Hello.

I just noticed, that on one of my S9 systems, some filesystems were
all of a sudden unmounted. I've got no clue at all, about why that
happened - earlier today, these filesystems were still mounted.
I did not umount them and there's nobody else, who could've done
that. In syslog, there's nothing at all to be found. I also checked
with "showlogs" the logs of the ALOM and with "consolehistory" the
output on the console. There's nothing at all. The filesystems reside
on a StorEdge 3310. The SE3310 is up and well and not all the filesystems
from the SE3310 were unmounted. Just some.

How can I possibly find out, /why/ or maybe just /when/ those filesystems
got dismounted?

The only unusual thing I did was, that I had "lucreate" running and
then killed the terminal in which it was running. This might have
killed lucreate. Could this maybe unmount those filesystems (they
were busy, BTW. Oracle is running on one of the filesystems and
the other is exported via NFS and also accesseed)?

Best regards,

Alexander

On Jan 3, 6:38 am, Alexander Skwar  wrote:

> I just noticed, that on one of my S9 systems, some filesystems were
> all of a sudden unmounted. I've got no clue at all, about why that
> happened - earlier today, these filesystems were still mounted.
> I did not umount them and there's nobody else, who could've done
> that. In syslog, there's nothing at all to be found. I also checked
> with "showlogs" the logs of the ALOM and with "consolehistory" the
> output on the console. There's nothing at all. The filesystems reside
> on a StorEdge 3310. The SE3310 is up and well and not all the filesystems
> from the SE3310 were unmounted. Just some.

Could be disk failure, if both file systems where on the same disk.
The correct log to check for that is /var/adm/messages.

> How can I possibly find out, /why/ or maybe just /when/ those filesystems
> got dismounted?
>
> The only unusual thing I did was, that I had "lucreate" running and
> then killed the terminal in which it was running. This might have
> killed lucreate. Could this maybe unmount those filesystems (they
> were busy, BTW. Oracle is running on one of the filesystems and
> the other is exported via NFS and also accesseed)?

Was that lucreate on the same disk?  I don't know if this could cause
the problem you are seeing.
--
Ren=E9 Berber

· René Berber <rberber@mailandnews.com>:

> On Jan 3, 6:38 am, Alexander Skwar  wrote:
> 
>> I just noticed, that on one of my S9 systems, some filesystems were
>> all of a sudden unmounted. I've got no clue at all, about why that
>> happened - earlier today, these filesystems were still mounted.
>> I did not umount them and there's nobody else, who could've done
>> that. In syslog, there's nothing at all to be found. I also checked
>> with "showlogs" the logs of the ALOM and with "consolehistory" the
>> output on the console. There's nothing at all. The filesystems reside
>> on a StorEdge 3310. The SE3310 is up and well and not all the filesystems
>> from the SE3310 were unmounted. Just some.
> 
> Could be disk failure, if both file systems where on the same disk.

Hm, but why was a 3rd file system not unmounted, which
was on the same "disk" (disk as far as Solaris is concerned,
ie. slices 1 and 2 were unmounted, but slice 3 was not umounted
(I don't know the exact slice numbers right now, but that doesn't
matter, does it?))?

> The correct log to check for that is /var/adm/messages.

Yes, in the syslog, there's nothing.

>> How can I possibly find out, /why/ or maybe just /when/ those filesystems
>> got dismounted?
>>
>> The only unusual thing I did was, that I had "lucreate" running and
>> then killed the terminal in which it was running. This might have
>> killed lucreate. Could this maybe unmount those filesystems (they
>> were busy, BTW. Oracle is running on one of the filesystems and
>> the other is exported via NFS and also accesseed)?
> 
> Was that lucreate on the same disk?

I don't quite understand what you mean with that. I had lucreate use a
metadevice d120, which was *NOT* located on the se3310. On se33310, I
have "non essential" filesystems /u01, /u02 and /u03. They were to be
shared by the old and new bootenvironments.

> I don't know if this could cause 
> the problem you are seeing.

Me neither. Does anyone have an idea about what Solaris all of
a sudden úmounts filesystems without leaving a trace at all in
the logs?

Alexander Skwar
-- 
Do not post without providing any useful information whatsoever. I
would be willing to try to guess if there was SOMETHING. You gave us
nothing to go on.
     -- /dev/rob0