[00:02:44] (PS2) Eileen: Update backoffice message template to allow receipt resends. [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312952 (https://phabricator.wikimedia.org/T137496) [01:08:03] (CR) Ejegg: "Sure, I'd be happy to chat about this. I agree it's probably a step backward to rely on the contribution ID linkage from the contribution " [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312563 (https://phabricator.wikimedia.org/T141477) (owner: Ejegg) [01:15:29] (CR) Ejegg: "Wait, can we not just return early from PayPal\Listener::execute if $request->getMethod() == 'POST' ? It looks like RequestHandler will ex" [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [02:17:17] (CR) Cdentinger: "> Wait, can we not just return early from PayPal\Listener::execute if" [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [02:41:36] thanks!! [09:53:11] fundraising-tech-ops: upgrade all frack servers to debian/jessie - https://phabricator.wikimedia.org/T146479#2670038 (Jgreen) [11:03:15] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, MediaWiki-extensions-CentralNotice, Operations, and 3 others: Banner not showing up on site - https://phabricator.wikimedia.org/T144952#2670227 (Pcoombe) >>! In T144952#2668547, @spatton wrote: > @awight our typical practice is to disab... [11:48:32] (CR) Daniel Kinzler: [C: 2] Remove old calls to WikiPage::doEdit() [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/310940 (https://phabricator.wikimedia.org/T145738) (owner: Reedy) [11:50:00] (Merged) jenkins-bot: Remove old calls to WikiPage::doEdit() [extensions/CentralNotice] - https://gerrit.wikimedia.org/r/310940 (https://phabricator.wikimedia.org/T145738) (owner: Reedy) [14:23:36] Fundraising-Backlog, MediaWiki-extensions-CentralNotice, Patch-For-Review, Technical-Debt: CentralNotice should not call WikiPage::doEdit() - https://phabricator.wikimedia.org/T145738#2670710 (enigmaeth) Open>Resolved a:enigmaeth [16:56:49] fr-tech: hehe maybe I picked a bad day to sync up, everyone seems to be |alt [16:56:57] ejegg: ohai! [16:57:36] hi awight ! [16:59:21] (CR) Ejegg: "I don't think we even need to call header() to end up with a 200 status, do we?" [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [16:59:49] I'm just taking some notes in the TT etherpad, iono if it's helpful to have some draft material in there or if it kills collaboration... [17:00:35] sure, it's nice to have an agenda [17:00:54] huh, etherpad autonumbering is all bad [17:01:07] aha [17:02:07] fr-tech: I'm in the hangout, if anyone wishes [17:05:40] https://hangouts.google.com/hangouts/_/wikimedia.org/fr-tech-talk?authuser=0 [17:11:48] let's IRC it then [17:12:05] Donno why it's like that [17:12:55] Um, so I think I see what you're talking about wrt some methods setting "completed" too early, e.g. PayPal [17:13:08] awight_: so, I think it's possible for the payments-init message to be consumed before the donation is imported [17:13:37] even for messages from amazon and astropay where there is a legit completed donation in that queue [17:13:40] Maybe we should also work on making payments-init statuses more granular [17:14:14] we already have a consistency issue between our init messages, and what PSPs think about our donations [17:14:34] awight_: I think we just need to send the correct status from the frontend (pending for pp) and send updates from the listener when we know it's changed statusa [17:15:00] That'll screw up all the reporting, though [17:15:00] since payments-init can understand updates now [17:15:32] oh? what reporting does it screw up? [17:15:53] I thought MBeat was relying on it, and maybe even campaign reporting [17:16:01] * awight looks in dread /srv/br [17:16:18] hmm. I think adyen reporting is already crappy for them then, since it's setting everything pending [17:16:44] in fact... let me just make smashpig send those update messages [17:16:51] one moment... [17:17:04] I think we need to tighten the definitions of each status [17:17:10] and possible have additional columns [17:21:23] sure was a lot of fail mail last night [17:21:27] missing db server? [17:21:47] what kind of additional columns? [17:23:51] yowza... [17:24:21] ops offsite ftw [17:24:49] It's just like our offsites... remember the fr-tech week in Paris? [17:24:50] j/k [17:25:33] :P [17:25:45] there is talk of one next year! [17:25:48] ejegg: yeah I'm not sure about columns, but we need to