Was this something that was working previously?

Has DNS changed in your environment?

When does the error occur?  Upon start-up of ltid or when the Master
server attempts to display status of the media server's devices?

-----Original Message-----
From: Ocalagan, Edmundo [mailto:eocalagan AT concordefs DOT com] 
Sent: Wednesday, July 16, 2003 9:29 AM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] ltid dies upon sockets errors

All,

Currently we are running Netbackup 4.5 mp2. Our problem is Sporadic
failure
of the ltid deamon on Media servers. Here are the errors we get from the
logs:

Edmundo, here are the errors being seen in the daemon.log. If you look
in
the log the errors happen many many times. 

++++
07:47:51.175 [17370] <16> put_string: (9) network write() error: Broken
pipe
(32)
07:47:51.176 [17370] <16> send_string: unable to send data to socket:
Broken pipe (32), stat=-5

08:01:52.075 [18487] <16> get_string: (7) network read() error:
Connection
reset by peer (131)
08:01:52.075 [18487] <16> read_string: unable to get record string from
socket: Connection reset by peer (131), stat=-3
08:01:55.634 [16917] <2> oprd: cannot open lockfile, assume ltid not
active
++++

Basically, the machine was not able to send or recieve to the ltid
process
over the network. Ultimately ltid went down because of the failure. I
will
contact you once I hear more from the Veritas support tech.

Does anyone seem this behavior on their media servers??. IF you have do
you
have a fix. Veritas support stated to be related to Network issues. We
are
running Solaris 2.6, 7 and 8 on different platforms, including the
Starfires. Any help is appreciated.


Edmundo Ocalagan
Concord EFS
eocalagan AT concordefs DOT com

_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu