zfs discuss - Feb 2011 - smbd becomes unresponsive on snv_151a

Hi!

We have some trouble with CIFS. Two servers (n20 and n30), both configured the
same way and running Solaris 11 Express snv_151a. From time to time CIFS service
on n20 locks up, becomes unresponsive and impossible to restart, even "kill
-9 <pid_of_smbd>" can''t kill it. Only system reboot helps to
solve the problem.

Servers had deduplication turned on ZFS datasets for some time and on n20 it was
only recently turned off and still contains deduped data. Servers periodically
create and rotate ZFS snapshots on shared filesystems. We cant get any definite
evidence on what is causing the trouble. Can somebody advise on how to debug
this? We can provide more information if necessary or even organize remote
access if there is interest in it.

Thanks!
-- 
This message posted from opensolaris.org

Are you running CIFS with any AD integration, or is it functioning in work-group
mode?

Do you have lockups only when you transfer a lot of data, or will it lock up
without any machine working on the CIFS share?

How long is "time to time"?

-----Original Message-----
From: zfs-discuss-bounces at opensolaris.org [mailto:zfs-discuss-bounces at
opensolaris.org] On Behalf Of Marcis Lielturks
Sent: Thursday, February 10, 2011 10:57 AM
To: zfs-discuss at opensolaris.org
Subject: [zfs-discuss] smbd becomes unresponsive on snv_151a

Hi!

We have some trouble with CIFS. Two servers (n20 and n30), both configured the
same way and running Solaris 11 Express snv_151a. From time to time CIFS service
on n20 locks up, becomes unresponsive and impossible to restart, even "kill
-9 <pid_of_smbd>" can''t kill it. Only system reboot helps to
solve the problem.

Servers had deduplication turned on ZFS datasets for some time and on n20 it was
only recently turned off and still contains deduped data. Servers periodically
create and rotate ZFS snapshots on shared filesystems. We cant get any definite
evidence on what is causing the trouble. Can somebody advise on how to debug
this? We can provide more information if necessary or even organize remote
access if there is interest in it.

Thanks!
-- 
This message posted from opensolaris.org
_______________________________________________
zfs-discuss mailing list
zfs-discuss at opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

It is integrated with AD who has a lot of domain trusts, lockups happen during
working hours, when CIFS is quite busy with a lot of users (~216) and files
open(~1723). I don''t know about IO load at time when lockup happens,
but usually it is not doing heavy IO. "time to time" is varying -
first lockup happened ~13 days after OS was installed, then it worked for
another 16 and hanged again. Third lockup (last so far) happened 5 days after
2nd lockup.

cifs-discuss suggested that this might be "6996574 smbd intermittently
hangs" which seems very likely.

Thanks!
-- 
This message posted from opensolaris.org

> Servers had deduplication turned on ZFS datasets for some time and on
> n20 it was only recently turned off and still contains deduped data.
> Servers periodically create and rotate ZFS snapshots on shared
> filesystems. We cant get any definite evidence on what is causing the
> trouble. Can somebody advise on how to debug this? We can provide more
> information if necessary or even organize remote access if there is
> interest in it.
How large is the dedeuped dataset? And how much memory/L2ARC do you have in the
system?

Vennlige hilsener / Best regards

roy
--
Roy Sigurd Karlsbakk
(+47) 97542685
roy at karlsbakk.net
http://blogg.karlsbakk.net/
--
I all pedagogikk er det essensielt at pensum presenteres intelligibelt. Det er
et element?rt imperativ for alle pedagoger ? unng? eksessiv anvendelse av
idiomer med fremmed opprinnelse. I de fleste tilfeller eksisterer adekvate og
relevante synonymer p? norsk.

Deduped dataset is 2.1TB, no L2ARC and server has 64GB RAM. We have currently
ruled out possibility that this is related to dedup and ZFS and working to get
fix for "6996574 smbd intermittently hangs".

Thanks!
-- 
This message posted from opensolaris.org