Sounds great, Sean! The following tables can be set to keeping 40 days of data:

MediaViewer_6054199
MediaViewer_6055641
MediaViewer_6066908
MediaViewer_6636420
MediaViewer_7670440
MediaViewer_8245578
MediaViewer_8572637
MediaViewer_8935662
MediaViewer_9792855
MediaViewer_9989959
MultimediaViewerAttribution_9758179
MultimediaViewerDimensions_10014238
MultimediaViewerDuration_8318615
MultimediaViewerDuration_8572641
MultimediaViewerNetworkPerformance_7393226
MultimediaViewerNetworkPerformance_7488625
MultimediaViewerNetworkPerformance_7917896

There's a good chance that some of the older ones will end up being empty, in which case they can be safely dropped.



On Mon, Oct 6, 2014 at 5:22 PM, Sean Pringle <springle@wikimedia.org> wrote:
On Fri, Oct 3, 2014 at 2:28 AM, Gilles Dubuc <gilles@wikimedia.org> wrote:
We can trim down our team (multimedia)'s tables considerably by getting rid of data older than 30 days. This could even be done by a daily cron. How would we go about doing that? Should we be the ones taking care of it? I'm not sure that the DB credentials I currently have can delete content.

We can automate purging using the MariaDB using the Event Scheduler[1] if you guys want a once-off-set-and-forget solution. Eg:

CREATE TABLE purge_schedule (
  table_name varchar(100) NOT NULL,
  days tinyint(3) unsigned NOT NULL
);

Then for each EL table you would do:

INSERT INTO purge_schedule VALUES ('MultimediaTiming_7193302', 30);

The rest would be left to me, or rather, to a couple of stored procedures :-)

_______________________________________________
Analytics mailing list
Analytics@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/analytics