ADSM-L

Re: [ADSM-L] Message ANR0481W, ANS1809W, ANS1810E: Session lost

2007-04-13 08:10:45
Subject: Re: [ADSM-L] Message ANR0481W, ANS1809W, ANS1810E: Session lost
From: Richard Sims <rbs AT BU DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 13 Apr 2007 08:10:00 -0400
On Apr 13, 2007, at 5:11 AM, Gerd Becker wrote:

Hello *SM*ers,
since we changed our network interface on the tsm server (IBM P630,
AIX5.3)
to Gigabit Eth, we get very often the error ANR0481W, Session lost.
On the
Client Error Log we get the message ANS1809W, ANS1810E very often.
On the
server side the option for commtimeout is 3600 and idletimeout is 60.
Has anyone an idea what the problem my be? Can I increase
commtimeout to
7200 and idletimeout to 120?

Some further thoughts on this...
We don't know what networking you had before, but probably 100 Mbps:
Did the introduction of gige substantially improve the elapsed time
of your backups - or are they actually taking longer now?  Taking
longer definitely points to networking problems.  Also, where backups
take longer, the RESOURceutilization value which is in effect may
result in Producer sessions timing out while Consumer sessions
struggle over a longer period of time to pump data through the
problematic pipe.  The termination of a Producer session is not
inherently bad, as the TSM client will start one afresh (ANS1810E)
when the Consumer session has concluded its backup work, when session
summary statistics need to be conveyed to the TSM server for it to
log as ANE messages in the Activity Log.  If you perform Query
SEssion through the backups period and continue to see viable client
sessions continually feeding data to the server, then you have good
Consumer sessions going on; and you may also see companion Producer
sessions for that node, and possibly see them disappear over time, as
they idle timeout.  You can do 'Query Node ____ F=D' to inspect
session values and get a quick sense of where delays were in the last
backup session, or run a report on your TSM accounting records for
before the gige change and then after, to see if any dramatic wait
time changes.

   Richard Sims

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