[SOLVED] Zimbra sometimes not responding

[willy85]

Dear all,

I'm a new member to this forum, so please let me introduce myself. I'm currently working as a network staff and implementing Zimbra Collaboration Suite as an email server in my office.

Recently, Zimbra sometimes stop responding as if it hangs and can't be telnet from the client. FYI, i'm using telnet to detect whether my email server is working or not. And some of our users are reporting that they can't send/receive messages. But later, the Zimbra is working normally...

I myself has also found that when Zimbra is not working, I have to restart the network service on the server (I'm using CentOS 5.1). Have anyone experience the same problem as I did? Please let me know how to fix this problem... Thank you very much.


Best Regards,

Will

[bdial]

what port are you telnetting to to see if it's up? the standard telnet port? if thats the case, and you have to restart all networking than it sounds like this has pretty much nothing to do with zimbra.

[willy85]

@bdial:
I'm telneting over zimbra's POP3secure (995) and SMTP (25), since both are mainly used as the services. Sometimes it showed the sign that it's ok, sometimes it's not OK, the connection failed...

So you thinks that this has nothing to do with zimbra? If that's the case, do you have any suggestions for me? Thank you.

[bdial]

well i guess i'd be curious if anything else works while zimbra isn't working. can you ssh to the box fine during the times where zimbra is out to lunch? maybe run a continuous ping and see if you have any packet loss that corresponds to the same time periods?

[willy85]

I can't SSH the box whe zimbra is out to lunch, and FYI this is happens quite alot even after I restart the network service, sometimes it will hang again after the next 20-30 minutes. Do you think the router is the problem or the connection from the ISP is the problem?

Because my email server is accessible publicly and on last friday it was attacked / spammed. As if it was receiving numerous of emails and my email server just down... After that I tried to fix the problem by deleting the junk emails and restart the zmcontrol, it works fine again. But on the saturday, I'm receiving reports that the server can't be accessed from the public IP, but it works normally on my private LAN.

Please, anyone... A help or two is very much appreciated. Thank you.

[sn00p]

Do you have enough memory installed? We encountered such problems with servers with too low memory onboard. Java VM works very incorrect and unpredictable in case of unsufficient memory.

You are welcome to post output of such utilities as
free
vmstat
top
netstat -anp
dmesg

Are there any zombies in process list?

[willy85]

@sn00p:
I've had enough memory installed. Thank you for your kind attention

Anyway, the problem's solved (for today), even though I'm still maintaining it until this very moment. But it seems that after I physically restart the physical server which running the Zimbra service (I'm using 2 server and a virtual machine running on top of it).

We'll see for the next few days ahead. Once again, thank you very much for your kind support Glad being here

[phoenix]

Originally Posted by willy85

I've had enough memory installed.

What, exactly, do you call 'enough memory' and does it at least meet the minimum required specification?

[sn00p]

By the way, what virtualization technology do you use for your installation?
We use OpenVZ.

In OpenVZ container you could set and monitor different system parameters and even lack of resources of different types. The problem is that JavaVM is working sometimes unpredictable in virtualized server environment.

For example: you have to set 'privvmpages' parameter incredibly high (comparing to normal setting, calculated on base of installed RAM), even unlimited, in order to make Zimbra work correct.

No errors, nothing interesting in logs, all services in process list, zmcontrol status is ok, ports are all opened. And very similar situation. No response of services and so on. Anyway, software and OS problem, not Zimbra

We have two servers with 8Gb RAM onboard each, 3+3 openvz containers with zimbra components and approximately ~20000 accounts with average post traffic approximately 12000 +/- 5000 mails for a day. All works fine. After carefull tuning of system parameters, including sysctl.conf.

[willy85]

@phoenix:
There are 2 physical servers with each having 1GB and 2GB memory installed. I think that's enough, right? Since the minimum requirement for zimbra is 1GB. CMIIW.

@sn00p:
I'm using Xen virtualization technology here. Wow... 8GB RAM, i suppose your servers are handling numerous email traffic everyday. Oyeah, do you protect your server with iptables? Can you give me a suggestions for what kind of ports should I block in my email server? Anyone? Thank you.