Slow Domino backups
My Domino backups are slow - average about 10 gb/hr. I'm wondering what I might have configured wrong or not optimally. Regular backup client 5.2.2 backups are not slow, just Domino. I don't know much about TDP Domino setups. Any ideas? TSM server 5.2.2 on AIX 5.2 Domino TDP 5.1.5 Domino ver 6.5 on Win2003 server domdsm.cfg BUFFers 3 BUFFERSIze 1024 LOGFile domdsm.log LOGPrune 30 REPlace yes SUBDir no MOUNTWaityes NOTESIniPath LastPruneDate 01/09/2005 11:02:48 dsm.opt commm tcpip tcpport 1505 tcpserveraddress 10.10.x.xxx tcpwindowsize 63 nodenamedsmail2-dom passworda generate schedmodprompted schedlogname c:\Progra~1\Tivoli\Tsm\domino\domsched.log schedlogretention 3 compression on Include * Wondering: Should I turn compression off Should I change buffer and buffersize settings Am I using problematic TSM/Domino/TDP/OS versions or combination Thanks, David
Restoring the oldest version
A customer wants restored the oldest version of every fileTSM has for a certain node. What's the best way to do this? Example: If there is only one copy of a file and it is active, they want that one. If there are 2 inactives and 1 active, they want the oldest inactive. Thanks, David
Data on my tapes
Hello, Ok, how much data are you getting on a 3590 K tape. From reading past posts, I know the Pct Util and Est. Capacity method is suspect at best. All I need is an average you guys have seen using the following: assume the tape is full and has had no files expired or reclaimed 3590 E1A drive devclass FORMAT is set to DRIVE (should I be using something else?) no client compression drive compression 3590 extended K tapes additional info if it helps: data per tape is either all NT or all UNIX mostly small files on NT about half small files on UNIX and half Oracle database files (2-20gb) The reason I ask is that my best guess seems to be such a wide range - I think between 60 and 150 gb per tape. Also, according to est. cap. and pct util, UNIX volumes shows about twice the capacity of NT volumes I know IBM says these are 40gb native and 120gb at 3:1 compression but what is your real world experience of just how much data fits on a tape? Thanks for your input David (still new, but learning everyday)
LTO and 3590
Hello, I have read some archives about LTO vs 3590. I get mixed feelings about which is better depending on the situation. I would like to ask for your opinions on my specific case. Here are the facts: I am a TSM newbie 6 months. Current hardware is: 3494 w/2 3590E1A drives 3466-C00 w/H50 server 288gb SSA 7133-D40 We are needing to about double our current hardware (rough estimate) The additional backups will be 70/30 small file/large file (all from NT) The are 3 options on the table 1) add 3590EA1 drives (and additional frame) 2) add LTO drives (and whatever frame) 3) sell 3590 system and replace with LTO system My first thoughts are to just add 3590's. Been working fine for a year. I'm still gathering price info. but if cost was basically the same what would you do? I don't think cost will be significantly less for any of the three. Maybe I'm wrong. Advantages/disadvantages to these options? If cost were much lower for LTO would you go with it? If complete LTO replacement, would data migration off 3590 be a nightmare? Also, I don't really need fast restore times Thanks for any advice David
MSCS and the GUI
Hello, Is it possible to see the shared drives in a MSCS through the backup GUI? I see only the local node drive. I have created schedules for the cluster groups that seem to be working(according to their log files). I have clusternode=yes in the opt file for the cluster groups and not in the opt file for the cluster node. If I put clusternode=yes in the opt file for the node the GUI errors out with invalid line clusternode=yes Thanks
MSCS and GUI oops
Please forgive my ignorance. clusternode=yes is the problem should be just clusternode yes (no =) Yes, I feel dumb. Thanks David
wrong stgpool
I have 20 NT servers with a dsm.opt that has an include * ntclass line. These run a nightly backup that has worked fine for months, backing up to the backup copygroup of mgmtclass ntclass with a dest. of ntdiskpool. This is all normal. Here is my problem. I created a new mgmtclass called notesclass, which has a backup copygroup with a dest. of notespool (this is a tape pool). Now, keep in mind the 20 NT servers have include * ntclass as the only mgmtclass. I'm seeing that q content of some of the volumes in notespool show data paths from most of the 20 NT servers. Notesclass is not a default mgmtclass and the notespool volumes were aquired empty from scratch. How could this data get on the wrong stgpool volumes? Also, the NT servers data paths on the volumes in notespool point to directories only, no files. I'm also seeing this with my UNIX mgmtclass (having data paths on notespool). I am new to this and have never run any audits or unloads or anything on the database or volumes. Could that help? Still too new and dumb to know why this is happening. Thanks.
db vols and disk pools
I have sixteen 18gb SSA disks (one adapter, two loops, one 7133-D40 shelf). I've estimated I'll need a 36gb database and a 4gb recovery log. The rest of the storage will be divided into two disk pools. My question is: How to setup physical disk wise? Should I keep the db on four 18gb disks (includes mirrors) and the disk pools on their own disks. Or should I spread out the db on more disks and share those disks with the disk pools. Which has less contention? looking for advice. Thanks.
block level backups
Hello, Does anyone know if TSM does, or will in the future do, block level backups? I want to back up large files (5 to 10gb and databases) that have only small daily changes. They will be backed up over slow links (128k lines). As far as I know, TSM does only file level but there are other products out there(VytalVault) who claim they backup only the changes to a file at the block level. I want to use TSM for this but I have to show the advantage. Anyone familiar with VytalVault? Sorry, but I am kinda new to all this backup stuff. Thanks