SAMBA 2.2.3a Server Does Not Refresh its WINS Registration

[Alzhy]

I'm at a loss here.. my NT Admin says the fault is with SAMBA as it is supposed to renew its WINS registration. When the clients are unable to connect to my SAMBA shares, my log.nbmbd would show:

2002/06/18 06:10:39, 0] nmbd/nmbd_responserecordsdb.c:(236)
find_response_record: response packet id 11204 received with no matching record.

Cycling down and restarting Samba fixes the problem but 24 hours later I again would have to recycle SAMBA to fix this WINS problem..

Any ideas?

Thank you.

[Keith Lautenschlager · Accepted solution]

Nelson,

This is a problem that has been identified in Samba 2.2.3a, which is what CIFS Server A.01.08 is based upon. Until a fix is available, which is being worked on, the following workarounds have been suggested:


To re-register to a WINS server should samba fail to do so, restart the nmbd daemon. To lengthen the time between renewal, increase the values of Renewal Interval, Extinction Interval, and Extinction Time-out on the WINS server. Also ensure that "max ttl" in smb.conf is less than the timeout setting in WINS server.

[rainer doelker]

Keith is right!
HP has an open ChangeRequest for this issue: JAGae28593

BTW: This affects High-Availibility-System which run the cifs/9000 server package in the previous active_active configuration (That means: in failover-case one server is netbios-alias of the other AND the netbios-alias might get lost from WINS-db) !!!

Rainer

[Alzhy]

Thanks to the the two replies.. I already logged a call with HP and will wait for their patch. The workaround that I employed (since I could not request static) is a daily cron entry to restart NMBD. This solved our problems.

What I am wondering is why the SUN builds are not behaving this way? I have not tried building one for HP yet (outside of the CIFS depot) but I suppose I will have the same problem?

Thanks.