2 Piwik VMware Servers behind a loadbalancer each 2 CPU 2,5GHz 8GB RAM (ESX 4.1.0)
running RHEL 6.5
RemoteDB MySQL 5.5.36
PHP-memlimit (php.ini and httpd-conf) 4GB
archiving via cronjob, 2000sec valid
keep logs for 42days, no removing of reports
DB size: 14,3 GB
9 sites: 8 small ones with less than 2500 per day,
one with up to 100000 visits.
since we ar running two instances, we run a modified archive.sh which takes
site an period for run to as parameter. so we can separate archiving of the
small sites from the big one and dont have to generate weekly, month and year-reports
dayly. and we can distribute the work to both servers.
report generation for the big site takes up to 2 or 3 hours, runns of the small sites some
minutes up to an 1/2 hour
will upgrade to 2.2.x the next days after solving the problem with the archive job:
in v 2.2.0 archive.sh and archive.php are deprecated - usage of console core:archive
will im be able to deploy the archive job to both servers with this new method?
what will happen if i still use the old archive.sh?
whats about transitions: the piwik-documentation saying, that the basic log-data can be deleted after
archiving/report generation since ‘all’ what is shown to the user is based on the reports - but this is
NOT TRUE for transitions - if i try to get transitions for a page for a day which is no longer in the
basic visitor-log i get a no data error …
should ALL functions in the user-interface be usable also for big sites? simple example:
i can choose the report period - day, week, month, year or free defined interval. all what is needed for the
first four choices is preprocessed by the archiving job, but not (certainly) for the last … so it may take
a very long time to get the result - maybe some hours - thats hardly ‘real time’ …
transitions are also unuseable for the big site! and there are many more examples …
i don’t ever have seen something like 'full realtime function of piwik is possible only for smal or mid-sized sites,
but NOT for big sites with many visits …'
so, i have the problem that some customers which knows the piwik promotion statements expecting full function
of the user-interface also for such big sites …
should piwik really offer this? or is the limited usability obligatory for such uses?
is there anywhere a statement which specifying the conditions for full realtime usability?
segmentation is a way to define some own reports - but for the user its also unuseable since he can define
the segment only in realtime mode and for his own … for the admin there is also the option to define the
segment preprocessed and visible for others too - so the segment based reports are also generated in the
background. is there a way to offer this options also to a normal user? or at least to a user which have
the admin-right for a site?
it would be very very helpfull if the normal user is able to define preprocessed segment-reports too …
so he will be able to use at least the basic functions for big sites …