Admin console server status page show stopped services

[carlswart]

Hi All,

Since there is no CentOS install, I used the --platform-override switch on install.sh with the RHEL5 distro.

It seems somehow the installer detected that I have rsyslog installed and proceeded to configure /etc/rsyslog.conf for me.

To fix, I did:

su - zimbra
zmcontrol stop
exit

vi /etc/syslog

Then I added: (from /etc/rsyslog.conf)

# Zimbra logs
local0.* -/var/log/zimbra.log
local1.* -/var/log/zimbra-stats.log
auth.* -/var/log/zimbra.log
mail.* -/var/log/zimbra.log

service syslog restart

su - zimbra
zmcontrol start
exit

Hope this helps.

[mowglie]

i already have these log entries in my log files & still get the errors/problems/issues posted earlier in this thread.

i also have the same entries but w/ @mail.domain.com instead of the /var/log parts. thats not an issue is it? it was automatically there after install.

[berjos]

We had a simular problem as carlswart, when the server was fresh and did the same as him. Worked as a charm.

Would also advice anyone that are having problems with services showing stopped, and/or logging not happening to see if the "zimbra" file in /var/spool/cron/ exists and that its not empty.

This file was for some reason empty on our server, so ones we restored it from backup services showed as running in admin console and logging/graphs etc worked as normal..

[heiko h.]

Hi,

I had exactly the same problem -> Server status and statistics

[ericortego]

Ran into this with an ldap slave. The issue for me was that the logs were getting to the zmlog server but the entries weren't being picked up by the parser due to a duplicate field not being logged by syslog servers.

The fix for me was to sed out the duplicate entries of the log file every so often.


in my case zcs0 is the host's name and my log entries looked like this:
Aug 18 23:53:34 zcs0 zcs0 zimbramon[5322]: 5322:info: zmstat convertd.csv: timestamp, utime, stime, cputime, rchar, wchar, read_bytes, write_bytes, rss, processes, threads:: 08/18/2011 23:53:34, 0, 0, 0, 0, 0, 0, 0, 10472, 6, 6

So my sed script does: 'sed -i -e 's/zcs0\ zcs0/zcs0/g' /var/log/zimbra-stats.log'

and now my check boxes are no longer red X's

[rojoblandino]

Same problem but i solved it:

1) I am using debian lenny using following zimbra version: Release 7.1.1_GA_3196.DEBIAN5 DEBIAN5 FOSS edition.
2) After moving to another clean server installation, averything work fine but gui admin status all green stop buttons.
3) I followed the /opt/zimbra/libexec/zmsyslogsetup procedure
4) I followed the /opt/zimbra/libexec/zmfixperms procedure

Then nothing happened, but then i got an idea like a light and i did following check:

5) ls -l /var/spool/cron/crontabs/
founding something like this:
-rw------- 1 10254 crontab 3151 ago 27 03:17 zimbra

6) I doubt about this perms and then i check my old server and found this file under zimbra user authority so i run as root of course:
chown zimbra /var/spool/cron/crontabs/zimbra

7) Then for any doubt i run again the crontab and syslog, i think were not necesary but i did it anyway:
/etc/init.d/cron restart
/etc/init.d/rsyslog restart

Sorry my english is not so good, hope it work for somebody here.

[sjangra]

hey man

run /opt/zimbra/libexec/zmsyslogsetup

[r1fle]

Dear colleagues,

For the new installations please check a sudoers file (run visudo) and change a line:
Defaults requiretty
to
#Defaults requiretty

Hope it helps!

Regards,
Andrey

[mgriego]

I was able to figure out why this was happening to us. I was noticing the same thing that ericortego was: in the zimbra-stats.log on our central logging host, the hostname field was being duplicated. After digging, I figured out why that was.

The issue for us was that our newly-installed MTA was installed on CentOS 6 whereas the logger host is running CentOS 5. CentOS 6 uses rsyslog while CentOS 5 uses sysklogd. rsyslog, even when using the RSYSLOG_TraditionalForwardFormat template, includes the timestamp and hostname in the message when sending it on to the remote host. When sending from a sysklogd logger, on the other hand, it only includes the syslog tag and message content. It does not send the timestamp nor hostname, so I can only assume this is what it's expecting to receive from a remote syslog.

So, to get around this problem, I had to create a custom template in rsyslogd on the CentOS 6 machine as none of the built-in templates seemed to do what was needed. Here are the lines I had to add to my rsyslog.conf file on the CentOS 6 machine. I may open a bug with Zimbra to add something similar to the zmsyslogsetup command. Anyway, here they are. They're basically the TraditionalForwardFormat without the timestamp and hostname fields:


$template ZimbraTemplate,"<%PRI%>%syslogtag:1:32%%msg:::sp-if-no-1st-sp%%msg%"
$ActionForwardDefaultTemplate ZimbraTemplate

[mgriego]

Actually, after finding out what the issue was, turns out there is already a bug open in Bugzilla (opened about 3 weeks ago): Bug 71141 – Server stats will not be updated if logger host is using sysklogd and senders rsyslogd

The solution in there is a tad better, but they both work. I have to say that I'm disappointed that Zimbra has basically decided that supporting RedHat 5 and RedHat 6 machines in the same environment is not in their best interest, as that's basically what they've done by saying that hosts with different syslog daemons aren't supported.