We recently had some mongo problems with advfs corruption to the point that
any attempt to mount anything in a certain file domain (which just happened
to be the home directories....) would panic the machine. Luckily, backups
were fairly recent and no user has threatened my life yet. The file domain
was easily recreated, I just installed some patches to help alleviate this
problem, so life is good [note: anyone ever have NSR wait for a new tape over
a period of time where you delete a clone set <not backed up> and reclone it
<from a fileset that's currently being backed up but waiting for a tape>, AND
have root logged into the console during all this? that's the "state
information" for this advfs corruption].
However, there still seems to be a bit of fallout, with the Prestoserve
acceleration:
state = ERROR, size = 0xffa00 bytes
statistics interval: 00:00:00 (0 seconds)
write cache efficiency: 0%
All batteries are ok
/usr/sbin/presto: PRSETSTATE DOWN failed: Device busy
Prestoserve acceleration has been disabled.
Starting Prestoserve: presto -u for the following:
/usr/sbin/presto: PRSETSTATE UP failed: Device busy
/usr/sbin/presto: PRENABLE failed /dev/rz3a: Device busy
/ - Presto enabled
/usr/sbin/presto: PRSETSTATE UP failed: Device busy
/usr/sbin/presto: PRENABLE failed /dev/rz3g: Device busy
/usr - Presto enabled
/usr/sbin/presto: PRSETSTATE UP failed: Device busy
/usr/sbin/presto: PRENABLE failed /dev/vol/homedir: Device busy
/home/jekyll - Presto enabled
/usr/sbin/presto: PRSETSTATE UP failed: Device busy
/usr/sbin/presto: PRENABLE failed /dev/vol/Female: Device busy
/usr/sbin/presto: PRENABLE failed /dev/vol/mailspool: Device busy
/usr/var/spool/mail - Presto enabled
Presto has been enabled.
Presto is *not* enabled for any of this, from the presto -l command. What
exactly does this "device busy" mean?
(/dev/vol/Female and /dev/vol/mailspool are LSM volumes used by the spool
file domain with a 'mail' fileset; homedir is obviously the home directories,
and rz3a/g are / & /usr)
Thanks for any info. I'm kinda worried about this; it looks like there is
stuff in the Presto cache that's waiting to be written out, but maybe
shouldn't be if it was there before the corruption caused the crash that
caused the homedir volume to be reloaded.
---
Mike Matthews, Mike_Matthews_at_sgate.com (NeXTmail accepted)
PGP public key available on request
Received on Wed Jul 17 1996 - 22:32:15 NZST