DSfW server shows samba cored in messages log, but no core exists

  • 7006443
  • 13-Jul-2010
  • 27-Apr-2012

Environment

Novell Open Enterprise Server 2
Domain Services for Windows
DSFW
(OES 2SP2)
(OES 2SP3)

Situation

Linux member servers are joined to the domain.

The /var/log/messages show the following:

=================================================
Nov 19 18:26:00 wgp-dt214 smbd[1062]:
===============================================================
Nov 19 18:26:00 wgp-dt214 smbd[1062]: [2009/11/19 18:26:00, 0]
smbd/server.c:exit_server_common(752)
Nov 19 18:26:00 wgp-dt214 smbd[1062]: Abnormal server exit: funnel closed on
write
Nov 19 18:26:00 wgp-dt214 smbd[1062]: [2009/11/19 18:26:00, 0]
smbd/server.c:exit_server_common(753)
Nov 19 18:26:00 wgp-dt214 smbd[1062]:
===============================================================
Nov 19 18:26:00 wgp-dt214 smbd[1062]: [2009/11/19 18:26:00, 0]
lib/util.c:log_stack_trace(1737)
Nov 19 18:26:00 wgp-dt214 smbd[1062]: BACKTRACE: 13 stack frames:
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #0
/usr/sbin/smbd(log_stack_trace+0x2d) [0x8023852d]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #1 /usr/sbin/smbd [0x8030f4dc]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #2 /usr/sbin/smbd [0x8030f6c3]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #3
/usr/sbin/smbd(ncalrpc_write_to_funnel+0xfa) [0x8017905a]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #4 /usr/sbin/smbd(write_to_pipe+0x6e)
[0x80175cae]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #5 /usr/sbin/smbd [0x80067c51]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #6 /usr/sbin/smbd [0x80068186]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #7 /usr/sbin/smbd(reply_trans+0x6c3)
[0x80068fa3]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #8 /usr/sbin/smbd [0x800c2340]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #9 /usr/sbin/smbd(smbd_process+0x84b)
[0x800c370b]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #10 /usr/sbin/smbd(main+0xbdd)
[0x8031151d]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #11
/lib/libc.so.6(__libc_start_main+0xdc) [0xb72df89c]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: #12 /usr/sbin/smbd [0x8004a461]
Nov 19 18:26:00 wgp-dt214 smbd[1062]: [2009/11/19 18:26:00, 0]
smbd/server.c:exit_server_common(757)
Nov 19 18:26:00 wgp-dt214 smbd[1062]: Last message was SMBtrans
Nov 19 18:26:00 wgp-dt214 smbd[1062]: [2009/11/19 18:26:00, 0]
lib/fault.c:dump_core(181)
Nov 19 18:26:00 wgp-dt214 smbd[1062]: dumping core in
/var/log/samba/cores/smbd
Nov 19 18:26:00 wgp-dt214 smbd[1062]:
Nov 19 18:26:01 wgp-dt214 /usr/sbin/cron[3163]: (root) CMD (ntpdate -s
164.99.147.25)

No core file exists in /var/log/samba/cores/smbd.

Resolution

This is a cosmetic error, but is still under investigation.

A quick explanation on why this is reported in the messages log:
If a connection is idle for more than 5 minutes the connection is closed on the server side (dcerpc).  For the error message to occur the client (smb) attempts to re-establish the connection on the closed pipe (due to being idle for more than 5 min.) instead of creating a new connection, thus resulting in the error seen in the messages log.

Additional Information

The samba member server does a search for highestcommitedUSN every 5-10 min.  The higestcommitedUSN value in DSfW is the timestamp of the partition.  This can change.  If the value of the highestcommitedUSN is changed the samba server does a dsenumtrustdomains call.  Samba can send a dsenumdomainTrust every 5-10 min.  After the initial pipe is created, the same pipe is used for the enumerate request.  If the connection is idle for more than 5 min. the dcerpc will close the pipe.  Samba on the member server detects the pipe is closed and attempts to  auto reconnect on the closed pipe.  This results in the error seen in the /var/log/messages.  No functionality is lost.  A connection will be re-establised on a new pipe.