Koozali.org: home of the SME Server
Obsolete Releases => SME Server 8.x => Topic started by: mike_mattos on August 18, 2015, 01:55:57 PM
-
I'm troubleshooting an oplock issue at a clients, and realized that the failures were not totally random! About 15 times over 5 days, a cluster of op-lock issues like
Oplock break failed for file shared/xxx -- replying anyway [2015/08/14 05:05:34.222777, 0] smbd/oplock.c:330(oplock_timeout_handler)
But in the past 5 days, there have been no failures before 5AM, but three times between 05:00 and 05:10 there have been problems. I can't see any application correlation, the housekeeping is done around midnight, and regular processing starts around 08:00 so what happens at 05:00? Also the shared file seems to be random. The application is monitoring data files and processing new entries, and there are no new entries at 5AM.
log.smbd last entry was at system reboot, and only one PC was running at 5AM so all 5AM failures relate to that PC. But the failures associate with almost all PC's at the site for the other dozen failures, including failures when the 5AM station was not running.
SO my question is, what is SME8.1 doing at 05:00 that affects samba?
-
more log analysis reveals clusters of the oplock issue occurring at 4 hours intervals, which I believe is the opslock KEEPALIVE interval, but almost daily failures at 05:xx, 09:xx, 13:xx , 17:xx , 21:xx
No failures 0-4, 6-8, 10-12 14-16
-
SO my question is, what is SME8.1 doing at 05:00 that affects samba?
It's not necessarily what SME is going. The log message talks about an oplock break operation. That's probably one client asking for the oplock held by another client. What are your clients doing?
This article suggests you could try "blocking locks = no".
http://www.linuxquestions.org/questions/linux-software-2/samba-oplock-break-failed-error-cannot-open-files-on-smb-share-from-network-4175446588/
-
AT 5am the clients are doing NOTHING! And nothing is on a 4 hour cycle, it is either daily at midnight, or several times per hour.
But the locks jam up every four hours. INterstingly, other sites have mostly EXCLUSIVE +BATCH locks, this site is largely NONE.
I've disabled AV on the stations, too. And in var/logs/samba not all station have log entries, BUT I've also had issues when I disabled suspect stations. Also replaced the network switch and network card in the server.
ONly just getting enough data to do the failure analysis and see the 4 hour repeating failure cycle now.