I’m afraid I am going to need more info in order to help you get to the bottom of this. You’ll have to provide a coredump.
Before I go into explaining how, it may also be worth trying to reindex the Kaltura indeces.
This can be done by running:
Note that this will cause an outage so if you can reproduce the segfault on a testing ENV rather than your production one, you should.
After reindexing, if the issue does not go away, here is what you should do:
- disable Monit’s Sphinx configuration by removing this symlink:
- stop the Sphinx daemon with:
# /etc/init.d/kaltura-sphinx stop
- make sure no orphan searchd procs are running with:
# ps fax|grep searchd
- run sphinx with GDB like so:
# gdb --args /opt/kaltura/sphinx/bin/searchd --config /opt/kaltura/app/configurations/sphinx/kaltura.conf --console
you’ll probably want to do that using screen, see:
When it crushes, in the GDB console, you will see the stacktrace and you’ll also be able to save generate a coredump file with:
The output will tell you the file’s name, for instance:
Saved corefile core.18423
Then attach this file so I can take a look.