<HTML><HEAD></HEAD>
<BODY dir=ltr>
<DIV dir=ltr>
<DIV style="FONT-FAMILY: 'Calibri'; COLOR: #000000; FONT-SIZE: 12pt">
<DIV>Greetings,</DIV>
<DIV> </DIV>
<DIV>When I first setup Koha, I was using a 3.6.3 tarball on Ubuntu with 512MB
of memory. I discovered that apache was leaking about 1MB per transaction in
Koha. This caused us a pain when indexing/reindexing. At that time, we couldn’t
even complete a full reindex with apache and MySQL running.</DIV>
<DIV> </DIV>
<DIV>So, if you seem to be running out of memory after 24 hours, I would suggest
a cronjob that restarts apache daily at an off hours time, if you are unable to
increase the amount of memory you have. The OS/Apache eventually recover the
leaks without intervention if you wait long enough, and/or are able to determine
better apache settings.</DIV>
<DIV> </DIV>
<DIV>You can use "free -m" to determine the total amount of memory, free memory,
and swap space that you have in megabytes. The size of your catalogue and the
number of transactions (check in/check out/clicking a button on a Koha page) are
usually used to determine the amount of disk space and memory you should be
using.</DIV>
<DIV> </DIV>
<DIV>It should also be noted that Zebra is a memory hog. This is why my
*temporary* solution was turning off apache, doing the reindexing, turning
apache back on, and restarting the zebra indexing daemon.</DIV>
<DIV> </DIV>
<DIV>We are currently running 3.8.6 off of packages now. When we increased the
memory to 1GB, this was no longer required as memory management stabilized. We
do not have high traffic on our site (yet), but it is growing. And as it grows,
this minimum of 1GB will likely need to be increased. It should also be noted we
have no swap space.</DIV>
<DIV> </DIV>
<DIV>GPML,</DIV>
<DIV>Mark Tompsett</DIV></DIV></DIV></BODY></HTML>