[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[cobalt-users] raq3 crashing [samba and netatalk]



I'm having a lot of difficulty with samba and netatalk on a mixed network of
win2k, winNT, win98, win95 and Apple Macs.

The raq was reinstalled on the weekend and before anything was changed it
had all the recommended patches laid down.  Since then the only changes have
been compiling php4 and some associated extensions, setting up a few sites
and linking smb and atalk into rc3.d

There are two main issues. The first is the nmbd keeps dieing
the logs are flooded with messages like

[2000/07/24 14:49:41, 0]
nmbd/nmbd_packets.c:find_subnet_for_nmb_packet(1412)
  find_subnet_for_nmb_packet: response record not found for response id
10147
[2000/07/24 14:49:41, 0]
nmbd/nmbd_responserecordsdb.c:find_response_record(240)
  find_response_record: response packet id 10147 received with no matching
record.

At peak logging times these get up around 2000 (2000 of each entry) times a
minute.  When the nmbd dies it still stays resident, just unresponsive.
This means that the raq keeps disappearing from the windows network.

The second is the apple talk daemon.  I'm getting an absolute flood of
entries in /var/log/messages to the tune of around 6,000 syslog messages a
minute at peak times and 67,862 entries in the last 6hrs (6.2mb) also the
syslog daemon has had over 50 minutes of cpu time in the last 17hrs.  The
messages I'm getting are of the form:

Jul 26 04:05:24 lightspeed atalkd[616]: zip skip reply 47999-47999 from
41166.140 (no query)

I'm not sure whether it's one or both of these daemons combined which are
crashing the raq (no traffic indicator lights flashing and the machine can't
be pinged, much less telneted to + the controls on the front do nothing),
however last night I shutdown the samba services and the machine was still
running this morning, albeit very badly (even simple telnet is
unresponsive).  Looking at free memory I have about 85mb of memory being
used however I can really only account for about 25 of that being used by
running processes I realise that a lot of it could just be dirty but if I
don't have either of these processes running, I haven't experienced usages
much over 60.

I'm hoping someone might have experience similar problems and knows a work
around?  I thought of compiling new versions of both products but I can't
find SRPMS for samba or atalk for the raq3, to find out how they were
configured and I'm not confident of just blindly using some 3rd parties
(even redhats) rpm and allowing it to clobber the existing setup. (from
memory there were conflicts when I tested the redhat samba rpm once before)

Thanks in advance
Andrew