Hello Zoltan,

  sorry, I misread this as damaged dedup pools, so next try:

--
Michael Prix

On 5/11/21 10:40 PM, Zoltan Forray wrote:
Your response does have relatable instructions but I have questions:

1.) Identify the damaged, or lost, containers ("volumes" of the
dedup pool) and mark them damaged or destroyed. Use "audit container"
for this. In your case, as you might have lost some filespaces
completely, "update stgpooldir" might be a faster solution.

We changed all managementclasses that pointed to this stgpool to use
other stgpools so daily backups could run.
We have been running lots of movedata and when they fail, the volume is
marked readonly. But there has been a substantial amount of data
"successfully" moved.
Running AUDIT VOLUME (s) right now

2.) add additional space to the container pool as needed. This can
be NFS storage, but be prepared for a massive performace drop.

Done.  But how does it know to restore *ONLY* to the new mountpoint/storage
area vs trying to write to the first, damaged storage area?
All directories a storagepool can write to are defined in the respective device class (dir=...). Just remove the damaged directories from the definition of the device class and they won't be used for further writes. This doesn't affect the readability of volumes residing therein, as the path and name are stored within the definition of the volume.

3.) checkin any offsite tapes of the respective copy container pool
Done.  Did some test "restore volumes preview=yes" and they happily found
the offsite tapes

4.) run "restore stgpool"
This is where I hesitate (back to my statement above -  But how does it
know to restore *ONLY* to the new mountpoint/storage area vs trying to
write to the first, damaged storage area?

As I wrote, adjust the device class.



On Tue, May 11, 2021 at 4:31 PM Zoltan Forray <zfor...@vcu.edu> wrote:

Thanks for the reply.  I should have clarified these are regular FILE disk
pools, not Containers......

On Tue, May 11, 2021 at 4:21 PM Michael Prix <mich...@prix.one> wrote:

Hello Zoltan,

recovering deduped pools after a desaster, even a partial one, has some
drawbacks.
Deduplication information is bound to the storagepool, not the
TSM-instance, so you have to restore the lost data to the same
storagepool. There is no "restore volume", as there are no volumes as
you know it with dedup pools. You lost part of unique data residing in
the pool and that may be affect every single backup object stored in the
pool.
So the only possibility you have is:
1.) Identify the damaged, or lost, containers ("volumes" of the dedup
pool) and mark them damaged or destroyed. Use "audit container" for
this. In your case, as you might have lost some filespaces completely,
"update stgpooldir" might be a faster solution.
2.) add additional space to the container pool as needed. This can be
NFS storage, but be prepared for a massive performace drop.
3.) checkin any offsite tapes of the respective copy container pool
4.) run "restore stgpool"
5.) after completion, run "audit  container stgpool=..". During the
audit, you can use the pool for backup and restore. If you run into
errors with restores,  wait till the audit completes and perform another
restore stgpool.
6.) Any damaged objects left can be repaired with additional backups or
maybe you can let the objects expire. Damages objects might "heal" by
themself during subsequent client backups into this pool. If the damaged
chunck appears during a client backup, all pointers to that chunck will
be rebound and the damaged original chunck will be expired.

--
Michael Prix

On 5/11/21 9:55 PM, Zoltan Forray wrote:
7.1.7.400 Linux server

We had a multi-disk failure event occur in one of our PowerVaults so an
array is damaged along with lots of volumes.

Tried to do a "restore stgpool" or "restore volume" to other stgpools
but
it says you can't do that for deduped pools/volumes.

We have been running movedata all night and it some cases it works or
works
for a while and fails making the volume readonly.

We have created an NFS mount to handle recovering what we can from
offsite
tapes but can't figure how to do this.

Trying to follow this procedure:

https://www.ibm.com/support/pages/recovering-lost-or-damaged-file-volumes-deduplicated-storage-pool
but have soooo many questions it doesn't seem to answer.

Following the above procedure, I am running AUDIT VOLUME but there are
thousands of volumes to process.

We can't just restore a volume since I am guessing it will try to put it
back into the same stgpool which we don't want to do since this PV is
barely running, haven't got the replacement disks, etc......

So how do I do this?

--
*Zoltan Forray*
VMware Administrator
IBM Spectrum Protect Administrator
Virginia Commonwealth University
UCC/Office of Technology Services
www.ucc.vcu.edu
zfor...@vcu.edu - 804-828-4807
Don't be a phishing victim - VCU and other reputable organizations will
never use email to request that you reply with your password, social
security number or confidential personal information. For more details
visit http://phishing.vcu.edu/
<https://adminmicro2.questionpro.com>

--
*Zoltan Forray*
VMware Administrator
IBM Spectrum Protect Administrator
Virginia Commonwealth University
UCC/Office of Technology Services
www.ucc.vcu.edu
zfor...@vcu.edu - 804-828-4807
Don't be a phishing victim - VCU and other reputable organizations will
never use email to request that you reply with your password, social
security number or confidential personal information. For more details
visit http://phishing.vcu.edu/
<https://adminmicro2.questionpro.com>


--
*Zoltan Forray*
VMware Administrator
IBM Spectrum Protect Administrator
Virginia Commonwealth University
UCC/Office of Technology Services
www.ucc.vcu.edu
zfor...@vcu.edu - 804-828-4807
Don't be a phishing victim - VCU and other reputable organizations will
never use email to request that you reply with your password, social
security number or confidential personal information. For more details
visit http://phishing.vcu.edu/
<https://adminmicro2.questionpro.com>

Reply via email to