Networker

Re: [Networker] more on strange 7.0 problems

2003-07-23 21:58:07
Subject: Re: [Networker] more on strange 7.0 problems
From: Matthew Temple <mht AT RESEARCH.DFCI.HARVARD DOT EDU>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Wed, 23 Jul 2003 21:48:02 -0400
Shelley,

        For what it's worth, I have also upgraded to 7.0 from 6.1.3 and
I've even taken a further irrevocable step and i'm running it on RH 9.0.
(I _had_ my reasons.)   Apart from the small damage I've caused myself
via invoking stinit, and apart from SCSI errors that preceeded the upgrade
as well,  my system has had no trouble appending to existing tapes and,
in addation, no trouble packing up those AIT-3 tapes, whew.   I don't use
NDMP.  Capacity remains high on the tapes as well.

                                        mht

> I've lamented on the list that I am having problems after upgrading to 7.0
> with tapes being marked prematurely full.  The tapes were *usually* being
> marked full when the backup started and tape loading/motion were involved.
> People suggested it was a jukebox sleep problem.  Well, I finally have found
> the time to look more closely at the problem, ie uninterrupted.  The reason
> that the tapes are being marked full is that the *previous* backup had
> problems.  I have a one to one correlation with an NDMP backup failure leading
> to a tape being marked full the next time networker attempts to load the tape.
>
> I have savesets with 0 files and 0 MB followed by "normal" backups.  Networker
> can continue to write to the tape during the same session.  I am guessing that
> networker is not using mtio commands to position the tape.  But, it "gets
> confused" about the number of records on the tape and when it does the mtio to
> fsf the tape to the end, it gets to the incorrect place and marks the tape
> full.  I think the tip off is that in 6.X the 0 length savesets are not shown
> in the Instances window and there is no indication in the logs that there were
> retries unless you carefully read daemon.log, which I had not done (slap).
> With NDMP backups, there is repeated instances of the NDMP resources being
> printed to the log while the backup is running and this makes it difficult to
> read the log daily.  OK, it's a pathetic excuse :-)
>
> I don't quite know how to confirm this, but maybe scanner with some debugging
> compared to the mminfo -V file number will confirm the problem.  I am thinking
> that I may have seen this same problem in 6.x but rarely and now with 7.X it
> is more of a problem.  Either that or for some reason my NetApp is suddenly
> more loaded and the backups are failing regularly.
>
> So, I think it is time to open a case with Legato to figure out why this is
> happening.
>
> If anyone is interested in hearing more, please contact me off list.
>
> Shelley
>
> --
>     Shelley L. Shostak, PhD                     sls AT qstech DOT com
>     Voice: (408) 574-3389
>     Lead Unix Administrator                     Quicksilver Technology
>
> --
> Note: To sign off this list, send a "signoff networker" command via email
> to listserv AT listmail.temple DOT edu or visit the list's Web site at
> http://listmail.temple.edu/archives/networker.html where you can
> also view and post messages to the list.
> =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
>

=============================================================
Matthew Temple                Tel:    617/632-2597
Director, Research Computing  Fax:    617/632-4012
Dana-Farber Cancer Institute  mht AT research.dfci.harvard DOT edu
44 Binney Street,  JF 314     http://research.dfci.harvard.edu
Boston, MA 02115              Choice is the Choice!

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

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