[Koha-zebra] Recent experiences with Zebra...

[Mark Tompsett]

Greetings,

I am sharing this in hopes that this helps people who are frustrated trying to figure out why Zebra indexing is not working for them.

When I initially did a standard install, I ran the full reindex as root. This was a mistake! This created a whole bunch of files with root.root as the owner.group in the subdirectories of the /var/lib/koha/zebradb/ directory. I corrected this problem with:

adminuser$ sudo chown –R –v koha.koha /var/lib/koha/zebradb

This, however, did not solve the problem. It was then suggested that I restart the koha-zebra-daemon (Thanks to jcamins):

adminuser$ sudo service koha-zebra-daemon restart

This corrected the problem temporarily, and everything was running as koha (the crontab to do a full reindex, etc.)

adminuser$ sudo cat /etc/cron.d/koha
[it had ‘koha’ in the ‘run as this user’ column]
adminuser$ ps aux | grep zebra
koha     11635  0.0  1.3  99632  7192 ?        S    10:40   0:00 /usr/bin/zebrasrv -v none,fatal,warn -f /etc/koha/koha-conf.xml
koha     12563  0.0  0.1  18296   584 ?        Ss   Jun29   0:00 daemon --name=koha-zebra-ctl.kohadata --errlog=/var/log/koha/koha-zebradaemon.err --stdout=/var/log/koha/koha-zebradaemon.log --output=/var/log/koha/koha-zebradaemon-output.log --verbose=1 --respawn --delay=30 -- /usr/bin/zebrasrv -v none,fatal,warn -f /etc/koha/koha-conf.xml
adminuser 12722  0.0  0.1   6160   680 pts/0    S+   11:49   0:00 grep --color=auto zebra

However, our VM was running low on memory (512MB total) and full indexes were triggering out of memory problems. I found that stopping the apache server, doing the full reindex, restarting the koha-zebra-daemon and then starting the apache server allowed for a successful reindex and searching to work properly.

adminuser$ sudo service apache2 stop
adminuser$ sudo su – koha
koha$ cd /usr/share/koha/bin/migration_tools
koha$ echo $KOHA_CONF
[this better be set correctly]
koha$ echo $PERL5LIB
[this better be set correctly]
koha$ ./rebuild_zebra –b –a –r –v
[long indexing/exporting output which could take several hours – DO THIS AFTER HOURS!! – unless it is currently broken]
koha$ exit
adminuser$ sudo service apache2 start

How did I know we were having memory problems? I opened a secondary window and watched while reindexing, and I noticed some errors in the logs
adminuser$ free –m –s 1
[output snipped]
^C
adminuser$ cd /var/log/koha
adminuser$ grep memory *
[you’ll recognize the problem when you see the output]

If you don’t have swap and can turn it on, do so. This is left as an exercise for sysadmin administrators.

I don’t know why I have to restart the koha-zebra-daemon after a full-reindex, but I’m hoping to get a memory bump to 1GB from 512MB, and then not have to stop apache. We sadly, can not turn on swap.

Anyways, I hope this helps someone somewhere. Also, if someone can explain why I need to restart the koha-zebra-daemon after a full reindex even as koha, that would be appreciated too. I was thinking that perhaps the daemon has filehandles open when the files are deleted, and so the filehandles it has are stale?

GPML,
Mark Tompsett
-------------- next part --------------
An HTML attachment was scrubbed...
URL: </pipermail/koha-zebra/attachments/20120703/7282e2d9/attachment.htm>