[Samba] SAMBA bringing NFS server to a halt

[Andreas Gaiser]

Hey Metthew,


I think you might have a chance.

Did you try clearing binary data stored by samba? There's a whole bunch
of stuff in tdb files in potentially more than one place (on Debian it's
under /var/lib/samba). Parts are temporary data and safe to remove.
Thereof cached data, connection data and I think file locking
information too. Some other files contain credentials and perhaps ID
mapping data you better preserve (was it secrets.tdb?). In my
experience, things can become corrupted and samba fails in some way. You
can always backup that stuff and try after 6 PM.


Andreas

On 06.03.13 12:33, Joseph, Matthew (EXP) wrote:
> Hello,
> 
> We have a Red Hat 5.3 SAMBA 3.0.33-3.7 Server that shares a few directories to 4 other servers.
> The other servers are Red Hat 5.3 and one Solaris 10 server.
> 
> I configured SAMBA to do the following for each share;
> 
> Force User: User1
> Force Group: Group1
> 
> Create Mask: 02770
> Security Mask: 02770
> Directory Mask: 02770
> Directory Security Mask: 02770
> 
> Inherit Permissions: Yes
> Inherit ACLS: Yes
> Inherit Owner: Yes
> Guest Okay: Yes
> 
> When the other servers mount the SAMBA shares they work fine until someone starts using SVN or Eclipse.
> This brings the SAMBA server to basically a halt. Looking at the processes I see about 15000 instances of SMB running. I try running top to see a list of processes but it takes about 10 minutes for it to start and then it will hang when it tries to do its first refresh.
> 
> Looking at the logs I don't see anything that really stands out on why it is slowing down.
> 
> Is there something I'm doing wrong in this configuration?
> 
> Thanks.


-- 
Andreas Gaiser
Berlin