One file system always runs at full

Question

Possibly this problem is down to the fact that this stuff is running on antiques but here is the issue I have:

Networker server 6.1.4 on Solaris 8
Networker client 5.5.3 on Solaris ? (hope you didn't swallow your gum when you read that, Hrvoje!)

Client has several file systems which are backing up okay; however one backup of file system /u00 (flat file data - the Oracle data in this file system is being backed up seperately by RMAN) first of all fails with an "exit on signal 11" error but then retries and completes successfully.

However, this file system always backs up as full - the probe returns the last full backup as 1/1/1970:

dbinf2:/u00 level=full, dn=3, mx=1, vers=pools, p=4
dbinf2:/u00 level=full, pool=UXREMsystem3m, save as of Thu Jan 1 00:00:00 GMT+0000 1970

mminfo returns details of this file system backups:

UB0732 dbinf2 02/22/06 395 KB full /u00
UB0732 dbinf2 02/22/06 195 KB full /u00
UB0732 dbinf2 02/22/06 1207 MB full /u00

I have noticed that there is a lot of variance on the size of data being backed up but I assume this relates to whether the Oracle data files are locked or not.

If I check the indexes tab in the GUI it does not list any indexes for /u00.

The other file system backups run fine, do incrementals and list index data.

I have ran nsrck -L6 twice on this client's indexes and nsrim -X without any success. After this I will look at a restore of the client indexes and if nothing there a media scavenge.

ble1 · Answer

Networker client 5.5.3 on Solaris ? (hope you didn't
swallow your gum when you read that, Hrvoje!)

You should try 5.5.5 at least

Can you just confirm that index gets saved under correct client index file? Is this some kind of cluster or just plain local file system on dbinf2 box (check /etc/vfstab for anything strange). Do you see the same if you try to do save with "save -c dbinf2" in remote backup command?

DavidHampson · Answer

I'll recommend we upgrade the client software but it is likely that it takes several weeks for me to get approval to do this!

I've already checked /nsr/index and there is only the one directory there for this client, and it is not on a cluster, just a plain local file system.

I can try running the save locally but at present I don't have local client access.

DavidHampson · Answer

I have restored the index with nsrck -L7 but still the same situation. I'll take a look at getting local access to test further.

ble1 · Answer

Yes, you would really need access to get more data... However, to get save output in debug you could the same from the server itself: just put in backup command save -D9 and this will go on server to local messages files I think (while running it will be in /nsr/tmp so you could tail while running).. The question is only do you have enough space on backup server end as I have no idea how big output will be (but usually it is big when running full backup).

DavidHampson · Answer

A further update before I leave for the day! Have ran save -D9:

root@ubback# save -n -D9 -c dbinf2 /u00
save: lg_lstat(): Calling lstat64().
/u00: No such file or directory

Returns No such file or directory!?! Will get local access to the box tomorrow and see what I can find out.

NetWorker

One file system always runs at full

Was this post helpful?