dsmcad processes: that's normal. Those aren't processes, those are threads belonging to one dsmcad process. Don't kill them. -- Jurjen Oskam
Re: expiration Oracle backups
On Thu, Oct 13, 2005 at 07:38:18AM -0700, Neil Rasmussen wrote: > When deleting Rman objects on the TSM Server, TDP Oracle uses the delete > function on objectsthe objects do not move to inactive they are just > simply deleted from the TSM Server. While the manual says to create nodes for RMAN with Delete Backup permission set to yes, the manual also says to use a managementclass that has Versions Data Deleted = 0 and Retain Only Version = 0. If what you say is true (and I have no reason to doubt that :) ), then these settings are not relevant, since there are no inactive RMAN objects. Am I right? -- Jurjen Oskam
Re: TSM testing
On Fri, Sep 30, 2005 at 08:51:31PM -0500, Stapleton, Mark wrote: > Telnet, ftp, and ping all use their own TCP ports, as does ssh. You need Minor nitpick: ping doesn't use TCP but uses two types of ICMP packets. This could be relevant when dealing with firewalls. -- Jurjen Oskam
Re: Scheduled session remains active
On Thu, Sep 15, 2005 at 12:47:53PM +0200, Jurjen Oskam wrote: > I have a problem with a scheduled command. The TSM server is 5.3.1.4 on AIX > 5.3 ML2, > the affected client is 5.2.2.0 on Red Hat Linux Advanced Server 2.1 (i386). > SCHEDmode is PROMPTed, the client has MANAGEDSERVICES WEBCLIENT SCHEDULE > and dsmcad is runnning. ... and I forgot to mention that POSTSCHEDULEcommand is not specified. Thanks to Richard Sims for mentioning this. Regards, -- Jurjen Oskam
Scheduled session remains active
2:13 Objects: 09/14/2005 21:42:13 Options: 09/14/2005 21:42:13 Server Window Start: 22:00:00 on 09/14/2005 09/14/2005 21:42:13 09/14/2005 21:42:13 Scheduler has been stopped. ... and the dsmwebcl.file shows: [...] 09/14/2005 21:42:14 (dsmcad) Waiting to be contacted by the server. I don't understand why session 2636 remains active for so long, while the schedule already ran successfully. That session seems to cause the TSM server to not prompt that client for the next session, which is a problem. I already searched ADSM.QuickFacts and the IBM Technotes and APARs, but couldn't find a solution. Maybe someone here ran into something like this and knows a solution? Thanks, -- Jurjen Oskam
Re: TDP for Oracle 8.1.7.4 & 9.2.0.6 on same server
On Wed, Aug 17, 2005 at 03:45:23PM -0500, Scott Grover wrote: > I have TDP 64 bit for Oracle AIX 5.2.0 running on an AIX 5.2 client > with > both Oracle 8.1.7 and 9.2.0. SHOWENV shows my configurations for both oracle I assume you have 64 bit Oracle 9, and 32 bit Oracle 8.1.7. (Oracle 8 64 bit won't run on 64 bit AIX 5.x) > Has anyone else run into this? (The Oracle 9i portion all works finejust > having problems with linking Oracle 8.1.7 to the 64 bit TDP binary). You should link Oracle 8 to the 32 bit version of TDP for Oracle. Although this works (we've done it ourselves), it's not supported by IBM (also first hand experience). Which means: if it breaks, you get to keep the pieces. :-) -- Jurjen Oskam
Re: LTO2 is unconfident
On Tue, Jul 05, 2005 at 02:02:02PM +0700, nghiatd wrote: > a.. 9 tapes have 'Access=Read-Only', 'Number of Write Errors: 1' > b.. 1 tape has 'Access=Unavailable', 'Number of Write Errors: 1 Perhaps that batch of LTO2 tapes was bad, but it's more likely that one of your drives has problems. Check your logs to see in which drive all these errors occurred. > In addition, Capacity of tape aslo display lot of different value (629,557.2 > MB; 372,665.2 MB; 481,247.3 MB) although capacity of tapes is 200/400 GB. This is normal. 200 GB is the maximum uncompressed capacity of a volume, 400 GB is the maximum compressed capacity *assuming you'll achieve a compression ratio of 2:1*. When your data is higly compressable, you can store more than 400 GB on a tape. When your data is already compressed, you can only store 200 GB. Since TSM doesn't know what data you'll be sending to tape, the indicated capacity of a tape is an estimate, *until* the tape is filled up. At that point, you know the exact amount of data written to the volume, so TSM changes the capacity to that amount. -- Jurjen Oskam
Re: using rman backup database to localdisk
On Wed, May 18, 2005 at 08:48:25AM +0700, Nghia Tran Dai wrote: > I want using Rman backup Oracle database to BACKUPDB directory on local disk This really is an Oracle question since it doesn't have anything to do with TSM. You might get better response on an Oracle-mailinglist or -forum. That being said, why are you planning to do this? If you use TSM for Databases you don't need to make a backup in a staging area and only then send it to TSM. -- Jurjen Oskam
Re: AIX client
On Tue, Apr 19, 2005 at 03:59:03PM +0200, Loon, E.J. van - SPLXM wrote: > The 5.3.0 client package does only contain a 64-bit API client and no 64-bit > BA client. Does anybody know why that is? Yes I do, and so do the creators of the accompanying README file... :-) -- Jurjen Oskam
Re: Delete a volume of a primary SP
On Tue, Apr 19, 2005 at 09:00:31AM -0300, TSM User wrote: > If I delete a volume of a primary SP (discard data = yes) that happens > with the data that are in the copy SP You can find out with the command HELP DELETE VOLUME. -- Jurjen Oskam
Re: 3583 Meltdown
On Fri, Apr 01, 2005 at 02:10:45PM -0600, Rushforth, Tim wrote: [3584 TapeAlerts] > But we too were getting too many of these useless errors so we decided > to turn tape alert off and only use if we suspect a problem. (Perhaps we > should leave tape alert on but not report on any of the errors ...) Did the useless TapeAlerts mostly occur on a drive which was also used as a control port? (They did here.) You might want to look into the useless TapeAlert-problem, though. In my case, when a bunch of silly TapeAlerts occured there was a small chance of the LTO2 index becoming corrupt, resulting in bad performance with that tape (but no I/O-errors), and genuine TapeAlerts on that volume from then on. Not letting TSM report the TapeAlerts would only have masked the problem, since no hard errors occured.. -- Jurjen Oskam
