Reclamation part in maintenance fails

[Eps]

Hi,

I have a maintenance script, that also do a tape reclamation.
Today script failed with this reason:

ANR4930I Reclamation process 163 started for primary storage pool LTOPOOL2 automatically, threshold=60, duration=None

ANR1044I Removable volume AUV179S is required for space reclamation. (PROCESS: 163)

ANR1044I Removable volume AUV180S is required for space reclamation. (PROCESS: 163)

ANR1176I Moving data for collocation set 1 of 1 on volume AUV179S. (PROCESS: 163)

ANR1082E Space reclamation is ended for volume AUV179S. There is an insufficient number of mount points available for removable media. (PROCESS: 163)

ANR0985I Process 163 for SPACE RECLAMATION running in the BACKGROUND completed with completion state FAILURE at 10:27:11. (PROCESS: 163)

ANR4936I Reclamation of storage pool LTOPOOL2 has ended. Files reclaimed: 0, Bytes reclaimed: 0, Files reconstructed: 0, Unreadable files: 0.

Device class mount limit in set to 100.
Tried to increase MAXNUMP value for node, up to 20.
Also tried library audit.

Tapes AUV179S and AUV180S are full and private. Bot tapes contain data.
Both tapes are read/write.

Regards,
Eps.

 

[moon-buddy]

ANR1082E Space reclamation is ended for volume AUV179S. There is an insufficient number of mount points available for removable media. (PROCESS: 163)

How many tape drives do you have? Are there two drives free to run reclamation?

 

[Eps]

My problem is that I have one drive only.

I made a RECLAMATION pool, type FILE.

Defined one scratch tape with "define volume..." under just created pool.

When I issue "q libv", I see that my tape is in Private state.

But "q stgpool" shows no Estimated Capacity under RECLAMATION pool.

And the same errors happen during reclamation.

 

[moon-buddy]

My problem is that I have one drive only.

I made a RECLAMATION pool, type FILE.

Defined one scratch tape with "define volume..." under just created pool.

When I issue "q libv", I see that my tape is in Private state.

But "q stgpool" shows no Estimated Capacity under RECLAMATION pool.

And the same errors happen during reclamation.

No can do! Reclamation works with sequential devices and as such, you need really two drives. Here is a snippet from "help reclaim stgpool:

This command can only be used with sequential access storage pools. The storage pool data format cannot be NETAPPDUMP, CELERRADUMP, or NDMPDUMP. Storage pools defined with a CENTERA device class cannot be reclaimed.

If you search this forum you will see that time and time again as discussed reclamation is always done using two tape drives

 

[Eps]

So.. there is no way to do reclamation with one physical drive?

 

[moon-buddy]

So.. there is no way to do reclamation with one physical drive?

You can try moving - move data command - all of the tape (assuming that the tape is a primary pool) data back to a disk pool (primary to primary but not copy to primary) and migrate it back to the tape pool.

Not a good way to do things.

Buy a second and third drive - this will make your life easier.

 

[JohanW]

If it's a primary pool, you can reclaim it with a single drive. See HELP UPD STGP, look for the RECLAIMSTGpool parameter.

If it's a copy storage pool, you can't. NO, really, you can't. TSM wasn't designed for single drive environments.

 

[Eps]

Ok.. I will try to organize reclamation with RECLAIMSTGpool.


Thanks for the answers.

Eps.

 

[moon-buddy]

If it's a primary pool, you can reclaim it with a single drive. See HELP UPD STGP, look for the RECLAIMSTGpool parameter.

If it's a copy storage pool, you can't. NO, really, you can't. TSM wasn't designed for single drive environments.

To be honest, I have never used this option as I am always blessed with multiple tape drives and that is why this hasn't crossed my mind.

It is good that you have pointed this out.

 

[BBB]

My thinking is ... if you only have 1 tape drive ... why are you using TSM?!!!

Tiny environment = use something simpler.
Big environment = TSM all the way...

 

[Eps]

When I came, the TSM environment have been already created. Before me...

"Tiny environment = use something simpler." For example?

Eps.

 

[JohanW]

My thinking is ... if you only have 1 tape drive ... why are you using TSM?

