To provide value to devs and consumers, it's crucial that our stats gathering/processing systems are always up and running. Things can and will go wrong (bug 781795). Let's provide alerts when data is inconsistent. We can start by adding simple data checks. For example, adblock-plus consistently has a lot of downloads so if we pinged it each day we could send an alert in case the downloads went to zero. Maybe use nagios for this?
A fine idea, but I'm marking [needs spec] because this is something we should do well and make sure it works for all the use cases. It's a spec we can write, I just want to go in with a plan.
For adblockplus stats, they come from metrics and they'll tell us when there's a problem with their cron jobs.
This is a very valid concern, especially after moving stats out to monolith: more moving pieces, more chances to fail. Needs a plan of attack.
I second (third?) the need for this.