0

my virtual server is used as a mailserver (with Zimbra 7.2.0_GA_2669) and webserver. Yesterday the Zimbra web interface was not accessible anymore, though all services were running. Since two hours, the whole server cannot be accessed anymore. Not via web, not via ssh.

It's a container, so I can get information through Parallels Power Panel. There it says:

CPU Usage 6.69% / Load Average 0.42, 0.46, 0.42

Memory Usage 28.72% (of total 8.8 GB)

Disk Usage 30.97% (of total 250.00 GB)

Disk Inodes 5.52% (of total 12,000,000)

Furthermore, the number of TCP sockets (numtcpsock) is at the maximum of 1,024 connections.

I don't understand this behaviour and therefore hesitate to just restart the whole container. What do you suggest? And how do you read these values?

Thanks a lot!

user2092982
  • 31
  • 1
  • 1
  • 7
  • What about network traffic/bandwidth usage? Is your server being DOSed? – Marki Sep 22 '13 at 10:12
  • Thanks for your instant answer! The traffic seems to be fine right now; the monthly average is about 200 MB for outgoing traffic per day (and ~50 MB for incoming). Yesterday, outgoing traffic was 131.44 MB (incoming ~30 MB), today 20.87 MB (incomig 4 MB). The only big difference was two days ago, there an extraordinary amount of 1,001.40 MB of outgoing traffic was recorded - five times more than the average. Incoming was regular that day. – user2092982 Sep 22 '13 at 10:24
  • Does the server still ping? – Marki Sep 22 '13 at 10:25
  • Yes, it responds within 12 to 22 ms. – user2092982 Sep 22 '13 at 10:31
  • Any ideas on the next steps? – user2092982 Sep 22 '13 at 12:35
  • Restarted the whole container, now everything's working again. Zimbra service needed another manual restart to work properly. – user2092982 Sep 22 '13 at 14:24

1 Answers1

0

Restarted the whole container, now everything's working again. Zimbra service needed another manual restart to work properly.

It seems the high amount of TCP connections was caused by a combination of unsuccessful outgoing Zimbra connections and a quite high timeout value. My assumption: Open connections failed but were not closed for a long time.

user2092982
  • 31
  • 1
  • 1
  • 7