decode what the statuses mean [17:27:40] ejegg: So, ingenico has the problem you're describing in point (6)? [17:28:24] awight: ingenico messages could be consumed out of order (especially if we pause the donations QC) [17:28:35] but for that gateway we don't depend on data in the pending db [17:29:19] the amount of meetings this afternoon [17:31:46] 10:28 < awight> i.e., how much work can we postpone until the stabilization step [17:31:58] 10:30 < awight> dang, irssi is not doing me right on this tethered connection [17:33:18] awight_: i assume your decision to not use an irc bouncer is deliberate and principled [17:33:31] I just don't normally need one due to sedentary [17:34:27] how do you mean? [17:35:07] At home & in the office, I rarely get disconnected [17:35:44] well, I'm almost out of time for the morning, and meetings like you said. But https://etherpad.wikimedia.org/p/fr-tech-talk [17:36:12] There seems to be a lot to do, and we've regrettably created a blocking situation [17:36:16] argh, sorry for blocking [17:36:27] Not at all yr fault [17:36:33] i will have this done in afew minutes [17:36:52] ejegg is right, i was overengineering this simple thing [17:37:04] hehe IMO it's not a matter of rushing anything through, but we need to have a pretty concrete picture of the steps that we'll be taking [17:37:48] naw you were plugged into a massively overengineered framework and u took pains to simplify, future generations will sing your praises ;) [17:38:33] well the last 200ok thing anyway [17:38:43] we can just return instead of the whole header() thing [17:38:46] aw, that was one line [17:39:01] +1, good CR [17:39:57] Maybe want to confirm that none of the other gateways are using GET, though. Or do a check for empty $_GET and $_POST [17:40:13] It's too fragile with just the is_GET [17:40:57] ejegg: lmk if you think we can wait on any of the pending-payments_init interaction fixups. I still don't understand the problem [17:41:24] Is the danger that we'll fail to import the donation, and will still need the pending data? [17:41:34] I don't see how that could happen [17:43:33] (PS1) Ejegg: Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 [17:44:13] awight: lemme diagram a problematic timeline in the tt etherpad [17:44:19] (CR) jenkins-bot: [V: -1] Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 (owner: Ejegg) [17:46:17] gotta get my teeth ground on, back in an hour [17:46:21] enjoy! [17:46:44] (CR) Awight: [C: 2] Update payments_initial when donation completes (1 comment) [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 (owner: Ejegg) [17:47:12] Fatal error: Syntax only allowed in Hack files [17:47:14] that's a new one [17:47:25] (CR) jenkins-bot: [V: -1] Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 (owner: Ejegg) [17:49:28] cwd ooh, where are you seeing that? [17:50:33] ejegg: just got it locally, it was a trailing comma in an argument list [17:50:40] weird [17:51:17] awight: oho! As long as we only send 'complete' payments-init messages when we're also sending fully-populated messages to the donations queue, we won't get in trouble [17:52:23] ejegg: do you happen to know if paypal always POSTs? [17:53:15] (CR) Ejegg: "Oops, test needs fixin'" (1 comment) [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 (owner: Ejegg) [17:56:19] cwd I think so! [17:56:23] pretty sure... [17:56:33] maybe it's safer/better to just not store blank messages [17:56:53] can't really imagine a reason we'd ever want to [17:57:00] hmm, maybe. gotta make an exception for that one 'p=' parameter is all [17:57:11] which seems a little flukey from inside that class [17:57:22] what parameter is that? [17:57:41] the querystring param that routes us to paypal/listener [17:58:51] oh right i do remember seeing that [17:59:06] what's the deal with that? aren't they supposed to be pretty urls? [18:02:57] cwd that's rewritten by apache or nginx [18:03:15] (PS6) XenoRyet: Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) [18:03:15] our entry point is just smashpig_http_handler.php?p=gateway/listener [18:03:29] (CR) XenoRyet: Move Unsubscribe Queue Consumer off ActiveMQ (1 comment) [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [18:03:33] ah ha, so smashpig excpects the p [18:04:17] in that case maybe there should be a way to separate it from the payload within SP. maybe a method on the request class to get the values without p? [18:04:54] (CR) Ejegg: [C: 1] "Rockin! Maybe change the logger tag?" (1 comment) [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [18:05:01] ejegg: what do you think of that? seems like p is conceptually different from the normal params [18:05:34] cwd Probably the thing that looks at it should expunge it from the request after use [18:05:39] in fact [18:05:45] i was just gonna say that [18:05:51] it should get stripped after routing is done [18:05:59] (PS7) XenoRyet: Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) [18:06:06] yah, Core\Http\RequestHandler is the only thing that needs to care [18:06:28] ok let me try to whip that up [18:11:56] ejegg: Pushed that last adjustement if you want to take another quick look. [18:18:02] (PS1) Cdentinger: Remove 'p' from the request after routing [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313046 [18:18:22] XenoRyet: woot! Just trying to smoke test [18:18:32] good times [18:20:50] nuts, calling php-queue\PDO&pop() on an empty table is a fatal on CI [18:21:06] it was an exception locally [18:30:03] i guess cause the redis backend creates them implicitly [18:30:14] so it thinks the table doesn't exist [18:37:04] (CR) Ejegg: [C: 2] Remove 'p' from the request after routing [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313046 (owner: Cdentinger) [18:38:00] (Merged) jenkins-bot: Remove 'p' from the request after routing [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313046 (owner: Cdentinger) [18:39:28] (PS3) Cdentinger: WIP Just respond 200 to blank requests [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 [18:40:19] (CR) jenkins-bot: [V: -1] WIP Just respond 200 to blank requests [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [18:41:35] fooding early... [18:49:40] (PS1) Cdentinger: Return null from php-queue\PDO&pop() when missing table [wikimedia/fundraising/SmashPig/vendor] - https://gerrit.wikimedia.org/r/313053 [19:13:18] Fundraising-Analysis, Fundraising-Backlog, Analytics, MediaWiki-extensions-CentralNotice: Provide performant query access to banner show/hide numbers - https://phabricator.wikimedia.org/T90649#2671656 (Nuria) This can be solved in two ways: 1. Process webrequest records every hour to a table tha... [19:58:16] (PS1) Cdentinger: Return null from php-queue\PDO&pop() when missing table [wikimedia/fundraising/php-queue] - https://gerrit.wikimedia.org/r/313059 [20:17:36] (CR) Ejegg: [C: -1] "Smoke test released smoke..." (1 comment) [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [20:22:17] (CR) Ejegg: [C: 2] "I'm conflicted - pop behavior now matches push, but get/set/clear are all still going to throw exceptions when the table is missing." [wikimedia/fundraising/php-queue] - https://gerrit.wikimedia.org/r/313059 (owner: Cdentinger) [20:22:48] (Merged) jenkins-bot: Return null from php-queue\PDO&pop() when missing table [wikimedia/fundraising/php-queue] - https://gerrit.wikimedia.org/r/313059 (owner: Cdentinger) [20:26:12] (PS2) Ejegg: Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 [20:26:51] (CR) jenkins-bot: [V: -1] Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 (owner: Ejegg) [20:28:00] ejegg: i agree that the inconsistency is lame [20:28:25] for some reason that was a fatal error on CI but an exception locally [20:28:33] i guess some error level thing [20:30:07] (Abandoned) Cdentinger: Return null from php-queue\PDO&pop() when missing table [wikimedia/fundraising/SmashPig/vendor] - https://gerrit.wikimedia.org/r/313053 (owner: Cdentinger) [20:31:52] cwd weird [20:32:07] oh... I still +2ed it! [20:32:16] did you need it for tests or something? [20:32:25] eh, probably better to set it up with createTable [20:38:44] haloo! [20:39:07] where are we at, now... Maybe I can help with CR? [20:42:14] hi awight ! [20:42:35] I just need to fix a test failure in that payments_initial thing [20:43:39] (CR) Awight: [C: -1] "Add entries under Schema/ to create the jobs-paypal table..." [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [20:44:06] ejegg, awight - there are quite a few changes to our fork of php-queue that are not in smashpig vendor [20:44:24] including my latest one [20:44:28] should i update? [20:46:07] cwd: Want me to add the jobs-paypal schema? [20:46:24] (CR) Cdentinger: "> Add entries under Schema/ to create the jobs-paypal table..." [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [20:46:53] awight: i don't understand, what