lsphp suddenly stopped working - returned error 500 on some websites

Discussion in 'General' started by iso99, May 16, 2014.

  1. iso99

    iso99 Member

    I don't know when but it's within 12 hours today, lsphp suddenly stopped working and configurations were reset back to the default config.

    lsphp5 Run on Startup went back to 'No' from SuEXEC Daemon
    A lot of abort error code 4 on logs
    Affected sites mostly Wordpress blogs

    I just restarted the server and everything seems to be fine for now.

    I have a working zend opcache with 1.5gb allocated memory. Just to be safe, I doubled lsphp5's soft and hard memory limits before I restarted the server (LSWS restart).

    --- update ---

    stderror log:

    2014-05-16 09:43:52.467 [STDERR] Fri May 16 09:43:52 2014 (545034): Fatal Error Unable to allocate shared memory segment of 1610612736 bytes: mmap: Cannot allocate memory (12)

    I guess this was caused by the opcache. Should I increase opcache's memory from 1.5gb to 3gb? Should lsphp5's memory limits be always bigger than the opcache's?
    Last edited: May 16, 2014
  2. NiteWave

    NiteWave Administrator

    >Should lsphp5's memory limits be always bigger than the opcache's?

    >Should I increase opcache's memory from 1.5gb to 3gb?
    I'd not recommend personally. there is tool to check opcode cache's status, most likely there is plenty of free space within the 1.5G memory. from the past experience, I have a expression that opcode cache can't be too large. I guess it's because after long time run, the fragment is becoming more and more, and opcode cache can't handle that well. to prevent fragment increasing, a simple workaround is to restart the web server after some time, e.g., every 1 or 2 or 4 ... hours. I have noticed this trick working well on a user's production server for a long time.
  3. iso99

    iso99 Member

    I don't think SuExec Daemon suits me, I am now trying ProcessGroup and lowered opcache size to 128M.

Share This Page