Re: 3583 Meltdown
On Fri, Apr 01, 2005 at 01:57:56PM -0600, Bill Kelly wrote: > I heard, both Jurjen and I have problems open with IBM hardware support > (personally, I suspect a firmware problem, but obviously that's just an > uneducated guess). I'm now running on 4AP0, a non-released version. The 'silly' TapeAlerts are gone now, only real TapeAlerts seem to remain. The only thing I'm still seeing is that some tapes keep producing "This tape is not data-grade. Your data is at risk."-TapeAlerts. But I don't intend to hijack the topic, which was about a 3583. Since it's been almost 1,5 years since I last worked with one of those, I don't think I can help there. -- Jurjen Oskam
Re: Deleting archive data using command line client
On Wed, Mar 23, 2005 at 03:19:46PM -0500, Richard Sims wrote: > therein. My original response suggested the ultimate high level: the > file system name, which should then show all your archived directories > and files within it. At a minimum, specify one higher directory level. I could have sworn I also tried the command with "/oracle/" and still getting nothing. I retried and now the directories did show up! What was the cause? Instead of using "query archive", I used "delete archive -pick" earlier on, but I failed to realise that "-pick" only shows files. D'oh! Thanks, -- Jurjen Oskam
Re: Deleting archive data using command line client
On Wed, Mar 23, 2005 at 10:37:24AM -0500, Richard Sims wrote: > dsmc query archive -des="gl-2004*" -subdir=yes // > > This should turn up all files and directories which > employed that Description. Hmm, that doesn't work as I expect: # dsmc query archive -des="pint-2004*" -subdir=yes /oracle/app/oracle/admin/pint/arch/ IBM Tivoli Storage Manager Command Line Backup/Archive Client Interface Client Version 5, Release 3, Level 0.0 Client date/time: 03/23/05 20:11:03 (c) Copyright by IBM Corporation and other(s) 1990, 2004. All Rights Reserved. Node Name: TRIPROD1 Session established with server SERVER1: AIX-RS/6000 Server Version 5, Release 3, Level 0.2 Server date/time: 03/23/05 20:11:03 Last access: 03/23/05 20:04:59 ANS1092W No files matching search criteria were found ... but see http://www.stupendous.org/dsmwebcl.jpg for a screenshot of the Web-client showing many archives that do match the wildcard. All those archives do not contain any files, though. The following command: # dsmc query archive -des="pint-2005*" -subdir=yes /oracle/app/oracle/admin/pint/arch/ ... does result in many files being shown (but no directories): IBM Tivoli Storage Manager Command Line Backup/Archive Client Interface Client Version 5, Release 3, Level 0.0 Client date/time: 03/23/05 20:19:10 (c) Copyright by IBM Corporation and other(s) 1990, 2004. All Rights Reserved. Node Name: TRIPROD1 Session established with server SERVER1: AIX-RS/6000 Server Version 5, Release 3, Level 0.2 Server date/time: 03/23/05 20:19:10 Last access: 03/23/05 20:18:26 Size Archive Date - TimeFile - Expires on - Description ------ 3,072 B 03/08/05 09:29:17 /oracle/app/oracle/admin/pint/arch/pint_1_22521.rdo 03/23/05 pint-20050308092917 4,096 B 03/08/05 09:59:56 /oracle/app/oracle/admin/pint/arch/pint_1_22522.rdo 03/23/05 pint-20050308095955 10,486,272 B 03/08/05 10:30:32 /oracle/app/oracle/admin/pint/arch/pint_1_22523.rdo 03/23/05 pint-20050308103032 10,486,272 B 03/08/05 10:30:33 /oracle/app/oracle/admin/pint/arch/pint_1_22524.rdo 03/23/05 pint-20050308103032 [... and so on ...] > If TSM balks at removing directories which do not have archived files > under them, look into the special CLEAN ARCHDIRectories command (APAR > IX89638). This doesn't seem to apply, since the archiving was done under 5.2 (now on server 5.3.0.2 and AIX client 5.3.0.0). The server doesn't recognize the commands CLEAN ARCHDIR and friends. > And, welcome to the more powerful world of CLI operations. :-) Thanks, but I'm already a CLI user, and TSM is no exception: when upgrading to 5.3, I didn't even *try* to bother installing the ISC. :-) -- Jurjen Oskam
Deleting archive data using command line client
Hi there, With the GUI-client, you have the option to manually delete archived data. The GUI presents archived data in packages, sorted by archive description. I'm using the 5.3 B/A client on an AIX node. I'd like to delete a whole lot of those packages. Using the GUI is not practically possible: selecting the first entry and then shift-click on the last entry to select it and everything in between doesn't work. I manually have to select each and every one. Also, I didn't find a way to select and/or filter on the contents of the description field. I didn't find a way to accomplish this task using the CLI. Of course, it's possible to delete archived data using the CLI, but it doesn't seem to be possible to delete archived packages using a wildcard pattern on the archive description. There are no files in the archived packages, they are all expired (as intended). Due to a configuration mistake, the directories were bound to a management class with a much longer archive retention setting, so they haven't expired yet. Eventually though, they will. The reason I want do delete these packages is that the GUI (specifically the Web client) gets real slow when there are a lot of archived packages to choose from. What I would like to know is how to achieve something like this (pseudocommand): dsmc delete archive -des="gl-2004*" This would then delete all archived data with a matching description, even when only directories are present in a package. Thanks, -- Jurjen Oskam
Re: DIRMC - Are copypool reclamation performance issues resolved or not.