schema? it's just a redis store [20:48:35] cwd: oh, good point! Something is trying to access it as PDO [20:48:48] (CR) Ejegg: "yeah, it should only be a php-queue table backend in sqlite for testing. We should just call createTable in the test setup" [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [20:48:52] What's the mock store? [20:49:11] ejegg beat us to it :) [20:49:13] should just be sqlite::memory: [20:50:15] why do we need to call createTable? i wasn't seeing any erros [20:51:00] running phpunit locally? [20:51:38] yeah [20:51:48] huh. PHPQueue\PDO is supposed to run createTable if ->push throws an exception [20:51:59] maybe you're hitting that path and for some reason CI is not. [20:52:11] (PS1) Ejegg: Get all queues with QueueFactory::getQueue [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313117 [20:52:38] awight: i'm confused, what error are you seeing? [20:53:04] awight: this happened https://gerrit.wikimedia.org/r/#/c/313059/ [20:53:34] awight: but if you pop first, you still get an exception instead of a nice null [20:55:17] huh, that's a sketchy workaround but it does have some symmetry with the push() method creating the table transparently. [20:55:27] ooh hi get it [20:56:10] awight fixed one of the FIXMEs you noted: https://gerrit.wikimedia.org/r/313117 [20:56:46] yes, that one looks fun! [20:57:14] oop, merge conflict with cwd though, let's get the listener merged first [20:57:41] i think if the table being missing is an error in one situation and not another there's some inconsistency that should be addressed [20:57:53] +1 good point [20:58:33] cwd yah that's why I +2ed [20:59:18] at least the parts of the same interface should act the same [20:59:35] yeah [20:59:53] ack, videoing [21:00:00] but i agree that it's not the real fix [21:08:38] Fundraising-Backlog, FR-Ingenico: Ingenico orphan rectifier failing with no logs - https://phabricator.wikimedia.org/T146677#2672049 (ggellerman) p:Triage>Normal [21:11:35] Fundraising-Backlog: Estimate requirements for DS ticketing system transition - https://phabricator.wikimedia.org/T146511#2663705 (DStrine) @MBeat33 this seems to be pretty open-ended and it might better to engage OIT on this. [21:12:13] Fundraising-Backlog, FR-Paypal, FR-Smashpig, Recurring-Donations: Should import new email address when PayPal recurring donor changes their PP profile contact - https://phabricator.wikimedia.org/T146309#2672061 (ggellerman) p:Triage>Normal [21:13:34] Fundraising-Backlog, FR-Ingenico, MediaWiki-extensions-DonationInterface: Orphan rectifier is breaking fredge consumer - https://phabricator.wikimedia.org/T146244#2672081 (ggellerman) Open>Invalid [21:14:04] Fundraising-Backlog, FR-Ingenico, MediaWiki-extensions-DonationInterface: Orphan rectifier is breaking fredge consumer - https://phabricator.wikimedia.org/T146244#2654783 (ggellerman) Per @awight this does not seem to be happening anymore [21:14:23] Fundraising-Backlog, FR-Paypal, FR-Smashpig, Recurring-Donations: When PayPal recurring donor changes their PP profile contact, update the email address in CRM - https://phabricator.wikimedia.org/T146309#2672089 (awight) [21:16:10] Fundraising-Backlog: Engage import failing to import certain significant fields - https://phabricator.wikimedia.org/T146295#2672093 (ggellerman) p:Triage>Normal [21:17:34] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM: create a dashlet for dedupe script stats - https://phabricator.wikimedia.org/T146229#2672105 (ggellerman) p:Triage>Normal [21:20:21] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM: Dedupe job failing - https://phabricator.wikimedia.org/T146607#2672112 (ggellerman) p:Triage>Normal [21:22:18] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, DBA, Performance: Spike: Look into transaction isolation level and other tricks for db contention - https://phabricator.wikimedia.org/T146821#2672116 (awight) [21:22:31] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, DBA, Performance: Spike: Look into transaction isolation level and other tricks for easing db contention - https://phabricator.wikimedia.org/T146821#2672134 (awight) [21:22:38] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM: Dedupe job failing - https://phabricator.wikimedia.org/T146607#2672135 (awight) [21:22:40] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, DBA, Performance: Spike: Look into transaction isolation level and other tricks for easing db