label verification failed

Question

hello, i recently upgraded NetWorker server from 7.3.4 to 7.5.1. i have configured separate media pool for index backups. after upgrade i have noticed that from time to time NetWorker reports errors during tape mount for that index pool. it tries to load tape and then reports 'label verification error'. then it re-tries it several times until finally succeeds. sometimes it tries to load, fails, unloads and fails again about 10 times before finally succeeds. tape itself is ok as it finally manages to verify its label and event write to it. re-labeling tapes does not help, it happens with 'brand new' tapes as well. we are using EMC Disk Library, so these tapes are actually virtual. i will also try to add screenshot were you can see multiple tries of loading tape EDLA0400 and finally successful load. it usually happens with index pool tapes, but i have seen it for other pools as well. do you have any ideas? regards, kaido

kipz1 · Answer

damn, seems that attachments are not allowed anymore...

kipz1 · Answer

i have also noticed that at the same time when there are those "label verification failed" errors under NMC -> Monitoring -> Operations (and also in daemon log) there are following messages under NMC -> Monitoring -> Log:

(volume id ) not found in media index

these messages do not appear to daemon log. so i'm little bit suspicios that these errors are not actually related to any SCSI errors but are somehow related to media, index or config databases.

have anybody seen this before? these volumes which are reported as missing from media index are actually present in media database and can even be browsed for save sets.

kipz

dugans1 · Answer

Yes i have had the same thing. I will check out my logs and see if they contain the same entries as you are stating.

DavidHampson · Answer

I have seen similar issues in the past when tape ordering gets screwed up - are your tape devices persistently bound?

kipz1 · Answer

actually, yesterday i had binding issues. it is brand new environment, persistent binding was not implemented yet. NW is running on Solaris 10. day before yesterday this server got rebooted and tape drives were discovered in different order. i had to re-configure library. so it might be related to that indeed.

today i implemented persistent binding in /etc/devlink.tab file and re-configured all libraries again. will see whether it gets better or not.

thanks for a hint.

kipz

kipz1 · Answer

i will have more information tomorrow. today i'm doing upgrade to EDL code and cannot perform much tests now. int the evening new backup schedules start and by tomorrow morning i have more information whether this problem still exist or not.

i'll keep you posted.

kipz

dugans1 · Answer

Please do update this thread as i am watching it very closely.

I have had to disable persistent binding as i have done some major changes to our environment.

I am looking to re-enable this but waiting for things to settle down first.

Note, When i disabled it i did not see a performance impact on the system.

And yes i have had the problem of having to re-create the library. Don't re-boot

Message was edited by:
Scott Dugan

kipz1 · Answer

seems that persistent binding helped. i did not notice any label verification errors during last night backup cycle anymore.kipz

NetWorker

Was this post helpful?