I don't know enough about UMAPI to have an opinion, but I can explain my rationale for doing this with EventLogging. Systems that rely on the engineers that maintain them to report downtime tend to, um, underreport downtime. The outage is short, or the upgrade only takes a little bit longer than you anticipated, and it's easy to trick yourself into thinking the data loss is not significant. Knowing changes to service status are publicly broadcast helps me be conscientious and honest, so I like it.
Ori Livneh
On Monday, April 15, 2013 at 8:08 PM, Dario Taraborelli wrote:
to expand on Diederik's concern, eventlogging-alerts currently serves a dual purpose as a list for automatically generated alerts and a list for (human) announcements. I understand Erik's suggestion to refer mostly to the latter. It makes sense to have a channel to communicate scheduled downtimes or any major code change affecting the responses and I agree it doesn't need to be the same outlet as automated alerts.
On Apr 15, 2013, at 8:03 PM, Diederik van Liere <dvanliere@wikimedia.org (mailto:dvanliere@wikimedia.org)> wrote:
Hi Erik,
I have some hesitations; it makes it less clear who is supposed to troubleshoot UMAPI and probably nobody besides Ops can actually do anything. I do agree that we should think about setting up automatic monitoring and alerting, preferably using our existing Ganglia / Nagios infrastructure but I will leave it to Ottomata to comment on the technical implementations.
So yes to monitoring & alerts, not sure if a mailinglist is the best outlet. D
On Mon, Apr 15, 2013 at 10:01 PM, Erik Moeller <erik@wikimedia.org (mailto:erik@wikimedia.org)> wrote:
As Ori put the EventLogging service into production he created an eventlogging-alerts list to keep developers and users of EventLogging data informed about any service or data quality issues.
The user metrics API is quickly becoming relevant to multiple teams in the org who will likely also need to stay informed about similar kinds of issues, whether through automatic notifications or manual emails. Not all of them will want to be on analytics@, and not all folks on this list will care about the nitty-gritty.
Would it make sense to setup umapi-alerts to serve the same purpose as eventlogging-alerts, but for the user metrics API?
Cheers, Erik
-- Erik Möller VP of Engineering and Product Development, Wikimedia Foundation
Support Free Knowledge: https://wikimediafoundation.org/wiki/Donate
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics