1

Topic: Problems with EAS

==== REQUIRED BASIC INFO OF YOUR IREDMAIL SERVER ====
- iRedMail version (check /etc/iredmail-release): 0.99
- Deployed with iRedMail Easy or the downloadable installer? downloadable installer
- Linux/BSD distribution name and version: Centos 7.5
- Store mail accounts in which backend (LDAP/MySQL/PGSQL): LDAP
- Web server (Apache or Nginx): Nginx
- Manage mail accounts with iRedAdmin-Pro? Yes
- [IMPORTANT] Related original log or error message is required if you're experiencing an issue.
====
We're having a lot of problems again with EAS.   I've made the following changes in an attempt to correct the problem:
WOWorkersCount is 625 (up from 550)
WOWatchDogRequestTimeout 60 (up from 30)
SOGoMaximumPingInterval 3540
SOGoMaximumSyncInterval 3540
SOGoInternalSyncInterval 120 (up from 30)

People are unable to connect and messages like the following are appearing in logs:
Aug 31 21:08:44 sogod [5117]: [WARN] <0x0x56332e6bdfb0[WOWatchDogChild]> pid 5803 has been hanging in the same request for 32 minutes
Aug 31 21:08:44 sogod [5117]: [WARN] <0x0x56332e642df0[WOWatchDogChild]> pid 5146 has been hanging in the same request for 5 minutes
Aug 31 21:08:44 sogod [5596]: <0x0x56332eb59f00[SOGoActiveSyncDispatcher]> Sleeping 120 seconds while detecting changes for user (account name) in Ping...
Aug 31 21:08:44 sogod [6493]: <0x0x56332ec1c2e0[SOGoActiveSyncDispatcher]> Sleeping 120 seconds while detecting changes for user (account name) in Ping...

I'm also getting the No child available to handle request messages again.

When I tried setting the worker threads to 675 iRedAdmin became very slow.

Any suggestions on how I can tune this to avoid this problem?

Thanks,
bob

2

Re: Problems with EAS

On CentOS/Debian/Ubuntu, parameter "WOWorkersCount" in /etc/sogo/sogo.conf doesn't work, you should increase "PREFORK=" in /etc/sysconfig/sogo (CentOS) or /etc/default/sogo (Debian/Ubuntu) instead.

----

Buy me a cup of coffee ($5) to support iRedMail:

buy me a cup of coffee

3

Re: Problems with EAS

Thanks for the reply.   I'm not relying on the WOWorkersCount parameter in sogo.conf.    I'm running two instances, one for EAS and one for HTTP, and wanted to have different parameters for each.   Because of this I've modified the settings directly within my systemd service files.   

For my EAS instance I have the ExecStart line as follows:
ExecStart=/usr/sbin/sogod -WOPort 127.0.0.1:30000 -WOWorkersCount 625 -WOWatchDogRequestTimeout 60 -SOGoMaximumPingInterval 3540 -SOGoMaximumSyncInterval 3540 -SOGoInternalSyncInterval 120 -WOPidFile /var/run/sogo/sogo-eas.pid -WOLogFile /var/log/sogo/sogo-eas.log

Based upon the process count I believe this is working.   Note the following:
[root@stmail system]# ps aux | grep eas | wc -l
629

This is a little higher than it should be.  It's matching three other processes that just happen to contain the letters eas.   These are unrelated to sogo.   It's also matching grep itself.

Bob

4

Re: Problems with EAS

ok, what's the issue now? Is it solved?

----

Buy me a cup of coffee ($5) to support iRedMail:

buy me a cup of coffee

5

Re: Problems with EAS

ZhangHuangbin wrote:

ok, what's the issue now? Is it solved?

It's not solved.   We have situations when the server becomes very slow.   Stopping the EAS service immediately fixes the problem.   After EAS is started all is well for awhile, then the errors appear in the log.   Also, many people complain they can't use e-mail via their phones (Activesync/EAS).

Is it possible to run the EAS server on a separate server?   This will allow the resources of the original server to be used toward iRedmail.   The second server resources can be dedicated 100% toward EAS.

Thanks,
Bob

6

Re: Problems with EAS

bdushok wrote:

Is it possible to run the EAS server on a separate server?   This will allow the resources of the original server to be used toward iRedmail.   The second server resources can be dedicated 100% toward EAS.

Sure you can install Nginx + SOGo on another server, and configure SOGo to connect to iRedMail server to query user accounts, etc.

btw, How much memory does iRedMail server have? SOGo is memory hungry.

----

Buy me a cup of coffee ($5) to support iRedMail:

buy me a cup of coffee

7

Re: Problems with EAS

Thanks for the info.   

The iRedMail server has 64GB of ram.   

I'm running iRedMail in a VM on ESXi and have been watching performance for potential problems.   CPU goes up a bit and very occasionally spikes at 100% for brief periods.   Overall the CPU utilization is low.   Memory also seems to be low, which isn't what I've been expecting.   

I have somewhere near 1000 EAS users.   I've noticed if I shutdown the EAS instance performance seems to increase significantly.   Do you feel adding more ram will help?   The memory usage stats are puzzling.   I personally feel the graph should be showing much higher usage.

I'll tinker with a second server.    It seems like I should be able to install the OS packages for Sogo and copy the config from sogo.conf/sogod-eas.service from iRedMail to get it running.   I'll, of course, need to make a few changes to the config.   The current config references 127.0.0.1 for ldap, etc.   Those references need to be changed to the iRedMail server.

Thanks,
Bob

8

Re: Problems with EAS

64GB for 1000 EAS users should be more than enough.

----

Buy me a cup of coffee ($5) to support iRedMail:

buy me a cup of coffee