lost data !

[nightcrawler]

Hi,

I ran into these IO errors

ANR9999D_2133050718 FileReadNC(pvrfil64.c:2028) ... Insufficient bytes read
ANR9999D_4186073719 FileReadNC(pvrfil64.c:2042) ... Invalid block header

while auditing two file volumes (NFS mount)

Is there any way to recover from this, knowing I don't backup the storage pool?

TSM 6.2.5 on an AIX 6.1 box

 

[Trident]

Hi,

If you are sure that your network is fine, retry opration.

Test file on the nfs server to see if it is OK.

See what files will be lost by q cont 'vol name' If they can easily be readded, then do that.

Open a pmr with IBM to see if they can aid you in reading more data from the affected files.

 

[nightcrawler]

The network is fine (judging y NFS statistics).
the damaged files are archive copies that no longer exist on client, they can't be re-archived.

unfortunately, we didn't extend the support contract for this server .

 

[Trident]

Hi,

Is it just one or two volumes that are lost, or is it more? If other people here are more skillfull about nfs server/client settings could say what they use, it may give you some setting to change.

https://www.ibm.com/support/pages/ibm-spectrum-protect-server-support-nfs

 

[nightcrawler]

HI Trident,

we are talking about two TSM file volumes with one archive withing each.

tsm: TSM>q cont 0002DBDC.BFS

Node Name           Type     Filespace      FSID     Client's Name for File
                             Name
---------------     ----     ----------     ----     --------------------------------------
***********         Arch     /sgbd/bac-        6     /PCARD/hc_back_and_exp_full/exp_full_-
                              kup                     pcard_.save


tsm: TSM>q cont 0002DEB2.BFS

Node Name           Type     Filespace      FSID     Client's Name for File
                             Name
---------------     ----     ----------     ----     --------------------------------------
******              Arch     /sgbd/bac-        1     /db***9i/cold_back_and_exp_full/exp_f-
                              kup                     ull_db***9i_2.save.Z

these two damaged archives are the only ones left in these file volumes.

I did a lot of testing before using FILE devclass with NFS (EMC Data Domain)
The mount options that worked well for me are:

vers=3,proto=tcp,auth=unix,hard,intr,dynamic,link,symlink,rsize=524288,wsize=524288,retrans=5,acregmin=30,acregmax=60,acdirmin=30,acdirmax=60

I didn't encounter any TSM NFS related issues since implementation (over a year now).

 

[marclant]

Redo the QUERY CONTENT, but with the option COPIED=YES. If these 2 archives still show up in your output, they can be restored from the copy pool.