contention - https://phabricator.wikimedia.org/T146821#2672116 (awight) [21:23:53] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, Spike: SPIKE: investigate level of completeness of civi deduped DB - https://phabricator.wikimedia.org/T146384#2672153 (ggellerman) p:Triage>Normal [21:25:29] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, DBA, Performance, Spike: Spike: Look into transaction isolation level and other tricks for easing db contention - https://phabricator.wikimedia.org/T146821#2672160 (ggellerman) p:Triage>High [21:32:31] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM: Periodically run Civi contact import performance tests, note trends - https://phabricator.wikimedia.org/T146338#2672191 (ggellerman) p:Triage>High [21:33:42] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM: Periodically run Civi contact import performance tests, note trends - https://phabricator.wikimedia.org/T146338#2657931 (awight) Run performance tests on staging for a baseline and to catch regressions before they're merged to deployment. Measure produ... [21:33:55] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, Performance: Store live donation input timing in database - https://phabricator.wikimedia.org/T146824#2672196 (Ejegg) [21:34:04] (PS1) Cdentinger: update php-queue to 7834573f5dc2ab8935684259661b4a1fa83978cf [wikimedia/fundraising/SmashPig/vendor] - https://gerrit.wikimedia.org/r/313124 [21:36:58] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, Performance: Store live donation input timing in database - https://phabricator.wikimedia.org/T146824#2672216 (ggellerman) p:Triage>High [21:38:42] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, FR-Adyen, Patch-For-Review, Unplanned-Sprint-Work: Adyen failmails on new message type - https://phabricator.wikimedia.org/T146447#2672221 (Ejegg) Open>Resolved [21:40:11] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, Unplanned-Sprint-Work: Ingenico audit is broken by queue trickery - https://phabricator.wikimedia.org/T145852#2672223 (Ejegg) Open>Resolved a:Ejegg [21:40:34] Fundraising Sprint Muggle Baiting, Fundraising Sprint Nitpicking, Fundraising Sprint Octopus Untangling, Fundraising Sprint Pretending This Isn't Happening, and 8 others: Rewrite orphan rectifier to use the pending database and WmfFramework - https://phabricator.wikimedia.org/T141486#2672227 (Ejeg... [21:40:37] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, FR-ActiveMQ, FR-Ingenico, and 3 others: New orphan rectifier balks on CVV="0" - https://phabricator.wikimedia.org/T145846#2672226 (Ejegg) Open>Resolved [21:40:47] (PS8) XenoRyet: Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) [21:40:56] Fundraising-Backlog, Wikimedia-Fundraising-CiviCRM, FR-ActiveMQ, FR-Smashpig, and 3 others: [Epic] Payments and other components should send a copy of outbound queue messages a Redis queue - https://phabricator.wikimedia.org/T130287#2672229 (Ejegg) [21:40:59] Fundraising Sprint Pretending This Isn't Happening, Fundraising Sprint Qwerty Thwacking, Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, and 6 others: Audit processors should mirror outbound queue messages - https://phabricator.wikimedia.org/T141484#2672228 (Ejegg) Open>Resolv... [21:52:20] awight: Cool! [21:52:24] spatton: Hi! great news that you captured a banner preview bug [21:52:45] Centralnotice-template-B1617_0928_enUS_mob_p1_lg_amt_fiv ? [21:52:51] Exactly [21:53:18] I see 4 loglines so far... okay thanks, there are two ways we can proceed here. [21:53:40] But first of all--the preview error looks like *nothing*, right? no "" at the top? [21:54:17] spatton: If you're up for it (removing this time slice from your test results), I say let's [21:54:33] let's turn that thing on and see what the impact is on readers [21:55:14] Right, after your latest work, there's no weird bracketed code; the banner simply fails to load. [21:55:35] Ok, sure, we can do that. So this was intended to be used in our weekly Big English test, actually scheduled for tomorrow. I'll enable this vs the control for 15 minutes in enUS? [21:55:47] cool--or attach it to the test_campaign [21:55:54] Yeah, I'll do that :) [21:56:11] You can change the targeting to get as much traffic as you'd like, too [21:56:22] it's currently set for about 0.01% of global impressions [21:57:04] Ok word, it's an ask string test specific to USD, so I will change the country targeting for