Skip to content

configuration: reduce # of DB reads substantially

Administrator requested to merge sg/conf-readless into master

Created by: slimsag

Previously, our configuration polling would read from the DB at a rate of about twice per 5s (once for critical config, and again for site config) per service using configuration. From a quick check, we have ~10 services polling configuration.

Bug originally reported by @mrnugget in Slack as this made it hard to debug other SQL queries going on.

Now we cache reads with a ttl of 2.5s which means we go from about 240 reads/min to just 24 reads/min

Merge request reports

Loading