imap server hangs

[sg]

hi,

one of our scalix installations had a very bad performance during the last few days.

After I solved several other problems on this instance I figured out, that the imap daemon was responsible for the performance problems.

After I restarted the server the load increased rapidly up to 40.

Now I have killed the imap processes with killall. After that I started the IMAP service again with omon imap.

By now, everything seems okay.

My question:

What is the reason that a restart of the the server won't help but killing all imap processes manually? Is it a bug in scalix?

I'm currently using version 11.0.0.431. Should this problem be solved with an update?

sebastian

[les]

hi,

one of our scalix installations had a very bad performance during the last few days.

After I solved several other problems on this instance I figured out, that the imap daemon was responsible for the performance problems.

After I restarted the server the load increased rapidly up to 40.

Now I have killed the imap processes with killall. After that I started the IMAP service again with omon imap.

By now, everything seems okay.

My question:

What is the reason that a restart of the the server won't help but killing all imap processes manually? Is it a bug in scalix?

I'm currently using version 11.0.0.431. Should this problem be solved with an update?

sebastian

Going up to 11.3 is highly recommended. There were some imap related bugs fixed in newer versions, have a look at the release notes.

one of the problems related to defunct imap processes staying around and eventually blocking user access was fixed in 11.2.

[obrodkin]

This might sound as a similar problem I am facing.

After updating to 11.3 (fresh new install and DB migration) imap processes exhibit behavor that I never encountered before. Once a week (or 3-4 days) imap processes go wild for a user (sometimes two). 'omstat -u all' shows reasonable number of imap processes for this particular user, but "ps aux | grep imap" reveals that there 40 or 50 of them. I monitor server and number of imap processes. As soon as imap spin-off begun, server become unresponsive and need to be rebooted. Killing imap and omoff/on does not help.

Here is my little bash script to catch run-away imaps:

Code: Select all

#!/bin/bash

sxbin="/opt/scalix/bin"

prevUser=""

usage="t - calculates <T>otal # of runaway imap processes, '0' for none"
usage1="u - shows <U>ser list omstat and total imap processes"

while getopts "tuh" options
do
  case $options in
    t )
    let timap=`ps aux | grep imap41d | wc -l`
    let oimap=`${sxbin}/omstat -u all | wc -l`
    if [ $timap -gt $oimap  ]
        then
        let res=($timap - $oimap)
        echo $res
    else
        echo 0
    fi
    exit 0
    ;;
    u )
    ${sxbin}/omstat -u imap | while read line
      do
      user=${line%% / *}
      if [[ $prevUser != $user ]]
          then
          uid=${line#* 605}
              uid=${uid:0:2}
              let omimap=`${sxbin}/omstat -u all | grep "$user" | wc -l`
              let totalimap=`ps U "605"$uid | grep imap41 | wc -l`
              if [ $totalimap -gt $omimap  ]
                  then
                  echo ">>>" $user $omimap $totalimap
              else
                  echo "   " $user $omimap $totalimap
              fi
      fi
      prevUser=$user
    done
    exit 0
    ;;
    * ) echo $usage
        echo $usage1
        exit 1;;
  esac
done


any ideas? Thanks

[sg]

At our problem server I think 2 mailboxes were responsible for this problem. Everytime the user logged in, the swa session timed out while receiving the inbox. After killing the imap processes and forcing them to restart, everything seemed okay.

After an update to 11.3 and reindexing the 2 mailboxes the problem was solved.