We don’t have tremendous traffic on a couple of sites (about 5000 visits per day with 25000 views) but use e-commerce tracking excensively (5 categories per view and two custom variables per visit). We also set datatable_archiving_maximum_rows_xxx to 50000 to have meaningful stats which tie somewhat in.
Piwik runs of an EC2 instance and has a seprate dedicated MySQL instance. Since the year archiving takes more than 1.5 GB of memory, we run this on a physical cheap high memory commodity server connecting to the database remotely, which obviously slows things down due to latency.
Question:
[ol]
[li] How can I split the archiving so that monthly and yearly figures only get computed once or twice a day while keeping the archive cron for the day and the week running every hour to get fresh results on the screen? Yearly archiving also takes too long to do every hour.
[/li][li] Could we run the “quick” hourly archiving on one server (low latency to the database and not very resource hungry) and the resource intensive monthly/yearly on another one? Should this be possible, how could I prevent scheduled reports to be triggered by both crons?
[/li][/ol]
Current Piwik Options are
Allow Piwik archiving to trigger when reports are viewed from the browser = NO
Reports for today (or any other Date Range including today) will be processed at most every = 3600
archive.php runs every hour
[General]
datatable_archiving_maximum_rows_referers = 50000
datatable_archiving_maximum_rows_subtable_referers = 50000
datatable_archiving_maximum_rows_actions = 50000
datatable_archiving_maximum_rows_subtable_actions = 50000
enable_processing_unique_visitors_day = 1
enable_processing_unique_visitors_week = 1
enable_processing_unique_visitors_month = 1
enable_processing_unique_visitors_year = 0
enable_processing_unique_visitors_range = 0
And archive.php runs only with --url parameter
Many thanks,
Robert