Veritas-bu

[Veritas-bu] RE: Linux restore problems (Veritas-bu digest, Vol 1 #4220 )

2005-08-03 05:31:53
Subject: [Veritas-bu] RE: Linux restore problems (Veritas-bu digest, Vol 1 #4220 )
From: Mark.Hickey AT hds DOT com (Mark Hickey)
Date: Wed, 3 Aug 2005 02:31:53 -0700
 
Wayne,
   Am I right in thinking that you are serving these files (the .ppt
extention gives me this impression) via SAMBA to windows hosts?  You may be
seeing a symptom of a problem downstream of your restore process; e.g.:
filesystem is full or read-only, a version of the file is in use,  etc.  The
socket write error often means that the landing spot for the file cannot be
accessed.  Have you tried an alternate path restore of one of these files to
see if the target destination is the issue?

Mark

--__--__--

Message: 3
Date: Tue, 02 Aug 2005 16:14:40 -0400
From: Wayne T Smith <wtsmith AT maine DOT edu>
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] Linux restore problems ...

My Solaris NetBackup server is v5.1MP3 and I have a problem with some 
Linux restores.  On a few machines a full restore results in some 
(many!) files not being restored.  Some files are restored; others 
aren't.  If I pick one of the "problem" files and restore just it, I get 
a status code 5, with a "Socket write failed" error, 9-11 minutes into 
the restore.  An example Admin Console log is listed below. Failures are 
always 9-11 minutes into the restore, with the below example almost 
exactly 10 minutes.

A restore to another Linux machine (version & MP do not seem to matter) 
results in the same error.  However, a restore to my backup server 
succeeds!   For what it is worth, the data was backed up with NetBackup 
v4.5MP5.

I could use suggestions for how to debug this problem (socket write 
failed is not very informative) and any suggestions on settings to 
tweak, should that be the problem.

(My master/media server has the following timeouts reported: Client 
connect timeout: 900 seconds, Client read timeout: 1800 seconds, Media 
server connect timeout: 60 seconds, Use OS dependent timeouts is 
checked.  On the client, the Client read timeout is 300 seconds).

Thanks and cheers, wayne


ash:~$ less /usr/openv/netbackup/logs/user_ops/wsmith/logs/jbpUJAmua4gi.log
Restore started 08/02/2005 14:45:49

14:45:57 (248339.xxx) Restore job id 248339 will require 1 image.
14:45:57 (248339.xxx) Media id NB0361 is needed for the restore.
14:45:57 (248339.xxx) Media id NB0263 is needed for the restore.
14:45:57 (248339.xxx) Media id NB0145 is needed for the restore.

14:46:02 (248339.001) Restoring from image created Sat 30 Jul 2005 
12:19:57 AM EDT
14:46:10 (248339.001) INF - Waiting for mount of media id NB0263 on 
server ash for reading.
14:53:56 (248339.001) INF - Waiting for positioning of media id NB0263 
on server ash for reading.
14:54:31 (248339.001) INF - Beginning restore from server ash to client 
cedar.unet.maine.edu.
15:04:37 (248339.001) The following files/folders were not restored:
15:04:38 (248339.001) UTF - 
/usr/local/blackboard/content/vi/bb_bb60/courses/1/00230_K_042/groups/_9839_
1/_276333_1/Presenta
tion.ppt
15:04:39 (248339.001) Status of restore from image created Sat 30 Jul 
2005 12:19:57 AM EDT = socket write failed

15:04:40 (248339.xxx) INF - Status = the restore failed to recover the 
requested files.


<Prev in Thread] Current Thread [Next in Thread>