ADSM-L

Antwort: Linux restore hangs in IdleW

2004-08-17 05:49:31
Subject: Antwort: Linux restore hangs in IdleW
From: Rainer Holzinger <Rainer.Holzinger AT UPM-KYMMENE DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 17 Aug 2004 11:32:30 +0200
Hi Alexander,

I had a similar problem.
Could you please try on your AIX system where the ITSM sever is running the
command 'no -a | grep sack'?
What's the setting for sack?

regards, Rainer



                      Alexander
                      Verkooijen               An:      ADSM-L AT VM.MARIST DOT 
EDU
                      <alexander@SARA.         Kopie:
                      NL>                      Thema:   Linux restore hangs in 
IdleW
                      Gesendet von:
                      "ADSM: Dist Stor
                      Manager"
                      <[email protected]
                      T.EDU>


                      17.08.2004 11:23
                      Bitte antworten
                      an "ADSM: Dist
                      Stor Manager"






Hello,

(Searched the list archives and mr. Sim's page, even read the manual)

We are seeing weird behaviour of the 5.2.3.1 client on
several Red Hat 7.3 systems.
(yes, we know 7.3 is very old but we can't upgrade for political reasons).

When we try to restore a file that resides on a tape
that is not mounted in a drive the server starts to mount the tape.
The restore session goes into MediaWait. The client reports
that it is waiting for a media mount. After a minute or so
the tape is mounted and the session goes into Run.
This lasts for about another minute. I guess the tape is being
positioned. Q ses doesn't show any data going to the client.
The the session goes to IdleW and stays that way.
The status of the tape is "IN USE" until after 13 minutes
the session dismounts the tape and immediately terminates
itself (connection severed). Our CommTimeOut is 15 minutes btw.
At this point the client still reports it is waiting for a
mediamount until. Of course no data has been restored.
We have seen client hang for hours like this until we killed
them manually.

Because of the old RedHat level we can't go to IBM for support
and we can't reproduce this problem on other systems with
more up-to-date OS levels.
And no, we really can't upgrade these systems.

This is clearly some kind of timing problem because when the tape is
in a drive (in IDLE state) before the restore is started or when the
data is on a disk volume instead of a tape the restore completes without
any problem.
It looks like the 60 second wait for the mount is causing this behaviour.

Has anybody seens this before?

(5.2.3.0 server on AIX 5.2, 3494 library with 3590E drives)

Thanks,

Alexander

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