Was all file access to the Netapp impaired or was it just this ajax/service.php? I am rather curious as it would confuse the heck out of me for just this to act up if your file storage was not acting up to par.
We had the same issue with long wait time/timeouts on our server (3.1.5+) in a lab environment without routing to the internet.
We found out that the problem was that we hadn't used the proxyserver in Moodle (in Server -> HTTP). After configuring this the problem was solved! The Ajax call via XML-RPC in the service.php was the bottleneck in this case.
We have the same problem: often, the 'service.php' called in Ajax is very slow: delay = 30 seconds..
We did add a proxy server but the problem still appears.
Is a cache engine (ie: memcached) really required ? Is the 'service' result cached in Moodle?
in the admin (Server -> HTTP), what the most powerful method, "HTTP" or "SOCKS5"?
Thanks for your help.
Thanks for your infirmations.
It seems we have a speed problem on the NFS mount for moodledata I will try to store sessions in memcached.
Have a good day.
Benoit.
Benoit we use Memcached and local caching on the webtier now and that's a huge performance improvement.
Don't cache on your (slow) NFS share!
Hello Alain,
We do that too: Memcached for sessions, Redis for application cache, in local on each node (so, no SPOF).
"Don't cache on your (slow) NFS share!" I agree with You!
And check the scheduled tasks: some plugins have bad scheduled huge tasks (ie: use_stats) and must run only the night.
Best regards.
We are experiencing the same situation albeit an upgrade to 3.2.4+ build from our 3.0.10+.
Running on RHEL 6.6 64-BIT on php 5.6.25 (Zend OPcache) + MySQL 5.5.41
No clustering, failover, no proxy services, etc... nothing fancy and noticed that the MySQL spiked to 67.5% cpu usage.
On VMWare with 16 GB of ram...
what else could be causing the bottleneck?
Thanks
What values does your MySQL configuration file have for innodb_buffer_pool_size and friends?
https://dev.mysql.com/doc/refman/5.5/en/innodb-buffer-pool.html
Also noticed the very high amount of http traffic so much so that no one can log in.
My moodledata is pretty big .. about 925 GB and ram is 16 GB and db size is about 275 GB
In mysql query...
mysql> SHOW VARIABLES LIKE '%innodb_buffer_pool_size%';
+-------------------------+-----------+
| Variable_name | Value |
+-------------------------+-----------+
| innodb_buffer_pool_size | 134217728 |
In my.cnf (I don't have that value)
[mysqld]
#datadir=/opt/rh/mysql55/root/var/lib/mysql
datadir=/srv/mysql
socket=/var/lib/mysql/mysql.sock
# Disabling symbolic-links is recommended to prevent assorted security risks
symbolic-links=0
max_allowed_packet=1G
wait_timeout = 31536000
connect_timeout=100
net_read_timeout=3600
net_write_timeout=3600
[mysqldump]
max_allowed_packet=1G
[mysqld_safe]
log-error=/var/log/mysql55-mysqld.log
pid-file=/opt/rh/mysql55/root/var/run/mysqld/mysqld.pid
Should I put in the "innodb_buffer_pool" in my my.cnf to say, 12 GB?
innodb_buffer_pool = 12G;
Jerry how is your caching configured and your temp directory?