On Thu, Mar 17, 2005 at 07:38:16AM -0500, Richard Sims wrote: > blocks of 256 KiB minimum...". Could you provide a documentation or web > site reference for that 5.3 change? No, sorry. Just the info I received through the PMR. I made the suggestion to include this in e.g. a README, and that suggestion was passed on. > According to 2004 Technote 1167281, the block size for FILE and DISK > devtypes has been a fixed size of 256 KB for some time. > > I'd like to see definitive information on this before committing my 5.3 > beliefs. Hmmm. The only thing I know for certain is that the exact same setup that works with 5.2 and below, doesn't work with 5.3 due to FILE volumes filling up way too quickly. In the PMR I provided trace information, etc. etc. Eventually the result was that this was working as designed, and that a design change in TSM 5.3 caused the volumes to fill up to only 1,5% before encountering end-of-volume. The description given in the PMR seems to fit the symptoms quite well. -- Jurjen Oskam
Re: DIRMC - Are copypool reclamation performance issues resolved or not.
On Wed, Mar 16, 2005 at 07:29:50PM -0600, Rushforth, Tim wrote: [DIRMC] > What in 5.3 warrants new consideration? Probably the fact that sequential volumes are written to in blocks of at least 256 KB, even when the data is only 1500 bytes. This can cause a lot of overhead, and the effective capacity of sequential volumes could be reduced by a factor of 60 or more. Note that a great number of factors influence the above statement. On the one end, it could cause a perfectly OK setup under 5.2 to be unusable under 5.3. On the opposite end, you might notice no adverse effects and even experience a performance improvement. In the case of FILE storagepools used as a DIRMC destination, you might find yourself in the first scenario (works under <= 5.2, doesn't work under 5.3). You might be able to alleviate this by adjusting the TXNGROUPMAX server setting and the TXNBYTELIMIT client setting. Unfortunately, this doesn't affect data that's already in a storage pool. And unfortunately again, storagepools used as DIRMC destination often have quite generous retention settings, as per the documentation and general recommendation. Part of the reason that these retention settings could be generous was that directories were so small that it wouldn't matter a lot. With the new handling of sequential volumes, it does start to matter (and sometimes a lot). -- Jurjen Oskam
Re: Minor gotcha on upgrade to 5.3
On Wed, Mar 16, 2005 at 03:21:23PM +0100, Remco Post wrote: > >to be aware of. In TSM 5.3, the handling of FILE volumes was changed. > >All writes to such a volume is now done in blocks of 256 KiB minimum, > >even when the object itself is much smaller. > and IBM sells this as WAD? Does this also affect tape storage (not likely). I quote from the PMR: "The over head should affect sequential access volumes in general, however there are other factors that can come in to play on a physical tape devices that would not affect a file devclass. These other factor can include volume capacity, hardware compression and the block or header sizes dictated by the actual device.Because of these conditions, the waste seen on a FILE volume may not have the same impact or even noticeable on tape volumes." So your mileage may vary, I guess... -- Jurjen Oskam
Minor gotcha on upgrade to 5.3
Hi there, some of you might remember that I encountered a problem when upgrading TSM from 5.2 to 5.3. The problem was that FILE-volumes reached a premature end-of-volume, causing a 64 MiB volume to only be 1,5 % used but still be full. The cause of this is now known, and it's not a bug but still something to be aware of. In TSM 5.3, the handling of FILE volumes was changed. All writes to such a volume is now done in blocks of 256 KiB minimum, even when the object itself is much smaller. In my case, the storagepool only contained directories. (DIRMC option pointing to a small DISK pool, which migrated to the affected FILE pool) In TSM 5.2 and earlier, this wasn't a problem. In 5.3 (and presumably later), each tiny object now takes up 256 KiB at a minumum, causing a *huge* overhead. The solution was to stop using FILE volumes for this pool, but only DISK volumes. Regards, -- Jurjen Oskam
Re: WinPE Image backup problem
On Sat, Feb 26, 2005 at 03:24:19AM -, Iain Barnetson wrote: > Thanks for your msg, sarcasim is still the lowest form of wit by the > way. It wasn't sarcasm, nor humour. Following those procedures is generally a good idea, and in many cases would greatly speed up the process of solving a problem. (Of course, in some cases, a document about how to properly *read* a bugreport would be handy as well.) -- Jurjen Oskam
Re: WinPE Image backup problem
On Fri, Feb 25, 2005 at 08:09:45PM -, Iain Barnetson wrote: > I can tell you that the procedure as it is now DOES NOT work. Read http://www.chiark.greenend.org.uk/~sgtatham/bugs.html for more info about how to effectively solve your problem. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Ideas re: by-hand reclamation...
On Thu, Feb 24, 2005 at 03:16:44PM -0500, [EMAIL PROTECTED] wrote: > If MOVE NODEDATA had a preview=yes, then we could parse the output of that > command in a fairly straightforward manner. Then you might be interested in QUERY NODEDATA, a new command in TSM 5.3. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: select query misbehaving?
On Wed, Feb 23, 2005 at 12:02:32PM -, Warren, Matthew (Retail) wrote: > Ok, I can do that. But, surely NULL != 'VAULT' is not like '%VAULT%' ? SQL has three valued logic: TRUE, FALSE and UNKNOWN (or NULL): tsm: SERVER1>select count(*) from volumes 108 tsm: SERVER1>select count(*) from volumes where pct_reclaim = 0.0 36 tsm: SERVER1>select count(*) from volumes where pct_reclaim != 0.0 53 tsm: SERVER1>select count(*) from volumes where pct_reclaim is null 19 tsm: SERVER1>select count(*) from volumes where pct_reclaim is not null 90 -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
dsmc hangs in AIX when using KRB5A authentication
Hi there, On an AIX 5.2 ML 4 64bit system we use KRB5A authentication to authenticate some local users against Active Directory using Kerberos. This works fine. There is a problem with dsmc though. When run as root (who is locally authenticated) the client works fine, but when run as any other user (even a user who isn't authenticated using KRB5A), it hangs. As soon as I put: KRB5A: program = /usr/lib/security/KRB5A options = authonly KRB5Afiles: options = db=BUILTIN,auth=KRB5A ... in /usr/lib/security/methods.cfg, the client hangs when run as any local user. The instant I remove those lines, the client will run as any user. The client hangs after the message: "Selective Backup function invoked." (this was when executing a selective backup, but the problem occurs on every type of backup or archive) When the client is hanging, Q SESS on the server shows: tsm: SERVER1>q sess Sess Comm.Sess Wait Bytes Bytes SessPlatform Client Name Number Method State Time Sent Recvd Type -- -- -- -- --- --- - 15,277 Tcp/Ip IdleW 2 S 6.3 K 458 NodeAIX TRIDW1 15,278 Tcp/Ip IdleW 2 S 63 4 Node I have opened a PMR about this, but I wondered if somebody here encountered this problem and might have a workaround. The client version is 5.2.2.5. -- Jurjen Oskam
Re: LTO2 corrupted index question
On Tue, Feb 15, 2005 at 08:03:57AM -0600, Bill Kelly wrote: > Thanks for the reply. Yes, I've got TapeAlerts enabled. And yes, we've > seen all sorts of 'silly' TapeAlerts in the past couple of months; some [...] > information show up. It sounds to me as if we're experiencing the same > firmware problems you've seen. Indeed the symptoms seem to suggest that you have the exact same problem as we have. Do the TapeAlerts mostly occur on the drive that acts as the control port? > I will contact IBM to see about these firmware upgrades you've been > testing; perhaps they're close to making the upgrade generally available. Please also mention hardware call number B1S5C48 (issued by IBM in The Netherlands), as I was repeatedly told that we're the only one with this problem. (I think the problem is relatively serious - I don't like being told that "my data is at risk" by my backup system) -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: LTO2 corrupted index question
On Mon, Feb 14, 2005 at 02:26:42PM -0600, Bill Kelly wrote: > A few months ago, we experienced several of the 'corrupted tape directory' > problems...more precisely described as a 'corrupted CM index', I think. [LTO-2 firmware/volume problems] > I'm seeing. I'm assuming that the firmware update *really* fixed the > problem...it's been out for several months with no subsequent updates. We also have (had?) problems with our LTO2 3584 library. We also experienced the symptoms you see. What we also noticed many "silly" TapeAlerts: errors about a failing firmware upgrade when no upgrade was attempted, errors about snapped tapes, errors about failed drives. No "real" errors (e.g. I/O errors) occured. When these "silly" TapeAlerts happen, there's a chance that the volume that's in the drive gets a corrupted tape directory. These later show up as "real" TapeAlerts and bad performance. This problem has been open for months and months, during which we received two special firmwares, one for diagnosis and one in which the problem should be solved. We now no longer see silly TapeAlerts, but every now and then a TapeAlerts mentions that "Volume XXX is not data-grade. Your data is at risk." Again without any real errors. We're now on firmware level 4AP0, which should contain a fix for our silly TapeAlert and tape directory corruption problem. Are TapeAlerts enabled in your setup? If not, try enabling them and look in the TSM log for any TapeAlerts. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Is there a TDP Plug-in into the BAClient at V5.3
On Fri, Jan 28, 2005 at 07:56:26AM -0800, TSM_User wrote: > I thought I saw someone post something about a plug in for the TDP into the > regular BA-Client. Maybe I misunderstood because I couldn't find anything in > the manuals. I thought I'd ask here to see if anyone has seen this or not. We run Lotus Domino on Windows 2000. With the 5.3 clients, the TDP for Lotus Domino is accessible through the Web-interface of the standard B/A-client. It's in the manual. Be sure to select the plugin when installing the TDP for Domino though. You might need to select the Custom install, not sure. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Anyone installed 5.3 yet?
On Mon, Jan 24, 2005 at 10:48:26AM -0500, Zoltan Forray/AC/VCU wrote: > When the 5.3 server starts existing FILE volumes in FILLING state are > marked as full. I read this to mean that when the 5.3 server starts *for the first time* after an upgrade, FILE volumes in FILLING state would be marked as full. However: that is not the problem I'm having. When I define a smallish FILE volume (64 MB), in certain cases it fills up after only 0,5% usage. This caused my stgpool that is used as the destination of the DIRMC option to *really quickly* fill up after migrating to 5.3. Also, offsite reclamation started issuing errors reporting it couldn't access onsite versions of the objects it was reclaiming on offsite volumes. All the objects it couldn't access on onsite volumes should have been in the affected stgpool. IBM has reproduced this problem and is analyzing it. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: unix sparse file backup?
On Tue, Jan 25, 2005 at 03:56:00PM -0700, Andrew Raibeck wrote: > > > file, here is allocation map". My understanding is that during restore, > > > TSM analyzes the data for lengthy patterns of zeroes in the data stream, > > > and that is how MAKESPARSEFILE works; but TSM does not do that analysis > > > during the backup. [...] > The TSM server is not actively "aware" of a file's attributes. The > "sparseness" of the file is determined at restore time, after the data has > been uncompressed, but before it is written to the disk. I've just been testing this a bit. I've created two files, both 10M large. Both are entirely filled with zeros. One is sparse, one isn't: $ dd if=/dev/zero of=test-real bs=1024 count=10240 10240+0 records in. 10240+0 records out. $ dd if=/dev/zero of=test-sparse bs=1024 count=1 seek=10239 1+0 records in. 1+0 records out. $ ls -l test-real test-sparse -rw-r--r-- 1 joskam staff 10485760 Jan 26 08:52 test-real -rw-r--r-- 1 joskam staff 10485760 Jan 26 08:52 test-sparse $ cmp test-real test-sparse $ du -k test-real test-sparse 10240 test-real 4 test-sparse Then I did (dsmc output compacted): $ dsmc sel test-real test-sparse Normal File-->10,485,760 /home/joskam/test-real [Sent] Normal File-->10,485,760 /home/joskam/test-sparse [Sent] Total number of bytes transferred:20.00 MB $ rm test-real test-sparse $ ls -l test-real test-sparse ls: 0653-341 The file test-real does not exist. ls: 0653-341 The file test-sparse does not exist. $ dsmc rest -makesparsefile=yes test-real Restoring 10,485,760 /home/joskam/test-real [Done] Total number of bytes transferred:10.00 MB $ dsmc rest -makesparsefile=yes test-sparse Restoring 10,485,760 /home/joskam/test-sparse [Done] Total number of bytes transferred:10.00 MB If what you're stating above is true, TSM would have no knowledge of whether a file was originally sparse or not. But it does: $ du -k test-real test-sparse 10240 test-real 4 test-sparse Both the above files were restored with MAKESPARSEFILE=YES, but only the file that originally was sparse is restored as such. I repeated this with 512 MB files, and got the exact same results. With these larger files however, it was easily observable that the sparse file and the normal file took the same amount of time to be sent to the TSM server. This also applied to restoring. So it would seem that sparse files are indeed sent to and stored on the TSM server without any "holes". However, it would also seem that the TSM server does indeed know whether a file was sparse or not. I did not test whether sparse files are restored with the same layout, i.e. whether the "holes" in restored sparse files are in the same place as the "holes" in the original files. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Anyone installed 5.3 yet?
