cancel
Showing results for 
Search instead for 
Did you mean: 

Lots of socket write failed on 8.3.0.1

pnobels
Level 3

Hi,

setup : 3 x media servers running Windows 2012 R2, backups going to disk and then stage to LTO7.

Since we've upgraded from 8.1.2 to 8.3.0.1 we've experienced a lot of full backup failures during the weekend.

The errors can be summarized as : 

(2074) Disk volume is down
(41) network connection timed out
(24) socket write failed

All seems to point to network timeouts.  But we've never had issues with 8.1.2.  This is the second weekend that we're taking backups with 8.3.0.1 and that these failures occur.  Some clients were still on 8.1.2 last week.  They are on 8.3.0.1 now, but this does apparantly not make any difference.

Anyone had similar experiences?

 

8 REPLIES 8

Alexis_Jeldrez
Level 6
Partner    VIP    Accredited Certified

The Master does not serve the role of Media Server?

Yes, the master is also media server.

Been doing a little bit of further investigation...  Turns out i was wrong and the issue does not appear on all media servers.  All above mentioned errors point to one media server, which happens to be the master.

There were some other errors on other media servers (snapshot error while reading file) but i've also seen these in the past and am not too worried about these ones.

No network bandwidth exhaustion during backup window.  Nothing strange in the event logs...

Alexis_Jeldrez
Level 6
Partner    VIP    Accredited Certified

Only for Windows Servers: https://www.veritas.com/support/en_US/article.100037977

"DO NOT configure MSDP on NetBackup (NBU) master servers..."

I learned it the hard way: MSDP generally worked fine but at random the volume went down with status 2074.

My workaround: reduce the number of duplications from disk to tape. For example, if backups to disk are done during the night, then the duplications to tape are done during the day. The real fix is migrating the MSDP resource to another Windows Server that is not a NetBackup Master Server.

It may be unrelated, but I experience 2074s on OST when a media server is unable to query the storage the disk pools are attached to. The problem is, the storage unit is reported as down for all media servers until the culprit re-establishes connectivity to the storage servers. Could be food for thought though.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

@pnobels 

Are you aware of the new sizing guidance for NBU 8.3 and later?

Please see the recommended minimum processors and memory:
 Size guidance for the NetBackup master server and domain

https://www.veritas.com/content/support/en_US/doc/21414900-146141073-0/v146051743-146141073

 

 

We do not use MSDP.

I've seen the requirements.  Master server only has 64 GByte of RAM in our case.  But we only have 3 media servers.  And i haven't seen memory or cpu shortage.

Tape_Archived
Moderator
Moderator
   VIP   

@pnobels please check usage of your sockets on the Windows Master server. Use netstat utility to figure out if there are more than usual or normal usage of sockets from or to the master. You may have to work with your Windows admin to find more on this.