[06:36:09] Missing retro tomorrow, need to drop housemate's dog off during the retro window [11:04:08] lunch [14:23:21] Guillaume and I were talking and there was a suggestion to repurpose today's retro slot to continue the roadmap planning piece. I have the responses in a spreadsheet that help with visualizing things, just have one last double-ranked entry and then it's ready to show, and I think it'll help us be able to anchor the discussion of top contenders for the nearer term (thinking like mid-Q2 and beyond, after other work winds down). [14:24:36] +1, makes sense [14:54:53] dcausse: after the meeting block today (finishes in about 4 hours), i'll take a look at updates to the patches up for code review. [14:55:08] dr0ptp4kt: thanks! [15:06:33] \o [16:13:17] pfischer: point A here is about seeing if we can limit the ip range to something narrower than just any kubernetes ip: https://phabricator.wikimedia.org/T357353#9716932 [16:17:22] back around now [17:45:16] i'm thinking for the next Google Form to reinstate the related pages item, given the note in the Etherpad on what happens with IRV. i think a semi-conclusion (of mine) from the meeting was that things would stay in a consultative mode and be fairly light touch. but, again, given the note about it, if it's the thing that folks think would be worth pursuit for optimization (irrespective of UI changes) knowing relative interest helps [17:47:51] i think i'm seeing netsplit artifacts here...so i'm going to post that last thing on the low-volume internal slack [17:57:43] got new laptop mostly up and working...just one minor visual glitch. For some reason it's rendering a partially transparent set of lines over ~20 pixels on the left and right of my second monitor. Where i usually have text. They seem to be visual indicators for window drop zones, but i'm not sure why they are always rendered :S [17:57:58] makes the first two letters in my shell look funny and badly rendered [18:37:39] doh, i think the problem is the monitor. The difference is previously there was a menu bar covering that area. On closer visual inspection there are three vertical bands against a single color screen :( [19:07:32] Thanks for the notes for the discussion! And thanks Trey314159 for the re-ranking. That's interesting... [20:32:14] titlesuggest is promoting bad indices :( T363521 [20:32:14] T363521: Completion suggester can promote a bad build - https://phabricator.wikimedia.org/T363521 [20:37:06] * ebernhardson guess bad error handling on failing requests, but hard to say [21:03:02] huh, upstream error% on the envoy appserver->search-eqiad went from ~0% to 1% yeserday at 21:40 [21:26:49] wow, so the error handling of an Elastica scroll is: perform next request. set scrollId to null. It the request worked, set the scrollId to the returned scrollId. [21:26:58] What happens if the error failed? Your foreach loop exits without a warning [22:27:44] Hooray, my bouncer is sorta fixed [22:27:56] see ya tomorrow! [22:39:18] making 60 requests over 60 seconds from mwmaint1002 to search.svc.eqiad.wmnet, 3 failed with no route to host :S [22:39:39] just requesting the banner