Logger says "running" on start but stopped in web admin

[Nico]

Hello and thank you Achille,

I tried that Achille has proposed :

1) I stopped zimbra services (zmcontrol stop) and killed all tail (killall tail)

2) I looked in /dev/shm : no files were mounted.

3) I looked the permissions of /var/log/zimbra.log : the owner was 'root' for user and group. I changed it for 'zimbra'. Now the owner (user and group) is zimbra.

4) I checked the size of /var/log/zimbra.log : the size is 0 !

5) (I didn't make a change in '/etc/syslog-ng/syslog-ng.conf' ; and inside it there is no line with #zimbra at the end)

6) In the files :
- /opt/zimbra/bin/zmsyslogsetup
- /opt/zimbra/conf/zmlogrotate
- /etc/logrotate.d/zimbra
I replaced 'killall -HUP syslogd' by '/sbin/rcsyslog restart'

7) I restarted syslog-ng with: '/sbin/rcsyslog restart'

8) I did 'mount -a' and restarted zimbra (zmcontrol start)

But the problem is still here, the same 'error' message appear in the web admin. And the size of /var/log/zimbra.log is still 0.

[Achim.Theobald]

Originally Posted by Nico

Hello and thank you Achille,
5) (I didn't make a change in '/etc/syslog-ng/syslog-ng.conf' ; and inside it there is no line with #zimbra at the end)
But the problem is still here, the same 'error' message appear in the web admin. And the size of /var/log/zimbra.log is still 0.

Read my reply !

####
Copy the zimbra changes from "/etc/syslog-ng/syslog-ng.conf.in" to "/etc/syslog-ng/syslog-ng.conf" (the last 5 lines with "#zimbra" comment at end)
####

The zimbra changes you can find in "/etc/syslog-ng/syslog-ng.conf.in" !!!

Or run as root /opt/zimbra/zmsyslogsetup. Or copy these lines to the bottom of your /etc/syslog-ng/syslog-ng.conf and restart syslog-ng !


filter f_local0 { facility(local0); }; # zimbra
destination zmail { file("/var/log/zimbra.log" owner("zimbra") ); }; # zimbra
log { source(src); filter(f_mail); destination(zmail); }; # zimbra
destination local0 { file("/var/log/zimbra.log" owner("zimbra") ); }; # zimbra
log { source(src); filter(f_local0); destination(local0); }; # zimbra


Cu

Achim

[Dirk]

Ok, I stepped through these instructions and made the same mistake as Nico, and I got exactly the same results!

So I tried again, this time I manually added those 5 lines to the bottom of both the files in the syslog folder, as they were not there, and running the zmsyslogsetup did not add them.

Then I started the syslog and checked the filesize of /var/tmp/zimbra.log, and finally it was non-zero!!

Time to start zimbra services back up, and when they completed (takes ages on this machine) I checked the filesize again, 9k! woo, lots of errors in there though, and as yesterday, I cant get to the web admin console to see what's happening but I expect the logger will be able to report at least something now.

Achim, thank you.

[Nico]

Thank you Achim !!

As Dirk, I copied the five lines at the botom of both the files :
- /etc/syslog-ng/syslog-ng.conf.in
- /etc/syslog-ng/syslog-ng.conf

And as Dirk, after having restart the syslog-ng, the filesize of /var/log/zimbra.log was not 0 anymore.

I launched the Zimbra services, and check /var/log/zimbra.log : now it works.

But not as Dirk, in my web admin, I haven't the error message anymore. I can see the list of the services running


Thank you very much for your help !

[Dirk]

That's great news Nico.

Now, this seems to be an issue caused by Suse10.1, so I wonder, do we need to post to bugzilla the details of this, or do the zimbra dev team read these forums enough to make the changes to the binaries for the next release?

[Achim.Theobald]

Originally Posted by Dirk

That's great news Nico.

Now, this seems to be an issue caused by Suse10.1, so I wonder, do we need to post to bugzilla the details of this, or do the zimbra dev team read these forums enough to make the changes to the binaries for the next release?

It's imho not a Bug in zimbra. The problem is your syslog-ng.conf btw. syslog-ng.conf.in is not "clean". Everyone have make changes in this files and so say the SuSEconfig this:

post:/etc/syslog-ng # /opt/zimbra/bin/zmsyslogsetup
Starting SuSEconfig, the SuSE Configuration Tool...
Running module syslog-ng only
Reading /etc/sysconfig and updating the system...
Executing /sbin/conf.d/SuSEconfig.syslog-ng...

ATTENTION: You have modified //etc/syslog-ng/syslog-ng.conf. Leaving it untouched...
You can find my version in //etc/syslog-ng/syslog-ng.conf.SuSEconfig...


Cu

Achim

[Dirk]

Well, to be fair, I installed the same version of Zimbra onto two computers, one running Suse 10.0 and the other running Suse 10.1

The logger works out of the box on 10.0 but does not work until I made the changes you suggested on 10.1

Given that the changes you suggest worked right away for two people (at least) is it not something that the installation package should be aware of, so that it can work right away on suse10.1 without someone needing to trawl the net for this thread to get it working?

I'm not moaning, I know the prebuilt binary is for suse10.0 only, I just think it would be nice to have a 10.1 install too.

[KevinH]

Originally Posted by Dirk

Now, this seems to be an issue caused by Suse10.1, so I wonder, do we need to post to bugzilla the details of this, or do the zimbra dev team read these forums enough to make the changes to the binaries for the next release?

We read them but to have things logged and fixed they must go into bugzilla. It would be too hard to fish through the 1000's of posts here and track the bugs. Bugzilla does a much better job of that. So please file it. Search first to make sure someone else didn't beat you to it.

[Dirk]

Thanks, I was just unsure of the protocol one should follow regardign logging bugs, as I've not done it before.

It's logged now at: http://bugzilla.zimbra.com/show_bug.cgi?id=7904

Hopefully I've worded it ok.

[frank_g]

Hi,

there is - up to Zimbra 3.1.4 - no tag for SuSE 10.1 in /opt/zimbra/libexec/get_plat_tag.sh, so it returns "UNKNOWN". This script is called by "zmsyslogsetup", which adds the needed lines to "/etc/syslog-ng/syslog-ng.conf.in" and calls SuSEConfig.

If it gets back "UNKNOWN" from "get_plat_tag.sh", it does nothing -> that's the bug.

Simple solution: just copy the five lines beginning with "grep "SUSE LINUX 10.0..." and replace "10.0" with "10.1" in the copy. If you run zmsyslogsetup now, the correct lines are copied to "/etc/syslog-ng/syslog-ng.conf.in".

Frank