Hi!

Stefan Fritsch:

Some processes are in "stopping" state but are not dying. They accumulate until the scoreboard is full.

First you should verify with ps if the processes with the PIDs in the "(old gen)" lines still exist. If not, it is a bug in apache itself (probably in mpm_event).

They do indeed still exist in the process table. Just looking at the command lines doesn't tell me much as they are all apache2 forks/threads, but see below:

  # apachectl status 2>&1 |awk '/old gen/ { print "tr \"\\0\" \" \" < /proc/" $2 
"/cmdline; echo" }'  | sh
  /usr/sbin/apache2 -k start
  /usr/sbin/apache2 -k start
  /usr/sbin/apache2 -k start
  /usr/sbin/apache2 -k start
  /usr/sbin/apache2 -k start
  /usr/sbin/apache2 -k start

If they do, one needs to find out why they are not dying. This is likely the fault of some module.

The server is running Trac with its default configuration, there might be something that has changed there?

  # dpkg -l trac
  Desired=Unknown/Install/Remove/Purge/Hold
  | Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Tri
  |/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
  ||/ Name           Version      Architecture Description
  +++-==============-============-============-===========================
  ii  trac           1.6-2        all          Enhanced wiki and issue tra

The processes do indeed seem to be from Trac, here is one example:

  # ls -l /proc/210869/fd
  total 0
  lr-x------ 1 root root 64 dec 14 09:19 0 -> /dev/null
  l-wx------ 1 root root 64 dec 14 09:19 1 -> /dev/null
  lrwx------ 1 root root 64 dec 14 09:19 10 -> 'anon_inode:[eventpoll]'
  lr-x------ 1 root root 64 dec 14 09:19 11 -> 'pipe:[35832724]'
  l-wx------ 1 root root 64 dec 14 09:19 12 -> 'pipe:[35832724]'
  l-wx------ 1 root root 64 dec 14 09:19 14 -> /srv/trac/log/trac.log
  lrwx------ 1 root root 64 dec 14 09:19 18 -> /srv/trac/db/trac.db
  l-wx------ 1 root root 64 dec 14 09:19 2 -> '/var/log/apache2/error.log.1 
(deleted)'
  lrwx------ 1 root root 64 dec 14 09:19 21 -> /srv/trac/db/trac.db
  lrwx------ 1 root root 64 dec 14 09:19 3 -> 'socket:[35833027]'
  lr-x------ 1 root root 64 dec 14 09:19 5 -> 'pipe:[35832707]'
  l-wx------ 1 root root 64 dec 14 09:19 6 -> 'pipe:[35832707]'
  l-wx------ 1 root root 64 dec 14 09:19 7 -> 
/var/log/apache2/other_vhosts_access.log
  l-wx------ 1 root root 64 dec 14 09:19 8 -> '/var/log/apache2/access.log.1 
(deleted)'
  l-wx------ 1 root root 64 dec 14 09:19 9 -> '/var/log/apache2/access.log.1 
(deleted)'

Should I reassign the bug to Trac instead?

--
\\// Peter - http://www.softwolves.pp.se/

Reply via email to