Edy Sulai wrote:

>I have very similar problem with samba 3.0.20b running on Redhat 9.
>in my case, smb process stacks up (within reasonable amount) and freeze.
>I had to restart smb to make it works again.
>the machine i'm using for samba pdc is quite low end (256 mb memory).
>
>through a few tests on the machine, i believe the problem was caused by
>lack of memory.
>

We have 1Gb of system memory here (512Mb of swap), and I think that is
plenty for a samba server. Weather your low memory was the cause in
your case or not you are on the right track to want to look into
upgrading. 256Mb is pretty low for serving up files and doing
authentication for 100 users.

In my case the process is stacking up out of control and for a reason I
cannot understand yet. Usually when this has happened there were only 7
users still connected and everyone else had gone home for the night. The
latest step I have tried was to set my deadtime to 15. It was set to 0
by default. Weather or not this corrects the runaway smbd processes
from happening again is still to be determined, but it should help with
clients that leave smbd processes open and do not close them.

>free memory was about a little over 15 mb. swap space
>was used up entirely.
>i'm replacing the server soon.
>
>btw, is there any minimum hardware spec for a samba PDC ? how much
>memory would samba requires for about 100 clients?
>
>thanks
>
>
>Edy
>
>On Wed, 2005-12-07 at 22:11, Simo Sorce wrote:
>
>
>>On Wed, 2005-12-07 at 09:50 -0500, Matt Lung wrote:
>>
>>
>>>To get to the point of the problem, this server will run fine for a
>>>period of time and then begin to build up SMBD processes until
>>>eventually our users can no longer access shares. The Samba server just
>>>stops responding. It does not even respond to STOP, START, or RESTART
>>>commands. Doing a RESTART on samba will look like it is restarting the
>>>service, but Samba will still be in the same locked state with shares
>>>still not available. Doing a status on the service then reveals that
>>>the STOP, START, or RESTART did nothing to clear out the old processes
>>>or the locked files it thought it previously had opened. We end up just
>>>rebooting the server to clear everything out. Right now we are just
>>>reading through all the documentation, posts, and waiting for this to
>>>happen again to hopefully capture some error in the log. When that
>>>happens I'll send more detail.
>>>
>>>

>>Instead of immediately restarting it you may attach a strace to the
>>spinning process and tell us where it dies.
>>
>>Meanwhile I suggest you to check the integrity of your tdb files
>>(killing with -9 may lead to corrupted tdbs and in some rare occasion
>>I've seen our code spinning on corrupted files).
>>
>>To check if a tdb is ok, you can tdbbackup it (no need to stop samba
>>for that) and see if the backup is ok. In case of error you have a
>>corrupted tdb and it is better to remove and restart it in case it is a
>>temporary db or plan adequate measures in case it is a persistent one.
>>
>>Simo.
>>
>>--
>>Simo Sorce - idra@samba.org
>>Samba Team - http://www.samba.org
>>Italian Site - http://samba.xsec.it
>>
>>

>
>
>


--
Matt Lung

Midwest Tool & Die, Corp.


--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/listinfo/samba