1. Please take a little time for this simple survey! Thank you for participating!
    Dismiss Notice
  2. Dear Pleskians, please read this carefully! New attachments and other rules Thank you!
    Dismiss Notice
  3. Dear Pleskians, I really hope that you will share your opinion in this Special topic for chatter about Plesk in the Clouds. Thank you!
    Dismiss Notice

7.5.3 upgrade on Suse 9.1

Discussion in 'Plesk for Linux - 8.x and Older' started by phorium, May 9, 2005.

  1. phorium

    phorium Guest

    0
     
    Just ran an upgrade on our development server to test out the new release. The upgrade completed successfully, but Apache wouldn't return anything but a blank page. Looking in the Services menu both Apache and Qmail had been turned off - manually restarting them worked fine... but Apache was still returning blank pages.

    To make a long story short, removing "python" from the Apache Modules list in /etc/sysconfig/apache2 brought Apache back up - otherwise it would segfault and return nothing (even though there were no errors returned on restart).

    So 7.5.3 is running fine on our server, but without the "python" or "webapp" modules enabled. (The exact same thing happened with the 7.5.2 upgrade - the only thing that stopped Apache from segfaulting was by disabling 'webapp').

    That being said, we're looking forward to 7.5.4 and trying to figure out which Apache Module has to be disabled to make that release run.

    We're running Apache 2.0.53 and PHP 4.3.10
     
  2. Linulex

    Linulex Regular Pleskian

    33
    80%
    Joined:
    Aug 4, 2001
    Messages:
    426
    Likes Received:
    61
  3. phorium

    phorium Guest

    0
     
    It is segfault 11, but we get the same results from two different servers with completely different architectures. I have a hard time believing that memory has anything to do with Apache modules being enabled or disabled, but I'll try out the tests you posted. Thanks for the tip!!
     
  4. Linulex

    Linulex Regular Pleskian

    33
    80%
    Joined:
    Aug 4, 2001
    Messages:
    426
    Likes Received:
    61
    hardware errors, specialy memory errors, can have this result.

    Lets say that both setups without the plug-in take 256000byte per child proccess and with plug-in they take 365000 byte per child.

    Then this means that the upper regions of the memory are not adressed without the plug-in actvated, but are addressed with the plug-in activated.

    It does not mather wath plug-in this is then, the only thing that mathers is if the specific memory areas are addressed or not.

    It also depends on what the plug-in tries to do with the memory. Some errors in memory only show with a mysql inodb. Not in apache or isam mysql.

    Thats the nice thing of memtest86: it does every possible action on every possible part on every bit of the memory.

    We have about 40 servers with all together about 300 memory modules and we test all off them very extensive, and even then we have still some servers that produce some faulty mem in some surcomstanses.

    memory these days is a quality product, but still it is a very cheap mass product so you have to calculate that 1 in 20 in faulty.

    regards
    Jan
     
Loading...