[12:14:09] <slyngs>	 Going through the final planning for the DC switch over, and it's a little early for jasmine_, is there any reason why we're not adding the scheduled maintenance to the statuspage until like an hour before. Can't I just do it like now?
[13:18:09] <Raine>	 slyngs: I don't see why not, other than "what if the plans change", which I don't think is very likely at this point
[13:19:32] <slyngs>	 I'll add it, we can remove it, if someone has a change or hearts
[13:31:06] <slyngs>	 Raine: If you have the time a sanity check on my note would be appreciated 🙂 https://www.wikimediastatus.net/ 
[14:02:27] <Raine>	 slyngs: lgtm :-) 
[14:02:40] <slyngs>	 Thank you :-)
[15:12:19] <_joe_>	 slyngs: re: switchover, I am slightly worried by the request rates to mw-web
[15:12:32] <_joe_>	 I am not sure we have enough capacity to run it single-dc as-is
[15:12:54] <_joe_>	 actually, I'm pretty sure we haven't, we will surely need to check how much we can expand mw-web in codfw to be able to do it
[15:13:05] <_joe_>	 that's the only potential real blocker to the switchover
[15:17:12] <slyngs>	 The upscaling is a lot https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/1332725/3/helmfile.d/services/mw-web/values.yaml
[15:46:57] <slyngs>	 I'll recheck the numbers tomorrow and pick a week in september and see if they even worse
[21:18:36] <jasmine_>	 s.lyngs: here's a paste with the other sept. estimates in case it saves you the time to rerun the projections script, [0] the particular week we're basing the capacity projections on includes the dolly parton incident but as Scott noted [1] that's only a small percentage (less than 1% of samples over 1m the course of a full week) and appears to be minimal impact in the grand scheme of things
[21:18:36] <jasmine_>	 [0] https://phabricator.wikimedia.org/P96396 
[21:18:36] <jasmine_>	 [1] https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/1332725/comments/0bd5909a_0c38d6c8