On Mon, Jan 24, 2005 at 01:15:47AM -0700, Paul Fielding wrote: > I'm interested to know if anyone has tried installing 5.3 yet and what > they've found... Yes, we're on AIX 5.2 and TSM 5.3 here. Very smooth upgrade (from 5.2.3), runs OK. We did hit a bug with FILE volumes, and cannot reliably use FILE volumes. Fortunately, it was quite easily worked around by temporarily using DISK volumes. Not running the Administration Center here: command line only. Nice new features are Collocation Groups, the (optional) elimination of the need to issue REPLY to commands and the Offsite Reclamation limit. Nice features on the client side: the ability for the TDPs to "plug-in" into the Web-interface of the standard client, so you can do TDP for Domino-restore from the regular Web-client. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Questions concerning file/user exits
On Fri, Jan 21, 2005 at 09:34:52AM -, Davis Adrian wrote: [ passing filetextexit output to a script ] > I presume that this has already been thought of - So I was wondering if > anybody has tried it? If so are there any problems? This is exactly what I'm doing, and it works quite well. I created a named pipe, and let the TSM server use that as a filetextexit (use the APPEND option). On the other end of the pipe a Perl-script is running, which reads from the pipe and takes actions depending on the messages it receives. Beware that such a setup could result in unwanted signals to the TSM server (SIGPIPE). Test this well. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: DIRMC stgpool: DISK or FILE?
On Mon, Jan 17, 2005 at 10:51:55AM -0600, Rushforth, Tim wrote: > >Also, ADSM.QuickFacts says that sequential volumes have > >advantages in a database restoral situation. > Yes. This is also documented in the 5.2/5.3 ADMIN guide describing > difference between DISK and FILE pools. I asked this here, because IBM support asked why we were using FILE volumes instead of DISK volumes for our DIRMC stgpool. He recommends DISK volumes, but only mentioned performance as a reason for this recommendation. I answered that we were using FILE volumes for better handling of a database restoral scenario, and (historically) for the offsite reclamation performance problems (but this is no longer a valid reason). Thanks, -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: DIRMC stgpool: DISK or FILE?
On Sun, Jan 16, 2005 at 02:02:08PM -0600, Stapleton, Mark wrote: > Storage pools that act as DIRMC management class destinations [...] > contain redundant data; the normal client data destination pool also > contains a copy of all directory and file structure data. I wasn't aware of this. Is this documented somewhere? If this is the case, then why have a DIRMC stgpool at all? From ADSM.QuickFacts I've understood that in current TSM versions the restore process doesn't necessarily restore the entire directory tree with all correct attributes first, but that (if necessary) the directory is created with default permissions and attributes, and the permissions and attributes are set when the directory is processed on the backup media. > For multiple customers, I just set up a smallish disk-based storage pool > from 5 to 7GB with cache turned on; once a week, a migration to the > regular primary tape pool takes place. Nothing fancy needed here. So the data from the DIRMC pool ends up on a sequential primary pool? Am I correct in concluding that such a configuration does not result in problems, e.g. performance-issues when doing offsite reclamation? Thanks, -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
DIRMC stgpool: DISK or FILE?
Hi there, I have a question about which devclass to use for a storagepool that is exclusively used as DIRMC destination. The server originally was set up with a large enough DIRPOOL of type DISK. Later, DIRPOOL was reduced in size, and a DIRFILE stgpool was added that consists of FILE volumes. Clients still back up to DIRPOOL, but that pool is migrated daily to DIRFILE. Is this sort of setup still necessary with TSM 5.3 on AIX 5.2? I remember DIRFILE being set up because of performance problems with offsite volume reclamation (mainly due to sequential volumes being able to handle many files in one large chunk). Also, ADSM.QuickFacts says that sequential volumes have advantages in a database restoral situation. Would it be a good idea to return to a large DIRPOOL of type DISK and eliminate DIRFILE, on TSM 5.3 and AIX 5.2? -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Enabling caching to shorten housekeeping
On Fri, Jan 14, 2005 at 10:06:49AM +0100, Loon, E.J. van - SPLXM wrote: > We only ran into a nasty bug when using the file device class, so I hope IBM > will turn my PMR in an APAR and fix it as soon as possible... Let me guess: FILE volumes only store a tiny bit of data before (incorrectly) encountering end-of-volume? At least, that is the problem I'm encountering. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: ISC on WinXP or Unix
On Sat, Jan 08, 2005 at 09:52:29AM +0100, Richard van Denzel wrote: > I hope you'll agree that the requirements for ISC/AC are ludicrous, we > were better of with the old webadmin, without all the new "functionality". Maybe it's just me, but my idea of the old webadmin is that it's just dsmadmc but with a scrollbar. I installed TSM 5.3 and didn't bother with the new console. There are new options in 5.3 that make it possible to separate administrative client access from backup/archive client access on the network level, that come in quite handy. What I would *really* like though: dsmadmc just using stdin/stdout/stderr instead of its own console magic. Now I have to use Expect when scripting dsmadmc stuff; it would be *much* handier if I could just open a pipe to and from dsmadmc. -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Installation instructions say to execute "kill -9"
Hi there, In the Installation Guide for the 5.3 AIX Server, the following snippet appears: "If you started the server as a background process, connect to the server as an administrative client and issue the HALT command. If you cannot connect to the server with an administrative client, you must use the kill -9 command with the process ID number (pid) that is displayed at initialization." As Randal L. Schwartz put it: > No no no. Don't use kill -9. > > It doesn't give the process a chance to cleanly: > > 1) shut down socket connections > > 2) clean up temp files > > 3) inform its children that it is going away > > 4) reset its terminal characteristics > > and so on and so on and so on. > > Generally, send 15, and wait a second or two, and if that doesn't > work, send 2, and if that doesn't work, send 1. If that doesn't, > REMOVE THE BINARY because the program is badly behaved! > > Don't use kill -9. Don't bring out the combine harvester just to tidy > up the flower pot. What is the reason the Installation Guide tells us to do "kill -9"? -- Jurjen Oskam "E-mail has just erupted like a weed, and instead of considering what to say when they write, people now just let thoughts drool out onto the screen." - R. Craig Hogan
Re: Different stgpool for RMAN archivelogs and datafile backups?
