Hi,
I have a samba (3.6.20) + ctdb cluster running on FreeBSD 8.2 and recently it
run into problems.
log.ctdb increased very quickly and consumed 100% of the disk space, the log is
full of message like:
2015/03/19 08:09:06.070897 [26053]: 00.ctdb: OPERATION: monitor
2015/03/19 08:09:06.278859 [26053]: 50.samba: [info] cy_samba_control monitor :
0
2015/03/19 08:09:07.054677 [vacuum-connections.tdb:21348]:
/usr/local/var/ctdb/connections.tdb.0:tdb_rec_read bad magic 0x42424242 at
offset=5838176
2015/03/19 08:09:07.054738 [vacuum-connections.tdb:21348]:
server/ctdb_vacuum.c:772 Traverse error in vacuuming 'connections.tdb'
2015/03/19 08:09:07.054759 [vacuum-connections.tdb:21348]:
server/ctdb_vacuum.c:1463 Failed to vacuum 'connections.tdb'
I searched on internet and there seems no answers to the cause of the problem,
the general advice is delete connections.tdb and restart samba.
I tried this but the problem happen repeatedly and soon log.ctdb is full again.
Is there any suggestion on this?
Another problem I met is fstat show smbd open more then 500,000 files but
smbstatus show only 600+ open files.
Is it possible that samba may leak fd?
Thanks in advance!
Best regards,
James Pan
