Veritas-bu

[Veritas-bu] cannot write data to socket, Broken pipe

2005-07-01 11:43:24
Subject: [Veritas-bu] cannot write data to socket, Broken pipe
From: jpiszcz AT servervault DOT com (Piszcz, Justin)
Date: Fri, 1 Jul 2005 11:43:24 -0400
Whoever is in charge of the Auburn mailing list should add a FAQ point
about this.

Basically, questions that need to be asked and answered are as follows:

1) What version of GLIBC are you using?  Does it support > 2GB files?
2) Are you running 5.1MP3 client (run: update_Clients) on the Linux box?
3) Does the 5.1MP3 client backup > 2GB files in Linux.

--

On a side note, have you tried restoring the file to the Solaris master
server if everything is 64bit? Does it work?


-----Original Message-----
From: veritas-bu-admin AT mailman.eng.auburn DOT edu
[mailto:veritas-bu-admin AT mailman.eng.auburn DOT edu] On Behalf Of Wayne T
Smith
Sent: Friday, July 01, 2005 11:24 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] cannot write data to socket, Broken pipe

I'm having trouble restoring some unix files.  The backups go perfectly,

yet restores have status code 1, with one or many files not restored.  A

single file restore of one of the failing files fails to restore with 
status code 5.

Environment: NetBackup Enterprise Server is v5.1MP3 on Solaris 9.  
Clients in question are (various) Linux.  In each case, some of a file 
is restored and then we get a failure, perhaps 9-11 minutes after the 
restore "began reading".  The job is left in "Incomplete" status.  The 
client is left with a partial file, always exactly the same size.

Restoring one of these files to a different machine also fails, except 
that I have been able to restore to the client on my (single) Solaris 
master/media server in every case!   The file shown here is quite 
compressible ... the 10GB file is restored with NetBackup reporting 4GB 
sent (on Admin Console Activity Monitor).

I believe this started with v5.1MP3, but I'm not sure.  Has anyone else 
seen this?

Anyone have debug or workaround suggestions? 

Here's an example of the "job detail" on one of the failed restores.  
Although this one was of a 10GB file, other failures are on much smaller

files (power point, etc. ... not just the vm-ware file shown here):

06/30/05 12:24:12 - begin Restore operation
06/30/05 12:24:14 - 1 images required
06/30/05 12:24:14 - media NB0351 required
06/30/05 12:24:50 - connecting
06/30/05 12:24:52 - connected; connect time: 000:00:02
06/30/05 12:24:52 - started process bptm (11445)
06/30/05 12:24:52 - mounting NB0351
06/30/05 12:25:50 - mounted; mount time: 000:00:58
06/30/05 12:25:51 - positioning NB0351 to file 53
06/30/05 12:27:10 - positioned; position time: 000:01:19
06/30/05 12:27:10 - begin reading
06/30/05 12:38:19 - Error bptm(pid=11449) cannot write data to socket, 
Broken pipe
06/30/05 12:38:20 - Error bptm(pid=11449) The following files/folders 
were not restored:
06/30/05 12:38:20 - Error bptm(pid=11449) UTF - 
/vmfs/vmhba3:0:0:5/Snarl.vmdk
06/30/05 12:38:22 - Warning bprd(pid=11149) Restore must be resumed 
prior to first image expiration on Tue 26 Jul 2005 04:06
:21 PM EDT
06/30/05 12:38:22 - end Restore operation; operation time: 000:14:10

Thanks and cheers, wayne
_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu