[19:50:30] Hi dbrant. I've noticed crashes with the latest bits (dev on master and alpha). It doesn't happen with the current prod version. https://phabricator.wikimedia.org/T181668 [19:51:02] bearND: thanks! will look [19:53:21] bearND: dbrant: i know why this is happening. I dont have a check for on-thi-day entry not having related pages. Fixing it in the short onthisday patch that I am making. [19:53:52] dbrant: sharvani: Here's a paste for the announcement text if that helps (also linked from the fundraising task). The only change you have to make in there is to change the start_time on the first one to something earlier than now. https://phabricator.wikimedia.org/P6393 [19:54:37] (the first entry is for Android in the US) [20:41:48] mdholloway: I'll make another change to the announcement, to move the date later, since there is no confirmation from stakeholders yet. [20:42:02] bearND: ok [20:51:11] mdholloway: ready [20:55:35] bearND: merged. are we trying to get this deployed this afternoon? [20:55:54] (this === the announcement patch) [20:58:44] mdholloway: hmm, I guess we'll do it tomorrow. [20:59:00] mdholloway: Not sure why the build failed [21:01:24] bearND: might have been transitory. it doesn't happen often but maybe once in the past couple weeks i've had several tests throw 500s on a test build, then pass fine on a recheck... [21:02:04] tests are looking fine locally [21:02:55] well, there's two in a row... :( [21:03:06] mdholloway: Did it automatically retry or did you do anything? [21:03:51] bearND: i think the gate-and-submit job ran again when you responded to tgr's comment while keeping the +2 [21:06:03] that's probably it [23:31:02] mdholloway: one final tweak https://gerrit.wikimedia.org/r/#/c/394205/, and then I'll probably deploy this one tonight. //cc:coreyfloyd [23:34:39] or tgr, if you are around to +2 a simple patch [23:34:54] bearND: i've got it [23:35:01] cool, thanks! [23:36:11] bearND: fyi, i've just been manually submitting these since without waiting for CI to pass since they look very safe (and the test failures on the other branch look specific to the media endpoint). [23:36:35] the CI backlog is >3hrs at this point [23:36:44] mdholloway: ah, that explains why my patches went through and your didn't. ;) [23:36:48] yours [23:36:57] thanks! [23:37:02] yw! [23:48:28] I think hand-submitting is generally discouraged when there is a large backlog as it will somehow interrupt tests and make things even slower [23:48:51] (I don't really understand how it works, just heard that being said in the past) [23:49:41] although if it's for an urgent deploy then can't really be helped [23:53:49] tgr: bummer. yeah, probably still would have done it in this case but that's good to know. [23:54:09] * mdholloway goes digging through the gerrit/zuul docs [23:54:34] James_F can probably tell if that's still the case [23:56:17] although now that I think about it that only makes sense for commits to the same repo which caused the congestion (which is presumably mediawiki/core) so don't mind me :) [23:58:17] hey, I'm investigating login failed notifications coming from the Android app - how often is it supposed to call action=clientlogin?