[Koha] rebuild_zebra daemon locks memcached?

[Alvaro Cornejo]

Hi

Since our update to 19.11.09 we start having problems with kona. We
initially thought it was apache since without any notice, koha stop
responding.

When doing ps-ef |grep apache we got hundreds of entries even though we
have very low load. Max 3 users at a time. Restarting apache solved the
issue for a while

On the server we only have koha and mod_pagespeed. Boith using the same
instance of memcached. We disabled mod_pagespped and for a while, it looked
like it was the problem, but after a couple of days, the issue returned.

using netstat we found that there were lots of CLOSE_WAIT connections to
port 11211. So  we decided to configure two instances of memcached in two
different ports. One for koha and one for mod_pagespeed.

The issue of CLOSE_WAIT connections is less often now. However they are
almost always related to perl
.
# sudo netstat -tulpan |grep 11211
gives us
tcp        1      0 127.0.0.1:54368         127.0.0.1:11211
CLOSE_WAIT  1411/perl

And checking the perl process:
#ps -ef |grep 1411
bibliot+  1411  1407  0 Sep18 ?        00:01:44 /usr/bin/perl
/usr/share/koha/bin/migration_tools/rebuild_zebra.pl -daemon -sleep 5

Also, we noticed that, at anytime, memcached port has at least 10-15
connections split between memcached process and starman workers

Is that normal?

mod_pagespeed didn´t have any problem since it is configured to use an
independant memcached instance

Koha 19.11.09
Ubuntu bionic
memcached  1.5.6

Regards

Alvaro


|----------------------------------------------------------------------------------------|
 Stay safe / Cuídate/  Reste sécurisé
*7* Switch off as you go / Apaga lo que no usas /  Débranchez au fur et à
mesure.
 *q *Recycle always / Recicla siempre / Recyclez toujours
 P Print only if absolutely necessary / Imprime solo si es necesario /
Imprimez seulement si nécessaire