Litespeed crashed

Discussion in 'Bug Reports' started by wanah, Nov 27, 2013.

  1. wanah

    wanah Well-Known Member

    Litespeed stoped responding today and I had to restart it. Some virtual hosts were still responding (ones with their own IP), others not.

    In my error_log file I had this at the moment it stoped responding :

    *** glibc detected *** litespeed (lshttpd): double free or corruption (out): 0x00000000044a39c0 ***
    2013-11-27 12:30:08.070 [STDERR] *** glibc detected *** litespeed (lshttpd): corrupted double-linked list: 0x00000000014df440 ***
    What is this ?

  2. wanah

    wanah Well-Known Member

    Hello, any news ? This seems like a serious bug and I'm not sure if you have even read my issue.
  3. NiteWave

    NiteWave Administrator

    if you can get core dump, then it'll help to identify the issue.

    #yum install gdb
    #ulimit -c unlimited
    #stop / start lsws

    see if you can capture a core dump if litespeed process when the issue come again.
  4. wanah

    wanah Well-Known Member


    I've just had the same crash.

    Here is the error message :

    gdb was installed some time ago.

    I've just run ulimit -c unlimited but I don't belive this is related to getting a coredump.

    Where would the coredump be created ? What option do I need to set in litespeed to allow coredumps ?

    Should I force an update of litespeed, have there been any fixes recently without changing the number that could fix this issue ?
  5. mistwang

    mistwang LiteSpeed Staff

    Yes, you should force update, if it is still happening, we will investigate.
  6. wanah

    wanah Well-Known Member


    It doesn't happen very often and so far hasn't happened during the night so I haven't been woken up yet by it… !

    Last time it happend I removed Cloudlinux's process limits to only leave entry limits as in Worker mode each user get's one entry process so I have enabled forked process limits. But I disabled this the previous time and as it happend again it's sign that it's not a process limit, I was getting 503 restarts when a user hit his process limit and disableling the CL process limits fixed this.

    I've forced litespeed to reinstall and am recompiling PHP to get the latest LSAPI version in case this was updated too without changing the numbers.

    It would be much easier if there could be some form of subnumbers to know if a newer version was available or not, (a build number and a changelog per build number or something) maybe just in your documentation so that users could see what issues/bugs were fixed and have an idea if an update could fix their issue.

Share This Page