On Wed, Nov 24, 2004 at 09:27:33AM -0600, Kowieski.Ingrid wrote: > INCLUDE /adsmorc/.../DB* DBTAPE_TDPNC (or RMAN_TAPE_POOL) > INCLUDE /adsmorc/.../LOG* DBDISK_TDPNC (or RMAN_DISK_POOL) Thanks! This is a much more elegant solution. I knew that INCLUDE/EXCLUDE statements were possible with TDP for Oracle, but apparently I wasn't clever enough to think of them. :-) Thanks again, -- Jurjen Oskam
Different stgpool for RMAN archivelogs and datafile backups?
Hi there, For several Oracle 9i databases, we use RMAN and TSM for Databases to store the backups in TSM. This works fine. However, both database backups and archivelog backups go to the same storagepool. The database is about 400 GB large, and this goes straight to tape. Archivelog backups are much smaller, perhaps 1 GB each time. Archivelog backups occur several times each day. These also go directly to tape. What this means is that one drive is occupied most of the time by the archivelog backups. This is inconvenient; the drive could be put to better use. How I think this problem could be solved: * create a disk stgpool, say RMAN_DISK_POOL * create a tape stgpool, say RMAN_TAPE_POOL * define the MAXSIZE parameter for RMAN_DISK_POOL * define the NEXTSTGPOOL paramater for RMAN_DISK_POOL to point to RMAN_TAPE_POOL * point RMAN to RMAN_DISK_POOL Will this work? Is there perhaps a better way of solving this problem? (Is it even a problem at all?) A possible problem with this setup might be that the TSM server doesn't know the size of an RMAN-backup ahead of time, so everything goes to RMAN_DISK_POOL and things break when too large 'files' end up in RMAN_DISK_POOL. (In our setup, the backup piece size is quite large, much larger than the typical archivelog backup. Choosing a correct value for MAXSIZE wouldn't be too difficult.) Thanks, -- Jurjen Oskam
Re: Oracle TDP and RMAN
On Fri, Nov 19, 2004 at 04:02:11PM -0600, Hart, Charles wrote: > In working with our Oracle DBA's they feel TSM< should manage the RMAN Oracle > Retentions etc... Is there a "Preferred method?" If so why? The preferred method is letting RMAN manage the retention policies of the backups. Especially so since Oracle 9i: RMAN works absolutely great for this. It was explicitly designed to be able to implement database retention policies. It keeps track of stored backups, and if you'd let TSM expire backups you'd lose some RMAN functionality like e.g. "CHANGE BACKUPSET foobar KEEP FOREVER". > "Note on page 85 of the RMAN Backup and Recovery Handbook it says: > > "If you are using a tape management system, it may have its own retention > policy. Note the "may". > If the tape management system's retention policy is in conflict with the > backup Note the "if". This "if" is quite important. The TDP for Oracle-documentation quite clearly explains how to configure the TSM retention policy so that it does not conflict with RMAN. > That is why I was leaving it up to the Tivoli retention policy to be used > instead of RMAN retention policies. If he keeps insisting on doing so, he'll be creating a dangerous situation for himself. For example, the "CHANGE ... KEEP FOREVER" will *not* work as expected: your backup *will* be gone when TSM expires it. In addition, he'll have to perform regular CROSSCHECK commands on the recovery catalog, and delete objects with the UNAVAILABLE status. RMAN still thinks it has the backups, while TSM expired them. So: configure TSM to not ever expire any backups made with RMAN. This is quite easy, and is explained in the installation documentation of the TDP for Oracle. Let RMAN take care of the database retention policy. That's what it's designed to do. -- Jurjen Oskam
Re: TDP for Oracle
On Fri, Nov 19, 2004 at 02:19:15PM +0100, Loon, E.J. van - SPLXM wrote: > 2nd quarter 2004. But I don't see any TDP (for Oracle) clients for Linux > platforms. It does exist, we downloaded ours from the Passport Advantage-site. It's in the same archive as the other platforms. -- Jurjen Oskam
Re: File space size is 0
On Wed, Oct 27, 2004 at 07:31:14AM -0500, Stapleton, Mark wrote: > If you'll read my response, I didn't say that the filespaces wouldn't > appear. I said that data backed up from that filespace doesn't show up; > i.e., the filespace shows 0 MB. Sorry, you're right. Missed that completely. -- Jurjen Oskam
Re: File space size is 0
On Tue, Oct 26, 2004 at 01:57:57PM -0500, Stapleton, Mark wrote: > >> I have taken backup of oracle database using TDP for oracle. The [...] > Data backed up through the TSM API does not show up in QUERY FILESPACE > commands. The TDP for Oracle does create filespaces that show up in QUERY FILESPACE. Even better than that, you can use the standard command line client to restore individual backup pieces to a filesystem: dsmc restore -virtualnode=TDP_ORACLE_NODE -pick /fsname/\* ./ If you restore the files (or symlink) to where RMAN expects them to be (as seen in the RMAN "list backup" command), you only have to allocate an RMAN channel of type DISK and RMAN will find the backup pieces where it expects them. This might come in handy if you want to duplicate or restore a not-too-large Oracle database to a host where you don't have the TDP for Oracle installed. (Of course, this is probably totally unsupported. :-) ) -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: MaxScratch Question
On Fri, Oct 22, 2004 at 03:34:46PM -0500, Hart, Charles wrote: > Lets say I have a tape stgpool that has 10 volumes and they all have > data on them, none being empty. To only let TSM allocate 2 more scratch > tapes to that pool do I set maxscratch to 2 or 12...? That depends on how those 10 volumes got defined into the stgpool in the first place. If the volumes were explicitly defined into the stgpool, the stgpool doesn't have any volumes in in that originally were scratch tapes. If you then want TSM to allocate 2 scratch tapes to the stgpool, MAXSCRATCH should be set to 2. If the 10 volumes were originally scratch tapes, and you want two more, MAXSCRATCH should be set to 12. If it's something in between, MAXSCRATCH should be set accordingly. (To find out whether a volume in a stgpool was originally a scratch volume, you can do a Q VOL ... F=D on the volume and look at the "Scratch Volume?"-line. If it says "yes" there, that volume is originally was a scratch volume, and "uses up" a MAXSCRATCH slot.) -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: TSM Licensing Question
On Fri, Oct 15, 2004 at 03:12:24PM -0400, Prather, Wanda wrote: > You can't download base 5.2, you will need the CD to get the license files. You *can* download base levels, including 5.2. If you have a Passport Advantage contract, you can download software from the Passport Advantage website. -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: Off-Topic: Question regarding IBM vs. EMC storage
