We recently transferred our Samba server to a cluster of two Sun systems running Solaris 9 with Veritas Cluster Server. We used Samba 3.0.7. Since then Samba seems to crash every 2 to 3 days, always around noon. The symptoms are that at first people who are connected lose the ability to access the shared filesystems, followed a few minutes later by nobody being able to create new connections either. Failing over the Samba service from one box to the other clears the problem up for a couple days till it recurs. Simply stopping and restarting Samba on the server where it resides does not help.

We did not have this problem at all with the old Samba server, which was running 2.2.8 on Solaris 8 with no clustering (on a single box).

We have tried downgrading the Samba version but the problem persists. We are not sure if it is an issue with Solaris 9, or with Veritas Cluster Server, or what.

I didn't find anything on this site related to running Samba on a cluster of Solaris systems. I will continue searching other sites as well, but I thought I'd post this here. Has anybody seen an issue similar to what I describe where Sambe periodically quits working?

Sorry, no, I've never seen that set-up. There are some patches for Samba out on Sunsolve - you might look into them. Also, if this is a cluster, do you have support? You might want to call them first.

The only thing sunsolve had on it that I could find with Samba and a cluster was for SUN's cluster software.