Networker

Re: [Networker] Permission denied SUMMARY

2003-01-22 14:53:02
Subject: Re: [Networker] Permission denied SUMMARY
From: Terrence Haupt <Terrence.Haupt AT ABBOTT DOT COM>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Wed, 22 Jan 2003 13:53:00 -0600
*** Original post ***

Ok folks...  I have a good one.

My environment

Sun E450 Legato server Running Solaris 8 and Networker 6.1.1 (with updated
nsrmmd binaries)

I have many clients and storage nodes 2 jukeboxes a STK 9740 and STK L700
all with DLT7000 drives.

The L700 has 6 of its drives shared using DDS

The shared nodes are 9 solaris 8 machines and 2 HPUX 11 machines. (all are
6.1.1 with updated binaries)

When labeling a tape I get the following error but the label completes
successfully...

Verifying label, error opening: Permission denied.

When I try to mount a tape manually i get the same error and the tape
ejects.

When legato try's to use the drive it gets the same error and then the
tape is stuck in the drive until i manually reach in and pull it out.  I
checked each drive on each storage node using mt -f status some of the
drives come back as  write protected or reserved. others do not...  i
manually went out and
tried to mt -f release each drive and still get these errors...  Has
anyone ran across this before?


Side notes.
In my st.conf i have set the no reserve bit I have not however set it in
the stdef.h (but that's used for building your own dev files right?)
I have also restarted networker on each of the affected storage nodes and
the server.

*** Solution ***

I really didn't get any concrete reason why this happens, or why it
affected some storagenodes and not others, but!  I did resolve the
problem...(sort of)  What I did was unload the st drivers.  i.e.

first I found which systems had drive problems...

# mt -f /dev/rmt/1cbn status
/dev/rmt/1cbn: write protected or reserved.

Then I shut down networker

# modinfo | grep tape
110 782e0000  10f05  33   1  st (SCSI tape Driver 1.205)
then
# modunload -i 110
then
# modload /kernel/drv/st

I restarted networker

I then ran the mt command again to see if it was still reserved.

# mt -f /dev/rmt/1cbn status
/dev/rmt/1cbn: no tape loaded or drive offline

VOILA!  Problem gone!  (until the next time...)

This worked for all but one of my storage nodes.  (which will be rebooted
soon, it complained that the driver was busy, could this be the culprit???
I may never know).

other things that were verified was the "do not reserve" bit was set on
all systems (st.conf, and it was on all the Solaris boxes), on HP-UX there is a 
kernel
parameter that affects the scsi reservation.  This parameter is
st_ats_enabled.
This value should be set to zero. The value was set to 1 it has since been 
changed to 0.  (I don't know if
this caused the problem or not, but one of the HP boxes crashed while a
backup was running and mysteriously this problem started the same night
hmmmmmm....)

The curious thing was that it only happens with my shared drives (DDS),
and it didn't affect all of the storagenodes...  for example drive 1 of
the shared pool was reserved by 4 of 7 storage nodes, drive 2 only 4 of 6,
and so on...  the networker server was not affected at all, and neither
were the 2 HPux storage nodes.



Terrence R. Haupt
UNIX Systems Administrator
Global Technology Solutions, Abbott Laboratories
Ph: 847-938-2375 Fax: 847-937-4160
Pager: 877-804-5379 (thaupt AT skytel DOT com)
terrence.haupt AT abbott DOT com





--
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>
  • Re: [Networker] Permission denied SUMMARY, Terrence Haupt <=