socket write failed error on tape backup only

[AAK]

I am getting socket write failed error only when backing up to tape. My enviorment is Master(win2008R2) Media(Appliance 5320) using tape Tape (IBM TS3200) Catalog backup to tape and a client backup to tape is giving similar errors. I collected the logs but they are not much help. Critical bpbrm(pid=98570) from client hidbak1.xenel.com: FTL - socket write failed 3/8/2016 7:57:19 AM - Error bptm(pid=98670) system call failed - Connection reset by peer (at child.c.1306) 3/8/2016 7:57:19 AM - Info dbclient(pid=5184) done. status: 24 3/8/2016 7:57:19 AM - Error bptm(pid=98670) unable to perform read from client socket, connection may have been broken 3/8/2016 7:57:20 AM - Error bptm(pid=98580) media manager terminated by parent process 3/8/2016 7:57:34 AM - end writing; write time: 0:51:25 3/8/2016 7:57:44 AM - Error bpbrm(pid=98570) could not send server status message 3/8/2016 7:57:46 AM - Info dbclient(pid=5184) done. status: 24: socket write failed socket write failed(24) Any guru out there to help me? Thanks.

[vjuhola]

Is the tape drive connected through SAN or is it directly attached to the appliance (afaik that is not even supported)? If it is connected through SAN, have you taken a look at your SAN for clitches or errors there? Have you cleaned fibre optic connectors? What are timeout values? What have you done to investigate the issue?

[nbutech]

Please check the data buffer settings

https://www.veritas.com/support/en_US/article.000074618

[tunix2k]

Default values for data buffers should be ok on an appliance.

What about the clients ?

- which OS

- which NBU Client version

- net buffer size ?

error(s) 24 are beasty and may have different root causes. Is this really only happen for tape backup ?

Since you are on appliance 2.6 please check: http://www.veritas.com/docs/000023273

If it not helps please provide log from client: bpbkar and bpcd and from media: bpbrm and bptm

ciao

Martin

[sdo]

@AAK, status?

[Mark_Solutions]

Please show us more of that job rather than just the failure

It may just be a simple timeout but we cannot tell without seeing the whole of the details

[sclind]

99% of our Windows backups started having status 24 errors late last week (I have an open case with Datalink).  So I am curious about your post.  Are just a few jobs failing or lots of them?  Any known changes to your network that could have caused this?

We are perplexed cause  it seems our problem just popped up out of the blue and so far we can't trace it back to anything.

We are Netbackup 7.6.1.2 on Windows 2008R2 writing to LTO4's on a Storagetek L700e silo.

Scott

[Marianne]

More /etc/sysctl.conf settings on the Appliance that can be checked in this post : https://www.veritas.com/community/forums/netbackups-failing-error-code-24-linux-clients

[AAK · Accepted solution]

Hi all,

Sorry for not posting more info, but my problem has been fixed.

I did the following:

1. Cleaned fibre

2. Switched fibre ports

3. Install latest drivers for tape.

I think the last one did the trick. Now compared to earlier, my tape backups are lightning fast.

[Marianne]

How did you install tape drivers on NBU Appliance?