ctl_cyrusdb is very long

[eric.doutreleau]

Hi

I'm using cyrus 2.1.5 on a redhat linux 7.3 machine.

I have a problem with ctl_cyrusdb 
I had a crash with my server and when i reboot ctl_cyrusdb is
taking
too much time.
Indeed it s still running for 20 hours.
In my /var/lib/imap/db directory i have the following files
-rw-------    1 cyrus    mail         8192 nov 20 15:02 __db.001
-rw-------    1 cyrus    mail       270336 nov 20 15:02 __db.002
-rw-------    1 cyrus    mail        98304 nov 20 15:02 __db.003
-rw-------    1 cyrus    mail     17063936 nov 20 15:05 __db.004
-rw-------    1 cyrus    mail        32768 nov 20 15:05 __db.005
-rw-------    1 cyrus    mail     10485747 nov  9 13:28
log.0000000155
-rw-------    1 cyrus    mail     10485751 nov 10 13:27
log.0000000156
-rw-------    1 cyrus    mail     10485736 nov 10 13:28
log.0000000157
-rw-------    1 cyrus    mail     10485755 nov 10 17:32
log.0000000158
-rw-------    1 cyrus    mail     10485736 nov 11 13:27
log.0000000159
-rw-------    1 cyrus    mail     10482458 nov 11 16:46
log.0000000160
-rw-------    1 cyrus    mail     10485643 nov 12 13:10
log.0000000161
-rw-------    1 cyrus    mail     10485735 nov 12 13:28
log.0000000162
-rw-------    1 cyrus    mail     10485705 nov 12 22:18
log.0000000163
-rw-------    1 cyrus    mail     10485674 nov 13 13:27
log.0000000164
-rw-------    1 cyrus    mail     10481898 nov 13 21:43
log.0000000165
-rw-------    1 cyrus    mail     10482734 nov 14 12:56
log.0000000166
-rw-------    1 cyrus    mail     10485662 nov 14 15:03
log.0000000167
-rw-------    1 cyrus    mail     10485758 nov 14 22:19
log.0000000168
-rw-------    1 cyrus    mail     10485715 nov 15 13:30
log.0000000169
-rw-------    1 cyrus    mail     10485760 nov 15 16:46
log.0000000170
-rw-------    1 cyrus    mail     10485648 nov 16 13:27
log.0000000171
-rw-------    1 cyrus    mail     10485650 nov 16 13:28
log.0000000172
-rw-------    1 cyrus    mail     10485746 nov 17 13:27
log.0000000173
-rw-------    1 cyrus    mail     10485760 nov 17 13:28
log.0000000174
-rw-------    1 cyrus    mail     10481830 nov 17 22:38
log.0000000175
-rw-------    1 cyrus    mail     10485749 nov 18 13:28
log.0000000176
-rw-------    1 cyrus    mail     10485726 nov 18 13:33
log.0000000177
-rw-------    1 cyrus    mail     10485749 nov 18 17:02
log.0000000178
-rw-------    1 cyrus    mail     10485760 nov 18 23:55
log.0000000179
-rw-------    1 cyrus    mail     10485703 nov 19 13:32
log.0000000180
-rw-------    1 cyrus    mail     10485721 nov 19 19:23
log.0000000181
-rw-------    1 cyrus    mail     10485708 nov 20 13:27
log.0000000182
-rw-------    1 cyrus    mail      7454187 nov 20 14:58
log.0000000183

The cvt_cyrusdb seem to take time to read all the files
log.000000*.

Is there a way to speed up that process as during this time
the imap
server is unavailable.

I have two backups directory
/var/lib/imap/db.backup1 and
/var/lib/imap/db.backup2

But these directories contains only the following files
[root at molure db]# cd ..
[root at molure imap]# ls
db          db.backup2  log           mailboxes.db.old  proc 
 sieve   tls_sessions.db
db.backup1  deliver.db  mailboxes.db  msg               quota
 socket  user
[root at molure imap]# cd db.backup1
[root at molure db.backup1]# ls -l
total 5364
-rw-------    1 cyrus    mail      4664227 nov  9 13:27
log.0000000155
-rw-------    1 cyrus    mail       811008 nov  9 13:27
mailboxes.db

As far as i can understand these directories should contain a
checkpoint
of the database every 30 minutes.
In the cyrus logfile 
I got Nov 20 11:57:46 molure ctl_cyrusdb[2914]: checkpointing
cyrus databases
Nov 20 12:27:46 molure ctl_cyrusdb[3318]: checkpointing cyrus
databases
Nov 20 12:57:46 molure ctl_cyrusdb[3929]: checkpointing cyrus
databases
Nov 20 13:27:46 molure ctl_cyrusdb[4329]: checkpointing cyrus
databases
Nov 20 13:57:46 molure ctl_cyrusdb[4857]: checkpointing cyrus
databases
Nov 20 14:27:46 molure ctl_cyrusdb[5378]: checkpointing cyrus
databases
Nov 20 14:57:46 molure ctl_cyrusdb[6008]: checkpointing cyrus
databases

But it seems never been done since 9 november.

does someone knows what happened and how i can speed up that
process?

P.S
all my databases are in DB3 format.






Accédez au courrier électronique de La Poste : www.laposte.net ;
3615 LAPOSTENET (0,13 €/mn) ; tél : 08 92 68 13 50 (0,34€/mn)"