Testing the 3.0.20pre, I found the following serious bug. My sambaserver is a Domainmember (no trusted domain) on a Solaris/Sparc Platform. I found this bug for the first time in the 3.0.20pre2-release, the 3.0.15pre2 works fine. The BUG seems to be similar to bug 1008. When I start winbindd, it gave up to get a lock on a file, which is already locked by smbd. After that winbindd crash. I cant't connect to the sambaserver. Without winbindd samba works. Here the important locations in the logfile: *** (smbd) *** [2005/07/29 14:07:39, 8] lib/util.c:(1815) fcntl_lock 16 34 0 1 2 [2005/07/29 14:07:39, 8] lib/util.c:(1850) fcntl_lock: Lock call successful *** (winbindd) *** [2005/07/29 14:07:39, 8] lib/util.c:(1815) fcntl_lock 16 34 0 1 1 [2005/07/29 14:07:39, 3] lib/util.c:(1826) fcntl_lock: fcntl lock gave errno 11 (Resource temporarily unavailable) [2005/07/29 14:07:39, 3] lib/util.c:(1845) fcntl_lock: lock failed at offset 0 count 1 op 34 type 1 (Resource temporarily unavailable)
Created attachment 1341 [details] log.smbd
Created attachment 1342 [details] log.wb
Two things: Apparently you are facing a segfault in winbind that has been fixed for 3.0.20rc1 which was just released, so please retry that. The log.smbd you sent does not really show anything relevant. It appears that you have 'log file = ...log.%m' or something else with %m in your smb.conf. If you still see the bug with 3.0.20rc1, could you please remove this line or alternatively send in a tar with all smbd logfiles? Thanks, Volker
(In reply to comment #3) > Two things: Apparently you are facing a segfault in winbind that has been fixed > for 3.0.20rc1 which was just released, so please retry that. > > The log.smbd you sent does not really show anything relevant. It appears that > you have 'log file = ...log.%m' or something else with %m in your smb.conf. If > you still see the bug with 3.0.20rc1, could you please remove this line or > alternatively send in a tar with all smbd logfiles? > > Thanks, > > Volker Hallo, I tested 3.0.20rc1 and all seems to be work fine. Thanks, Olaf
closing.
sorry for the same, cleaning up the database to prevent unecessary reopens of bugs.