test_campaign [21:58:14] sounds good [21:59:18] (CR) Awight: "See Tests/QueueConsumerTest.php" [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 (owner: Cdentinger) [22:00:27] ejegg: so this is the one I want to deploy later today https://gerrit.wikimedia.org/r/#/c/310219/ [22:00:41] I'm just giving it a last re-read myself out of paranoia [22:01:21] spatton: It's live? I saw one more logline [22:02:15] It's live! Targeted to US, 1% of traffic to start [22:02:33] okay--There was only one failure to display, AFAICT [22:03:04] (CR) Ejegg: [C: 2] "The smoke has cleared!" [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [22:03:19] eileen: looking! [22:03:24] spatton: I know this is hard to do after the fact, but can you remember the exact steps leading up to the failure to preview? [22:03:32] Might be me; I've tried a couple more times. Just did with language = it, previously was language = es [22:03:35] (CR) jenkins-bot: [V: -1] Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [22:04:14] d'oh! [22:04:15] spatton: gotcha. I saw one in `en` at UTC 22:01, and one in `it` just now. [22:05:00] Going to bump the traffic limit up to 10% [22:05:30] k [22:05:39] spatton: Can you hit the preview link again? [22:05:58] I'm trying not to jiggle anything myself... [22:06:31] I see a cache miss in `en` [22:06:38] x2 [22:06:47] x3 :) but still not a flood [22:07:13] Could all be me; just tried a few times in a couple browsers, still not getting the banner [22:07:20] *weird*. [22:07:27] Last attempt was language = he [22:07:33] yep I saw that [22:07:54] ok I'll deny my superstition and try the link myself [22:08:25] WLM is at 110% of course... [22:09:15] Haha, right. I set the priority to emergency to hopefully supersede it [22:10:23] huh, I'm still getting exclusively WLM banners. Private tab with no cookies, en/US... [22:10:46] (PS1) Cdentinger: Explicitly create test tables in sqlite [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313127 [22:11:10] Tried about 40 times, that's very unlikely. [22:11:20] (CR) Ejegg: [C: -1] "Couple left joins missing 'ON' clauses" (2 comments) [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/310219 (https://phabricator.wikimedia.org/T96410) (owner: Eileen) [22:11:27] there's a way to force a random number... [22:12:45] Interesting. I'm happy to cut out the country targeting, reset to language = nl? [22:12:55] Just to see if we can get some impressions [22:13:15] Seeing 17 clicks, now, in statler [22:14:41] Yeah, definitely some data now coming in; impressions are always delayed, but seeing 6 donations, 33 clicks [22:14:50] (PS9) Ejegg: Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [22:15:12] spatton: That was you trying a bunch of languages, right? [22:15:23] (CR) jenkins-bot: [V: -1] Move Unsubscribe Queue Consumer off ActiveMQ [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [22:15:29] Yeah, sorry. No one else should be trying anything but en. I did pl, ja, es, fr, it, sv [22:15:39] okay yeah, that's why I was wondering [22:15:44] perfect [22:15:59] huh. So it's working fine, just the preview is borken, I suppose. [22:16:24] Well! That is an interesting development. [22:16:53] I guess I did muddy the results up ... if it happens again, should I try a live test in the test_campaign, I won't touch anything at all; just let it run 15 mins and see if we are getting clicks/dons? [22:17:09] That would be a good test. [22:17:29] Meanwhile, I'm disturbed that I can't serve myself the banner... I'll work on that, and also on the preview failure [22:17:45] Yeah, for some reason it didn't occur to me that it might just be a preview thing ... I figured we'd have a flood of failures and mine would just be drops in the torrent [22:18:31] Are you ok w/ me disabling the campaign, for now? [22:19:25] yes, thanks! [22:19:36] Thanks much for working on it, adam! [22:20:26] (CR) Ejegg: [C: -1] "Think you need another arg to ensure it's always the same instance." (2 comments) [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313127 (owner: Cdentinger) [22:32:45] (CR) Ejegg: "recheck" [wikimedia/fundraising/crm] - https://gerrit.wikimedia.org/r/312084 (https://phabricator.wikimedia.org/T145419) (owner: XenoRyet) [22:35:42] (PS1) Ejegg: WIP use redis for Adyen jobs [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313130 [22:39:40] (CR) Cdentinger: "> (2 comments)" (2 comments) [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313127 (owner: Cdentinger) [22:41:02] boy is that gerrit quote feature confusing [22:41:05] cwd: It's nasty, but the $config->object call inside the regular logic will create a new instance unless we've already marked that node as persistent [22:41:18] check out the source for Configuration::object [22:42:01] (PS3) Ejegg: Update payments_initial when donation completes [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313042 [22:42:20] i'm not sharing any data between tests [22:42:36] right, but you want to share the instance over the course of the test [22:44:08] hrm, so why did this fix the errors? [22:46:23] errr ... [22:49:17] waitaminute... are we no longer resetting the config after each test? [22:49:26] we've got that Context::set call in the main tearDown [22:49:33] i don't see it referring to the same db twice [22:49:35] in the same test [22:49:52] unless you count in setup and the test [22:50:05] ugh, I think we /may/ inadvertantly be sharing instances [22:50:24] that's the only way it makes sense that would fix it [22:50:37] yeah... we need to null out the default config object in the teardown too [22:50:48] let's see what that breaks [22:51:54] in testConsume it refers to 2 different queues so i don't think that should matter [22:51:58] (pls add that sort of logic to TestConfiguration) [22:52:12] but you are saying createTable in setUp won't affect the tests? [22:54:58] (PS1) Ejegg: Null out default config in tearDown [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313135 [22:56:28] you have to pass a boolean to make config persist at all? [22:56:45] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, MediaWiki-extensions-CentralNotice, Operations, and 3 others: Banner not showing up on site - https://phabricator.wikimedia.org/T144952#2672426 (awight) So, @spatton and I just ran the experiment suggested above, where we add a broken b... [22:56:54] cwd if you don't pass true to ->object(), you get a fresh instance the next time [22:57:15] if you do pass true, then every following call to ->object() with the same key will return the same instance [22:57:42] so for sqlite::memory: backends, we need the same instance all through the test [22:58:03] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, MediaWiki-extensions-CentralNotice, Operations, and 3 others: Banner not showing up on site - https://phabricator.wikimedia.org/T144952#2672430 (awight) Tricking the HTTP cache with extra parameters like "foo=1" doesn't change the respo... [22:58:14] because they don't last longer than the lifetime of the object? [22:58:24] exactly. [22:58:26] cwd right [22:58:28] but...shouldn't sqlite be persistent for the session? [22:59:11] if it was a file, yeah [22:59:16] but each time you connect to a 'memory' backend it creates a new db in ram [23:00:04] Fundraising Sprint Rocket Surgery 2016, Fundraising-Backlog, MediaWiki-extensions-CentralNotice, Operations, and 3 others: Banner not showing up on site - https://phabricator.wikimedia.org/T144952#2672432 (awight) For some reason, this query is giving zero results. I must be in a parallel univer... [23:00:44] is that smashpig behavior? [23:03:31] cwd: naw that's sqlite. seems like a good choice [23:09:58] (CR) Cdentinger: [C: 2] Null out default config in tearDown [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313135 (owner: Ejegg) [23:14:01] (Merged) jenkins-bot: Null out default config in tearDown [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313135 (owner: Ejegg) [23:15:02] (PS4) Cdentinger: WIP Just respond 200 to blank requests [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/312941 [23:15:18] (PS2) Cdentinger: Explicitly create test tables in sqlite [wikimedia/fundraising/SmashPig] - https://gerrit.wikimedia.org/r/313127 [23:20:06] ejegg: i rebased on the config reset and now the first one passes too :S https://gerrit.wikimedia.org/r/#/c/312941/ [23:29:12] uhhh, dang. did I not null it hard enough? [23:31:34] needs some spl_object_hash investigation... [23:40:17] Fundraising-Backlog: Estimate requirements for DS ticketing system transition - https://phabricator.wikimedia.org/T146511#2672563 (MBeat33) Open>Resolved a:MBeat33 Thank you @DStrine I've opened a thread with them and will close this task. [23:52:05] ejegg: don't understand why this passes: https://gerrit.wikimedia.org/r/#/c/312941/4 wrt pop() on a non-table [23:52:26] yeah... that's freaky