fc4 zcs-3.0.1 -> 3.1 upgrade worked great.. almost

14319KevinH · Post by **14319KevinH** » Fri Apr 14, 2006 11:08 am

zimbra user. Is port 22 open on your firewall? Is this a single node install?

adobrin · Post by **adobrin** » Fri Apr 14, 2006 11:14 am

its a single node install, the ssh port is open, and i changed the zimbraRemoteManagementPort to the correct ssh port.
running zmqstat as the zimbra user now works from a shell, after changing the spool/ dir to chmod 440 (zimbra is in the postfix group, which is the file group).
getting the same error in the admin console (which accurately reflects the correct ssh port).
do i have to do a tomcat restart?

adobrin · Post by **adobrin** » Fri Apr 14, 2006 11:18 am

tomcat restart didn't fix it either.

also

[zimbra@mail postfix-2.2.9]$ /opt/zimbra/libexec/zmqstat all

Can't chdir to /opt/zimbra/postfix-2.2.9/spool/all at /opt/zimbra/libexec/zmqstat line 182.

[zimbra@mail postfix-2.2.9]$

adobrin · Post by **adobrin** » Fri Apr 14, 2006 11:40 am

clicking on the individual queue tabs works.. so it updates each queue individually, just not the summary data.
is it sending the command "zmqstat all" and what is it looking to recieve?
ive tried hacking it to return the summary data for each queue, and setting $readfiles=1 for all.
both throw the same exception.
Message: system failure: exception executing command: zmqstat all

17629anand · Post by **17629anand** » Fri Apr 14, 2006 1:22 pm

adobrin,
zmqstat is invoked with sudo.
$ sudo /opt/zimbra/libexec/zmqstat
Does that work for you?

adobrin · Post by **adobrin** » Fri Apr 14, 2006 1:26 pm

yes, from a shell.

17629anand · Post by **17629anand** » Fri Apr 14, 2006 2:25 pm

adobrin,
it is very strange that individual queues work, but not the global queue count. They all use the same underlying calls (pretty much). I don't know why your zmqstat (the all version) command is erroring with 127.
can you run this command (with your server name ofcourse). If it runs, then it there is a problem in the Java ssh client level otherwise it is in the sshd/zmrcd level. Let's find out.

$ /opt/zimbra/libexec/zmrc bolton.liquidsys.com

HOST:bolton.liquidsys.com zmqstat all

(out) hold=5071

(out) corrupt=0

(out) deferred=2

(out) active=0

(out) incoming=0

Note that the lines marked OUT are the lines you should get - in your case the counts are probably 0 though.
Since you seem to be hacking zmqstat and what not you should be aware that the "zmqstat all" gets translated into plain "zmqstat" by zmrcd.

adobrin · Post by **adobrin** » Fri Apr 14, 2006 2:34 pm

anand,

the zmrc command is saying "port 22: Connection Refused"
note that neither zmlocalconfig zimbraRemoteManagementPort, nor zmprov ms mailserver.tld zimbraRemoteManagementPort are set to 22.

adobrin · Post by **adobrin** » Fri Apr 14, 2006 2:39 pm

if i manually add -p [zimbraRemoteManagementPort] to zmrc, the following error:
bash: /opt/zimbra/bin/zmrcd: No such file or directory
and a symbolic link from libexec/zmrcd to bin/zmrcd solves the issue.
thanks!

17629anand · Post by **17629anand** » Fri Apr 14, 2006 2:43 pm

yeah zmrc is hardcoded to use port 22 (hmm that might have to be bugzilla'd).
Try this instead (changing port number and host name):

$ echo "HOST:bolton.liquidsys.com zmqstat all" | ssh -T -i /opt/zimbra/.ssh/zimbra_identity -p 22 zimbra@bolton.liquidsys.com
I get:hold=5071

corrupt=0

deferred=2

active=0

incoming=0