2.4.5 -> 2.6.2 slow course restore

2.4.5 -> 2.6.2 slow course restore

by Rikard E -
Number of replies: 0

We recently upgraded from 2.4.5 to 2.6.2 and now we have started to see some instabilities in 2.6.

At numerous occasions one of the moodle nodes spawns apache processes so that the apache max clients value is reached. Server load hits the sky.
After a while everything goes back to normal. Our other cluster nodes are still reachable during this.

This is what I know for sure right now:

Course backup and restore seems to have performance problems in Moodle 2.6 and as far as I can see this is what causes our nodes to reach a critical load sometimes. I can see that the problem is triggered when our teachers do course backup or course restore.

I have seen some of these errors in the apache error log:

[error] Potential coding error - existing temptables found when disposing database. Must be dropped!, referer: https://mymoodlesite.com/moodle/backup/restore.php
[error] Potential coding error - existing temptables found when disposing database. Must be dropped!, referer: https://mymoodlesite.com/moodle/backup/backup.php

With Debug:
Mon Aug 18 13:57:09 2014] [error] [client 10.2.12.34] instantiating restore controller ee1d62e087132914ba3bc59e64017592, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:09 2014] [error] [client 10.2.12.34] setting controller status to 100, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:09 2014] [error] [client 10.2.12.34] loading backup info, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:09 2014] [error] [client 10.2.12.34] loading controller plan, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] setting controller status to 300, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] checking plan security, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] setting controller status to 500, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] checking plan security, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] checking plan security, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] checking plan security, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] saving controller to db, referer: https://mymoodlesite.com/moodle/backup/restore.php
[Mon Aug 18 13:57:10 2014] [error] [client 10.2.12.34] calculating controller checksum eb17b6a2b9a8032a55bae2f074d8b14b, referer: https://mymoodlesite.com/moodle/backup/restore.php



Now I would like to know what you all think about this.

Do you think that this problem will be solved if we configure our moodle cluster to use memcached?

These links are probably related to what I have experienced:
https://moodle.org/mod/forum/discuss.php?d=244817
https://tracker.moodle.org/browse/MDL-44813
https://tracker.moodle.org/browse/MDL-40579




-------

Apache settings:

Timeout 60
KeepAlive On
MaxKeepAliveRequests 100
KeepAliveTimeout 5

<IfModule mpm_prefork_module>
    StartServers          5
    MinSpareServers       5
    MaxSpareServers      10
    MaxClients          70
    MaxRequestsPerChild   30
</IfModule>

-------

Environment:

LAMP (Ubuntu 12.04)
Three application nodes with four cores each and 8G of RAM.
MySQL server with 4 cores and 12G of RAM
NFS share for moodledata. Standard cache settings
No. users: ~10000
No. courses: ~1200

Average of ratings: -