samba - Jan 2005 - Best practices for long-running Samba server

Hi all,

I have been having the weirdest network browsing problems on my network the 
past week.  Come to find out the wins.dat file got corrupted.  Removing that 
file fixed everything.  Because our systems typically have uptimes measured 
in months rather than days, and it is difficult for me to restart Samba 
except as a scheduled task (which I'm afraid to do because there are users 
who come in earlier than I do in the AM and users that work later than me in 
the PM, and I don't want them having to call me at home :D ), I need to know
what the best practice is for the .dat and .tdb files.  Is it normal for them 
to become corrupt over time?  Is there some best practice for how often to 
remove them and let them re-populate?  Anything else I should know for a 
Samba instance that typically runs for months on end without interruption?

Thanks in advance,
Misty

misty,

i, too, have experienced some browsing issues since upgrading to samba
3.0.7-2 on an FC1 server.  the server stays up in terms of months than days
(thanks samba team).

restarting smb on the wins server every day seems to have eased this problem
for me. for a baling wire fix, i set up a cron job to restart smb on my
samba wins server when my users are not on my system every day.  this seems
to have "fixed" my browsing issues.

even when i have restarted smb on a server with users on the system, the
effects on my users have been almost nil.  my setup is not very
sophisticated, so i do not know how it would effect your situation.

if you have users on the system all hours of the day, then this may not
help.

stuart

----- Original Message ----- 
From: "Misty Stanley-Jones" <misty@borkholder.com>
To: <samba@samba.org>
Sent: Tuesday, January 25, 2005 7:37 AM
Subject: [Samba] Best practices for long-running Samba server

> Hi all,
>
> I have been having the weirdest network browsing problems on my network
the
> past week.  Come to find out the wins.dat file got corrupted.  Removing
that
> file fixed everything.  Because our systems typically have uptimes
measured
> in months rather than days, and it is difficult for me to restart Samba
> except as a scheduled task (which I'm afraid to do because there are
users
> who come in earlier than I do in the AM and users that work later than me
in
> the PM, and I don't want them having to call me at home :D ), I need to
know
> what the best practice is for the .dat and .tdb files.  Is it normal for
them
> to become corrupt over time?  Is there some best practice for how often to
> remove them and let them re-populate?  Anything else I should know for a
> Samba instance that typically runs for months on end without interruption?
>
> Thanks in advance,
> Misty
> -- 
> To unsubscribe from this list go to the following URL and read the
> instructions:  https://lists.samba.org/mailman/listinfo/samba

The client running our first Samba3 PDC production server had an uptime of 115
days, then the uptime got spoiled by them moving to a new building.

I have all client side needs automated, so maybe since they do not use Windows
Explorer to rove the network we have the lack of issues?

I have seen the warning in the docs to back up those TDB files each time you
stop the daemons... but I never stop them! ;-) So, would be interested in
reading suggested best practices just in case
life gets interesting here at some point.

-- 
Michael Lueck
Lueck Data Systems

Remove the upper case letters NOSPAM to contact me directly.

On Jan 25, 2005, at 7:37 AM, Misty Stanley-Jones wrote:
> Because our systems typically have uptimes measured
> in months rather than days, and it is difficult for me to restart Samba
> except as a scheduled task (which I'm afraid to do because there are 
> users
> who come in earlier than I do in the AM and users that work later than 
> me in
> the PM, and I don't want them having to call me at home :D ),
I restart my samba/cups services at 4:00AM every morning, using the 
following really, really aggressive script.

#!/usr/bin/perl
system("/sbin/service smb stop");
sleep(3);
system("killall -9 smbd");
sleep(3);
system("killall -9 nmbd");
sleep(3);
system("/sbin/service cups stop");
sleep(3);
system("killall -9 cupsd");
sleep(3);
system("/sbin/service cups start");
sleep(15);
system("/sbin/service smb start");

Never had a problem with things not restarting OK, and I've had this 
scheduled nightly for months.  In my experience, samba and cups will 
crash at some point, so I'd rather do this in attempts to avoid middle 
of the day type outages, but not everyone has the luxury of little to 
no users active at 4:00AM, so YMMV.
--
David Schlenk
Operating Systems Analyst
Bethel University
Saint Paul, Minnesota
david-schlenk@bethel.edu