remaining files /tmp/zmcontrol.[error¦status].*

[mfehr]

Hi,

I run Zimbra 5.0.5 on Ubuntu 6.06 LTS Server since 3 weeks with 6 e-mail addresses in 2 domains.

I did a first analysis of how zimbra is running in my live environment and I noticed that I have a lot of file pairs /tmp/zmcontrol.error.* and /tmp/zmcontrol.status.* with a file size of 0 or 20 - 40 bytes long. BTW: The server never was stopped / started or restarted manually.

Checking the /var/log/zimbra.log* files does not show anything in common. As I receive a lot of spam (which luckily gets filtered out by Zimbra), there is activity on the incoming port and also at the same minute while the status and error file pair was created (according to the time stamp of the file).

This is not a disk space issue but there is some clean up missing and makes it more complicated to keep track on the files in /tmp - any hints?

[uxbod]

What is the ownership on those files ? Are you able to post some of the content from those files ?

[mfehr]

all files are owned by user zimbra group zimbra.

Out of the 43 file pairs, 14 zmcontrol.status.* files have a file size of 0 bytes, others are 24, 26 or 35 bytes long and have content in it. The content is as follows:

"saslauthd is running with pid 6225" (14 times) or "amavisd is running with pid 5328" (13 times) or "slapd is running with pid 3717" (2 times).

The PIDs are consistent for the associated deamons. This means that the server did not get restarted automatically nor manually.

all zmcontrol.error.* files have a file size 0 bytes

[uxbod]

By the looks of things they are created by /opt/zimbra/bin/zmcontrol, but in that it does perform a unlink of those files. Very odd.

[uxbod]

It may be related to Bug 23668 - zmcontrol throw exceptions when it calls 'zmprov -l' before ldap master is in ready state or Bug 19563 - zmcontrol fails without printing an error
In the output file which is the last daemon that is listed ?

[mfehr]

There is only one line per status-file.

I modified the zmcontrol file with additional logging. After the zmcontrol.status* file is created there is a return without unlinking the 2 files. I'll check with the log to find a potential correlation.

[mfehr]

I altered the code of zmcontrol to log a line like "Checking status FOR ERROR-STATUS" when entering the function doStatus.

What I found out is that the error condition occurs in a timeout situation. Should a timeout occur, a log entry is created "zmstatuslog timeout after 60 seconds" and the script terminates. At this time, the file does not get closed.

See part of my log when the error occured (BTW: The file not being deleted has a time stamp of 05:51 hrs):

May 10 05:50:14 mail zimbramon[6482]: 6482:info: Checking status FOR ERROR-STATUS
May 10 05:50:16 mail amavis[20979]: (20979-04) Blocked SPAM, [xxx.xx.xxx.xx] [xxx.xx.xxx.xx] <xxxxxxxxxxxx@xxxxxxxx.com> -> <xxxxxx@xxxxxx.xx>, Message-ID: <xxxxxxxxxxxxxxxxxxxxxxxxxx@xxxxxxxxxxxx>, mail_id: bwUKmDjLXK-7, Hits: 25.608, size: 7591, 7228 ms
May 10 05:50:16 mail postfix/smtp[6635]: 6CAC6AFF95: to=<xxx@xxx.com>, relay=127.0.0.1[127.0.0.1]:10024, delay=11, delays=4.1/0.03/0.03/7.2, dsn=2.7.0, status=sent (250 2.7.0 Ok, discarded, id=20979-04 - SPAM)
May 10 05:50:16 mail postfix/qmgr[6212]: 6CAC6AFF95: removed
May 10 05:50:30 mail postfix/smtpd[6054]: connect from yyy.com[xx.x.x.xx]
May 10 05:50:38 mail CRON[6450]: (pam_unix) session closed for user zimbra
May 10 05:50:38 mail postfix/smtpd[6054]: NOQUEUE: reject: RCPT from yyy.com[xx.x.x.xx]: 550 5.1.1 <xxx@xxx.com>: Recipient address rejected: xxx.com; from=<zzz@zzz.com> to=<xxx@xxx.com> proto=ESMTP helo=<yyy.com>
May 10 05:50:40 mail postfix/smtpd[6054]: lost connection after DATA from yyy.com[xx.x.x.xx]
May 10 05:50:40 mail postfix/smtpd[6054]: disconnect from yyy.com[xx.x.x.xx]
May 10 05:50:47 mail zmmailboxdmgr[6763]: status requested
May 10 05:50:47 mail zmmailboxdmgr[6763]: status OK
May 10 05:50:58 mail clamd[5520]: Database correctly reloaded (450460 signatures)
May 10 05:51:01 mail zimbramon[6456]: 6456:info: zmstatuslog timeout after 60 seconds
May 10 05:51:03 mail postfix/smtpd[6054]: connect from yyy.com[xx.x.x.xx]
May 10 05:51:05 mail CRON[6455]: (pam_unix) session closed for user zimbra
May 10 05:51:06 mail postfix/smtpd[6054]: NOQUEUE: reject: RCPT from yyy.com[xx.x.x.xx]: 550 5.1.1 <xxxm@xxx.com>: Recipient address rejected: xxx.com; from=<uuu@uuu.com> to=<xxxm@xxx.com> proto=ESMTP helo=<yyy.com>
May 10 05:51:09 mail postfix/smtpd[6054]: lost connection after DATA from yyy.com[xx.x.x.xx]
May 10 05:51:09 mail postfix/smtpd[6054]: disconnect from yyy.com[xx.x.x.xx]
May 10 05:52:01 mail CRON[7036]: (pam_unix) session opened for user zimbra by (uid=0)
May 10 05:52:06 mail zimbramon[7046]: 7046:info: Checking status FOR ERROR-STATUS
May 10 05:52:08 mail zmmailboxdmgr[7174]: status requested
May 10 05:52:08 mail zmmailboxdmgr[7174]: status OK
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: antispam: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: antivirus: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: ldap: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: logger: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: mailbox: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: mta: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: snmp: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: spell: Running
May 10 05:52:11 mail zimbramon[7037]: 7037:info: 2008-05-10 05:52:02, STATUS: mail.xxx.com: stats: Running
May 10 05:52:11 mail CRON[7036]: (pam_unix) session closed for user zimbra


The file zmcontrol has defined a line $SIG{ALRM} = sub { print "Timeout after $timeout seconds\n"; exit(1) };

The files should be unlinked before exit(1) is called.