media list shows 39 phantom entries 1

[itgetsmybackup]

I have a problem with media in my environment. I am receiving lots of errors of the form "Media Manager could not deassign media id , retaining it in netbackup database" followed by "Media Manager error 8, invalid media id, host = <srv name>" and the media list report for the affected server shows 39 'phantom' entries with no media id, and very few other attributes, as if I have a corruption in the media db. As ever, the veritas site seems to offer nothing, so I am looking for realworld assistance !

 

[Stumpr]

If NetBackup was trying to de-assign the media then all images on the media have reached expiration. NetBackup attempted to reach the media server to de-assign the media because all images were expired but it was unable to connect to the database on the media server for that particular media. Please review the technote:
How to manually remove a tape completely from NetBackup/Media Manager.

http://seer.support.veritas.com/docs/208420.htm

I would do some testing by creating a new volume pool and running backup/restores and media list report on this server.

 

[itgetsmybackup]

I don't believe I have a problem connecting to this part of the media db on the media server in question, since the media list report also shows the 'real' media that I would expect to be present, (and that also shows in the corresponding media pools for that server). The crux of the matter is that NONE of these 39 apparent volumes (which are NOT shown in any way through media and device management) have a media ID (!) and so I cannot use commands to manipulate them. I believe they are corrupt or gash entries as they show very few attributes, and those that do have values are usually zero or nonsense (e.g. they are listed as qscsi media which we don't have and have never had)...

 

[PGPhantom]

An error that too many people make - Expiring a tape manually without deassigning it or vice-versa. Another error, people expire servers, replace them etc but do not transfer the media database over and as such invalid entries are created. At one point I had over 1100 invalid tapes because of a combination of these errors.

Run the following:
bin/vmdb_compact /?
This will give you the following instructions:
How to use this database compaction utility:
1. stop the NetBackup Volume Manager Service
2. rename <Volmgr_Folder>\database\volDB as
<Volmgr_Folder>\database\volDB.old
3. start the NetBackup Volume Manager Service
4. execute vmdb_compact (reads old DB and inserts into new)

I have successfully recovered quite a few databases this way. One note - Run this on the media server that is reporting the error as it is specific to media servers and not to the master server, unless they are one and the same. Let me know if it works.

 

[itgetsmybackup]

My volDB file is only 8 kbytes, and when I follow the process you describe it basically tells me there are no entries. A comparision with another similar server (but with no similar problem) on another site reveals a volDB of 1k ? Am I missing something ?!

 

[PGPhantom]

My apologies - This is run on the master server - That should definitely be larger.

 

[itgetsmybackup]

Well, the master compact worked (276 entries) but the process did not, alas, fix my problem ! This activity obviously deals with the volume manager db, but this looks OK if I run a vmquery (i.e I see no sign of any gash entries) and my erros says 'retaining in NETBACKUP database', so is there a similar set of tools to deal with netbackup\db ??