Yes, /opt/kaltura/web/logs/ should exist on your NFS mount. It should be mounted on all front servers [and batch as well, but only front nodes are relevant for DWH].
/opt/kaltura/web/logs belongs to the kaltura-dwh package as can be seen when running:
# rpm -qf /opt/kaltura/web/logs/
or for Deb:
# dpkg -S /opt/kaltura/web/logs/
My guess would be that you installed kaltura-dwh before mounting the NFS volume, hence, /opt/kaltura/web/logs was created on the local disk on the DWH machine, instead of on the NFS volume.
That’s easy enough to correct, simply mount it on the DWH machine and create the directory.
Just so that you further understand the flow of things:
- When a user hits play, an API request similar to the below [service=stats&action=collect], is made by the player:
- This in turn is recorded in the Apache access log for the Kaltura VHost [/opt/kaltura/log/kaltura_apache_access.log]
- The access logs from each front node are then rotated to /opt/kaltura/web/logs using /etc/logrotate.d/kaltura_apache, in particular:
mv /opt/kaltura/log/kaltura_apache_access.log-`/bin/date +%Y%m%d`.gz /opt/kaltura/web/logs/`hostname`-kaltura_apache_access.log-`/bin/date +%Y%m%d-%H`.gz
service httpd reload
su root kaltura
- On the DWH machine, the relevant scripts are run by crond, because of this file: /etc/cron.d/kaltura-dwh
- As a first step, the following config file is looked at /opt/kaltura/dwh/.kettle/kettle.properties, in it, the path to the log dir and the pattern to look for are defined like so:
EventsLogsDir = /opt/kaltura/web/logs
EventsWildcard = .*kaltura.*_access.*.log-.*
- Make sure /opt/kaltura/web/logs exist on the NFS
- Make sure the volume is mounted on all front and batch machines and that at least /opt/kaltura/web/logs is mounted on the DWH machine [it does not need the other dirs though you can mount the entire volume]
- Make sure /etc/logrotate.d/kaltura_apache exists on all front machines and the access log files are rotated
Note that analytics are not updated in real time but rather processed by running the cron jobs defined here /etc/cron.d/kaltura-dwh
If, after 24 hours after validating all the above, you still don’t see stats, please follow the checks here:
And report back with the results.