replace L32 frame in TS3500

[ChrisRees]

Hi

The current environment looks like this..
1 x TS3500 with a L32 library frame plus
1 x D52
2 x D53
1 x S54 frames
1174 cartridges

The library is used by a library manager and 3 library clients running TSM 6.3.4.2
(old I know but its nearly decomissioned)

The library has 22 LTO5 tape drives

So the people that support this are planning to replace the L32 library frame and I wanted to get down a plan of action for this procedure.

Will the tape drive WWPN's stay the same ?

High level I guess its along the lines of

checkout all the tapes
del paths / drives / library definition on lib manager
swap hardware
redefine libary / drives / paths
checkin scratch tapes
checkin private tapes with correct client owner


Thanks alot

 

[RajeshR]

There is a possibility if the frame get changed tape drive WWPN's also might change with my recent experience in a different model tape library.

Save your library configuration before start of replacement (Using GUI)
Save volhist,devconfig files as well.
Delete paths,drive,libr paths, and library.
swap hardware ( deleting any logical libraries before swap will do good, sometimes drives might not detect in new frame if the library does not get deleted logically of course it's HW support team call)
check tape devices from OS end.
redefine library/drives/paths
checkin scratch tapes
checkin private tapes with respect to server
checkin cleaning cartridges if assigned to library instead of system.
once everything looks good run audit library from tsm server otherwise your tape drives state might show as unknown instead of empty and its recommended to perform audit.

 

[RecoveryOne]

Chris,
Are they replacing the L32 with an L53? Or they swapping in a new/used L32 and importing the same config?
In all likelihood you will need to have the WWN's rezoned to the TSM server(s). At least I did when we went from an L32 to L53.
Quick overview of what we did. Going from memory here so I may have missed something critical. This is assuming you are keeping the same library name.

1. Full DB backup with all the trimmings (paranoid).
   
2. Take the drives/paths/library off line. At this point it is safe to delete DRIVES and their PATHS. Don't delete anything else if I recall correctly.
   
3. CE Pull tapes / drives from L32 (and other frames as needed) put in new frames/empty cells
   
4. CE installs new frame, checks firmware etc etc.
   
5. Boot up new frame , get WWN's, have san team rezone the WWN's to TSM.
6. Walk the bus, find your devices/media changer
   
7. "UPDATE LIBRARY <libname> ser=autodetect" This step clears out the serial number. If you try to do the next step, in my environment it kept complaining.
   
8. UPDATE PATH <tsmserver> <libname> srct=server destt=libr device=<device> autodetect=yes -- actlog should return: Library <libname> with serial number is updated with the newly discovered serial number 000xxxx (Notice the double space after 'number')
9. define your drives
10. update the library path online.
11. AUDIT LIBRARY <libname> refresh=yes -- You will likely get a lot of ANR8470W's "Initialization failure on drive"
12. DEFINE PATH <tsmserver> <drive> srct=server destt=drive libr=<libname> device=<device> online=yes
    
13. AUDIT LIBRARY <libname> checkl=bar
14. At this point you should be able to checkout/checkin any volumes that audit lib didn't like. Think I had one, that I had to do that to and then ran an audit vol on it.

I want to say when we first defined the drives, we had to define the paths to with autodetect=yes to clear out the serials. Not 100% sure.
At no point in this procedure did I have to take our TSM server offline.

Hope this helps.

 

[marclant]

Full DB backup with all the trimmings (paranoid).

I would call that sane.

 

[RecoveryOne]

I would call that sane.

"Strange how paranoia can link up with reality now and again." (Quote from Philip K. Dick)

One thing I forgot to mention, it took two days to fully swap my L32, D32 frames. That was with two CE's onsite. It partly was due to where my library was located. They had to disassemble all the frames to remove the base and secondary units. That was fairly time consuming.
If you are keeping the TSM server running, you will want to make sure you have enough disk space and perhaps some extra should things go wrong if you migrate data to tape or use it as part of HSM.
And obviously pause any admin schedules that do tape operations.

 

[ChrisRees]

Thanks so much for the replies and advice, I'll definitely be taking full db backups and multiple copies of volhist and devconfig... I'll engage with the SAN team to prepare for rezoning all the drives.

I like your procedure RecoveryOne, a lot easier if I don't have to checkout all the tapes and completely delete / redefine the whole library

 

[RecoveryOne]

Yeah, I cringed at the thought of having to check out a thousand tapes as well.
Oh don't know what frame you are going to but there was a snafu with using Virtual IO and the 2nd IO slots(?) being active. If I get some time I'll dig though my notes see if I can find out exactly what it was.

Please don't take what I posted above at 100% what you should do, going from memory/scattered notes. But it should at least point you in the right direction.

One last thing if you can, make sure you accomplish bringing the library back in/online while the CE's are on site. There was an issue with the OP panel and the 2nd IO. Thankfully that was discovered during our testing, and IBM was able to bypass the 2nd IO to keep the library running until a replacement OP panel arrived.

 

[ChrisRees]

We completed this change, with no issues.. the tips were invaluable so thanks a lot.

 

[RecoveryOne]

Welcome and good to hear no issues!