On Thu, Sep 23, 2004 at 07:33:23AM -0500, Mike wrote: > > For tape storage, we have an IBM 3584 library, fibre-connected but not via > > the SAN: each drive has its own HBA in the AIX box. > > This is the same tape configuration we're using. Have you thought > about moving the tape fiber to SAN and attaching from TSM that way? Well, we thought about it, but decided it wasn't worth the trouble. :-) -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: Off-Topic: Question regarding IBM vs. EMC storage
On Wed, Sep 22, 2004 at 03:29:11PM -0400, Thach, Kevin G wrote: > I know there are numerous people on this forum using EMC storage, and > I'd appreciate it if a few of you wouldn't mind giving me your opinion For disk storage, we have an EMC Symmetrix in an Open Systems-environment. The SAN consists of EMC-labeled Brocade-switches. Attached to the SAN are several AIX boxes and IBM xSeries. The AIX boxes have rootvg on internal disks, the xSeries are completely diskless. For tape storage, we have an IBM 3584 library, fibre-connected but not via the SAN: each drive has its own HBA in the AIX box. We are very pleased with the performance and support of the Symmetrix. It works (and is supported) on every relevant AIX version. A nice touch is the TimeFinder-feature (a la FlashCopy): it can split off mirrors without needing to quiesce the original in a few tenths of a second. For an Oracle database that means: only having to put the tablespaces in BEGIN BACKUP for a second. Also, the EMC technical support is quite good. Contrary to IBM technical support (at least our experiences with it), EMC technical support can quite easily (and often on-the-spot) involve a higher level of support. This is not often necessary though. As for software support: our Symmetrix has caused no problems on any version of AIX we've used, and that's with PowerPath (the EMC equivalent of SDD). Unfortunately, we weren't that lucky with our 3584. Driver and firmware problems have caused our TSM machine (AIX 5.2) to crash hard several times. (Fortunately this is solved now, but now we're getting nonsensical TapeAlerts from the 3584. This case is now open for several months.) -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: TDP vs. FlashCopy thoughts
On Tue, Sep 14, 2004 at 10:59:02AM -0600, James Choate wrote: > When a FlashCopy is initiated, the DB/Mail server that you are flashing should be > quiesced prior to the FlashCopy being initiated. Us Symmetrix users don't have such a restriction. :-) But for the original discussion: think about FlashCopy (or your storage system's equivalent) as a way to restore your datafiles (in case of a database) very fast. By restoring them from the FlashCopy, you don't need to restore them from TSM. -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: TDP for Oracle --exception raised in RPC: ORA-19602: cannot backup or copy active file in NOARCHIVELOG mode ----- Stuck -- Help
On Fri, Aug 27, 2004 at 03:45:19PM +0500, Muhammad Sadat wrote: > RMAN-10035: exception raised in RPC: ORA-19602: cannot backup or copy > active file in NOARCHIVELOG mode > Please advise where i go wrong? Your database is in NOARCHIVELOG mode, and you're trying to perform an inconsistent backup. This is not possible. You should read the Oracle documentation about backup and recovery for an explanation; the error you're encountering is not a TSM error. It's important that you understand the basics of Oracle backups, otherwise you might be unpleasantly surprised when you need to restore something. -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
TapeAlerts on AIX and 3584 library
Hi there, Some time ago, someone mentioned that many TapeAlerts were appearing in the log. Some/many/all of those alerts were "weird", in the sense that the error was e.g. that a firmware update failed, while no firmware update was in progress. For some time now (a few months), we're experiencing similar symptoms. We get a daily dose of TapeAlerts of varying importance. The most serious ones say things like "The tape is not data-grade", "Your data is at risk", "The tape has snapped", "There is a problem with the library mechanism", but we also get less serious ones that say things like: "The firmware update failed", "The drive is outside allowed temperature range", etc. I've read the documentation about TapeAlerts, and as far as I can tell we've had them all. This happens on a 3584 library with firmware version 4090 and four LTO2 drives with firmware level 4770. The problem also occured on every earlier level we've had (the library is in our possession since the beginning of the year). The drives are direct-attached to four 6778 fibrechannel HBAs in a 7026-M80 running AIX 5.2 ml2 with the latest Atape (but here too the problem occured with earlier levels). The strange thing is that this problem only occurs on the drive of which the controlport is used. Normally, we use one controlport, and that is /dev/smc0 attached to /dev/rmt0: the standard (and required) 3584 controlport. During troubleshooting, another controlport was created on another drive, which appeared as /dev/smc1 in AIX. TSM was configured to use /dev/smc1 instead of /dev/smc0, and from that moment the TapeAlerts were logged against /dev/rmt1 and /dev/smc1. So, only the drive of which the controlport is in use, complains about faulty tapes and/or operating conditions etc. The same tape in any other drive doesn't cause any tapealerts. These TapeAlerts *seem* to be only cosmetic errors, because no other errors are logged at all. No I/O errors or anything in TSM, and nothing whatsoever in the AIX error report. Even the drivelog on the library doesn't say anything weird, as far as I can tell. (Note that I say *seem*, because in the past tapes *were* corrupted by (probably) this issue. More correctly: the tape's memory had to be recreated by reading it entirely using tapeutil. This corruption seemed to have gone away either because of a drive replacement or a higher firmware level.) I already have a call open with IBM. That call is progressing at glacial pace however, and they make me jump through several hoops. So, I'm interested in any experiences of other people. Thanks for reading this far. :-) -- Jurjen Oskam "I often reflect that if "privileges" had been called "responsibilities" or "duties", I would have saved thousands of hours explaining to people why they were only gonna get them over my dead body." - Lee K. Gleason, VMS sysadmin
Re: new Services look like DSM but are they?
On Tue, Jul 13, 2004 at 02:03:05PM +0200, Martin Krau? wrote: > One of our file servers(W2K) was hacked last night, and I discovered > some new services that were not there before. they have the names: The only safe way to handle a compromised machine is to completely erase it, and restore from known-good media and backups. You cannot trust anything on that machine now, unfortunately. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance."-MS Q308417
Re: 3582 supported on TSM 5.2.2 AIX ?
On Mon, Jul 05, 2004 at 10:32:22AM -0400, Richard Sims wrote: > It cannot be stressed enough that direct feedback to IBM is vital to helping > improve things for all of us - and, unfortunately, almost no customers take the > time to do that, which can leave lingering problems for others to run into. While I fully agree with you, it *is* frustrating that a somewhat common response to a problem is "works as designed", and the impression is left that feedback is not something IBM support is waiting for. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance."-MS Q308417
Re: admin password on 5.2
On Wed, Jun 30, 2004 at 07:34:20PM +0200, Otto Schakenbos wrote: > try to shutdown the tsmserver (sending a kill signal should do the trick Don't send the KILL signal, send a TERM, INTR or HUP instead (in that order, until one works). Only use KILL as a last resort. The "kill" command sends TERM by default, so "kill " should work. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance."-MS Q308417
Re: Hey Big Blue....Why not MOVE DATA across copy pools?!?
On Wed, Jun 09, 2004 at 12:10:32PM -0500, Stapleton, Mark wrote: > I used to think the same thing, but it occurred to me that, while there > is a one-to-one correspondence between a file in an copy pool and the > copy of the file in a primary pool, there is not necessarily a > one-to-one relationship in the other direction. Especially with MOVE NODEDATA used to move data from one primary pool to another, where the first primary pool is backed up to COPY01 and the second primary pool is backed up to COPY02. The data in COPY01 is now (sort of) 'orphaned', and cannot be removed. (Not easily, at least) Maybe a DELETE NODEDATA would be nice. (However, a MOVE NODEDATA that would work on offsite volumes would be *much* nicer. That way, you can once in a while 'collocate' all files of a node manually, and greatly speed up a DR when all you got is the copypool.) > The more I look into this, the more I am inclined to never use more than > one copy pool for any given TSM server. Especially so with a MOVE NODEDATA that would work on offsite volumes. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance."-MS Q308417
Re: Question about the Move NODEDATA Command
On Wed, Jun 09, 2004 at 08:43:19AM +0100, Copper, Steve wrote: > pools? i.e will the copy storage pools, after doing a backup of the primary, > have all the data for that client on one tape as well, No, the copypool will not be affected. > or do I have to issue > the move node data command for the copy storage pools as well? Yes. Keep in mind that MOVE NODEDATA is not very useful for offsite copypools, since it requires the copypool-volumes to be onsite and available. (It doesn't work like e.g. MOVE DATA does, which gets the files from the primary pool) -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance."-MS Q308417
Re: dsmcad output redirection to dev null not effective on AIX.
On Mon, May 24, 2004 at 12:50:34PM -0700, Robert Clark wrote: > We have a few systems where the dsmcad appears to be outputting a countdown > on /dev/console, despite the output being directed to /dev/null in inittab. Was both stdout and stderr redirected? (e.g. something like dsmcad 2>&1 1>/dev/null) Otherwise, you could check with truss to see what dsmcad is doing. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance." - 308417
Re: SSH
On Wed, May 12, 2004 at 08:41:01AM +0200, Crawford, L. (Lindy) wrote: > This is off the tsm topic...Please could someone assist me. I have ssh setup > on our Aix server, but when a user password expires, it does not prompt the > user to change it.is there a setting I should put in place. You're much more likely to get an answer to your question on the AIX Toolbox mailing list. If you read that list, you would've seen that I had the same issue and that upgrading to the SSH released by IBM on May 4 fixed it. :-) -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance." -MS 308417
Re: 3584/lto2 code
On Mon, Apr 26, 2004 at 10:59:17PM +0200, Stef Coene wrote: > > It is a shame if IBM is no longer providing this microcode to the > > customer. If they are, then it is also a shame that the FTP site is so > > out-of-date. > 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. And, yet again, the latest versions on that page are 38D0 and 3480. *The original poster already *has* those versions*. -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance." - 308417
Re: 3584/lto2 code
On Mon, Apr 26, 2004 at 07:53:16PM +0200, Uwe Schreiber wrote: > did you look at > > service.boulder.ibm.com/storage/358x/??? Did *you* look? 38D0 is the latest there, and he already has that version. Anyway, I also was on 38D0 and Atape 8.4.1.0. When one LTO2-FC drive had problems, it took down the AIX 5.2 machine it was attached to with it. It crashed hard, multiple times. Support told me that it was a problem with 38D0 in combination with Atape 8.4.1.0. Atape 8.4.1.5 was released to circumvent this, but the CE also placed 42D0 on the drives and 3484 on the library (a 3584). -- Jurjen Oskam "Avoid putting a paging file on a fault-tolerant drive, such as a mirrored volume or a RAID-5 volume. Paging files do not need fault-tolerance." - 308417