Because it's a DR site. Or a test site. Or because it's what you have. Or know.

At home, I run a small but full TSM using disk only. At my brother's, I run a small but full TSM using a single tape drive, but he doesn't do copy pools (sometimes one backup is enough). Excellent reporting, useful testing for the corp environment, efficient use of disk space because of incremental forever, superior versioning of backups. If we had the bandwidth, we'd be offsiting our data to each other as well. One of these days .. As it is, if the house burns down, I hope I can save the cats. Meanwhile, I have a couple of clients (read: my parents, and comparable) that do backups over the Internet to my server, and, after the initial backup, incrementals forever sure are nice.

But really, for serious production environments, either get multiple tapes drives, or none at all and just a lot of disk. Both work great. If you're offsiting to disk, you need enough bandwidth, but not ridiculously much, because you can just saturate a dedicated link for a good part of the day. TSM'll just chug happily along doing its thing. For a DR site, one tape drive is enough until you are going into full recovery after an actual disaster, then you want to have more than one available after a single phone call of course. But you can plan for that.

 

[BBB]

Yohan I understand, however I assumed that this was for a normal production environment. Not that many people run TSM at home, and I figure not too many of those would use LTO drives with label names going up into the high 100's and be posting on ADSM.org As the site is doing backups and reclamation, I figure its most likely a production site where this is being experienced.

I still believe that for a small site that doesn't understand TSM well, it may not be the best solution *for them*. It might be *technically* the best solution, but TSM needs a lot of care and maintenance and most small sites don't have the resources or skills for this.

I also don't feel that using only disk is appropriate for an enterprise environment if data security is critical. There are failure scenarios where you can lose your primary and secondary disk copies at the same time, and you are then a goner... For example, a forum website just recently lost 12 years of postings from 60,000 users when both their primary server and backup server were hacked into and destroyed. No they weren't using TSM but if they were and were storing everything on disk, they still would have lost everything. I think with non-accessible offsite tape, you get better protection - if you need it. And yes it is more painful but what can you do.

 

[Eps]

Solved.

So I try to do reclamation...

I have tape pool - LTOPOOL2. Defined a new poll RECLAMATION.

Then:

update stgpool LTOPOOL2 reclaimstgpool=RECLAMATION

But, because RECLAMATION pool is DISK, it can update the reclaimstgpool.

My question. Can I(if yes, so how) make a sequential pool with volume under this pool that it will be DISK?

Eps.

 

[JohanW]

Not DISK. FILE. DISK is random access, FILE is sequential access. See HELP DEF DEVC.

 

[JohanW]

.. I also don't feel that using only disk is appropriate for an enterprise environment if data security is critical. There are failure scenarios where you can lose your primary and secondary disk copies at the same time, and you are then a goner .. I think with non-accessible offsite tape, you get better protection ..

How would you feel about two TSM Servers at geographically dispersed locations, with a dedicated WAN link between them, firewalled so they can really only talk TSM to each other, with one acting as copypool for the other? Sure, there are scenarios where inaccessible tape will save you while separate servers won't, but that was what I had in mind for the disk only scenario. And with disk and wires, you don't have to have couriers who f*ck things up with physical tapes ..

(Sorry for the thread drift guys. Back to reclamation now.)

 

[BBB]

How would you feel about two TSM Servers at geographically dispersed locations, with a dedicated WAN link between them, firewalled so they can really only talk TSM to each other, with one acting as copypool for the other? Sure, there are scenarios where inaccessible tape will save you while separate servers won't, but that was what I had in mind for the disk only scenario. And with disk and wires, you don't have to have couriers who f*ck things up with physical tapes ..

(Sorry for the thread drift guys. Back to reclamation now.)

Ok yes I agree, using 2 separate tsm servers like that, with geographic distance between them will cover most scenarios. Its a lot better than 1 server with everything on disk, even if it has an offsite disk copypool via dedicated link as its still controlled by one server.

The situation is still open to a security event affecting both servers (eg. security flaw in tsm leaves both servers vulnerable - firewall or not), a login account being hacked (assuming remote access is still allowed through firewall) or a destructive virus/worm which both servers get from their own local networks. Not saying these cases are very likely though, but they are possible