SUMMARY: Netbackup error: socket write failed

From: Rob Thompson <>
Date: Tue Jun 14 2005 - 13:12:17 EDT
Solution: Had to contact Sun support to get this one figured out.
After working things through, the autonegotiation settings on the
switch turned out to the be the problem.  With autonegotiation set to
OFF on the switch (forced to 100MB/full), the backup completed


------Original Message---------

Hello Sunmanagers,

I am having a problem that has me stumped with Netbackup.  I get the
following error on one of my machines both with backups initiated by
myself from the client or scheduled backups.  Any insight would be
greatly appreciated, I of course will summarize.

(from file)
11:28:41 ERR - Cannot write to STDOUT. Errno = 32: Broken pipe
11:28:45 INF - Backup by root on client blazer: socket write failed.

1) I installed the most recent patches etc. to the netbackup client
and the OS recommended patches (Solaris 9).  Here is my NetBackup
version: NetBackup-Solaris9 5.0MP5

1) The backup always fails with "socket write failed".

2) The backup always fails after getting to the same point in the
file system and logging the same file "cwsearn-03w.doc", however the
backup is hanging on the next file (learned this from lsof)

3) The backup hangs for approx 10 minutes on the file before outputing
the above error message.

4) There seems to be no relation to the size of the backup job, I
started a very small backup that only grabbed the few files before and
after this file, and it still failed on this file... I ran a large
backup and got all the way to this file, then failed.

5) I can select a single large tar file of the entire machine to
backup and the backup completes successfully, but a regular full or
incremental backup of the filesystems always fails on the above file.

6) There is nothing different wrt/ permissions on this file.

7) I ran FSCK on the filesystem and there are no problems.

8) There seems to be no problems with the network. I
routinely do rsync backups between buildings with no problems:

9) I ran a snoop of the network traffic during the failure and the only
interesting line in it is here:

loghost -> netbackup1 RPC R (#0) XID=0 Vers mismatch (low=1175552, high=93)
loghost -> netbackup1 RPC R XID=1175556 unknown
loghost -> netbackup1 RPC R XID=1175556 unknown

10) The backup log for the jobs show the socket write failure above.

11) The bpcd logs in /usr/openv/netbackup/logs/bpcd/... dont seem to
show anything interesting, at least not any errors.

Here is my bp.conf:

SERVER = netbackup1
CLIENT_NAME = blazer


Wayne State University
sunmanagers mailing list
Received on Tue Jun 14 13:12:51 2005

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:43:48 EST