[01:28:12] (PS1) Gergő Tisza: Add original file clicks to the TSV data generation [analytics/multimedia] - https://gerrit.wikimedia.org/r/140629 [01:31:25] (PS1) Gergő Tisza: Show original file view count on graph [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140630 [06:32:42] (PS1) Gergő Tisza: Display opt-out/opt-in events [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140657 [06:37:15] (PS1) Gergő Tisza: Collect data for opt-in/opt-out events [analytics/multimedia] - https://gerrit.wikimedia.org/r/140659 [07:35:50] hola qchris [07:35:55] Hi nuria [07:36:08] a question regarding the backups [07:37:08] the 'hourly' backup is removing [07:37:20] the prior hourly backup ... [07:37:49] for the db and redis... is that correct? [07:38:12] Yes. [07:38:21] In the hourly directory, they do. [07:39:13] Lines 59+60 in https://gerrit.wikimedia.org/r/#/c/139557/7/files/backup_hourly [07:40:15] The relevant comments are in PS3 and the change happened in PS4. [07:40:53] i saw those [07:41:23] but that implies that if db gets wiped out (and we do not notice) we are backing up an empty backup [07:41:45] is that intended [07:42:03] we woudl be relying on the prior daily backup in that case [07:42:29] I assume that by wiped you mean "The rdb file gets removed", not "The rdb gets empty". [07:42:30] *we would be [07:42:45] either one [07:42:51] ? [07:43:02] If the rdb file gets empty, [07:43:09] many things can fail: virtual instance crashes [07:43:27] Regardless of whether we remove it in the destination, copying the file over makes the previous rdb file dissappear :-) [07:43:28] someone types "drop database" by mistake [07:43:48] Ja. Make the script more robust. [07:43:53] Totally do it. [07:44:04] There are currently many points where things can fail. [07:44:05] * YuviPanda waits for people to finish up and test a redis backup script so he can steal it later [07:44:15] haha [07:44:20] It ours. We're close-sourcing it :-P [07:44:29] also are you guys using redis for persistant data that shouldn't be lost?! [07:44:31] our "redis backup script" is "maximum sofistication" [07:44:46] like it is 1999 [07:45:00] YuviPanda: no not redis [07:45:05] oh [07:45:06] mysql yes [07:45:09] ok [07:45:21] redis is for convenience as data is stored for 30 days only [07:46:07] ah, cool [07:48:10] But nuria ... just to make sure you're not implementing things that will get thrown away afterwards ... [07:48:19] recall the discussion of yesterday after standup. [07:48:33] And milimetric's corresponding comment on [07:48:47] line 67 of https://gerrit.wikimedia.org/r/#/c/139557/6/files/backup_hourly [07:49:01] i am not going to change any thing but the obvious, and that is that teh script as it is now will always fail teh first time is run [07:49:37] milimetrigs make the problem that "-f" in missing from the "rm"s go away. [07:50:01] s/milimetrigs make/milimetric's comment makes/ [07:50:23] * qchris prays for magically learning how to type :-( [07:50:36] * qchris else will jump off a privi ledge. [07:51:00] juas [07:52:58] but wasn't that just for the public_files? [07:53:45] No, that would not make any sense. [07:54:01] Why would we tar the public files, only to have them tared again at the outer layer. [07:54:24] Also in the comment, milimetric refers to "[everything we need to back up]" [07:54:49] and pipes it into last-hourly-backup.tar.gz. [07:55:45] Also in the discussion yesterday, we said to only do one tar of all the relevant files. [09:23:36] qchris, other question [09:23:39] where is $::wikimetrics_backup [09:23:42] defined? [09:24:02] on teh environment on labs as well? [09:24:04] *the [09:24:32] Looks like a global to me. Let me have a look at the change again. [09:27:08] ok, global in labs, https://gerrit.wikimedia.org/r/#/c/139558/1..2/manifests/role/wikimetrics.pp [09:27:08] The change does not seem to set this global variable. [09:27:19] no it does not thus my question [09:27:42] but as long as that is intentional that is fine [09:28:20] I think that it is intentional. [09:28:49] But code review for those two changes is not finished. [09:29:29] So once basic items on the scripts are settled, expect more changes on the outside :-) [09:30:01] Improve, wherever you think it fits. It'll only help the code review process. [09:30:53] ya, i'd like to see that at least puppet runs 7 setups crons as expected [09:31:16] but the backup.pp file does not parse yet, i need to change a few things [09:53:10] "puppet runs & sets up" crons as expected [10:04:07] (CR) Gilles: [C: 2] Add above-fold filepage link to stats fields [analytics/multimedia] - https://gerrit.wikimedia.org/r/140263 (owner: Gergő Tisza) [10:04:11] (CR) Gilles: [C: 2] Add original file clicks to the TSV data generation [analytics/multimedia] - https://gerrit.wikimedia.org/r/140629 (owner: Gergő Tisza) [10:04:14] (Merged) jenkins-bot: Add above-fold filepage link to stats fields [analytics/multimedia] - https://gerrit.wikimedia.org/r/140263 (owner: Gergő Tisza) [10:04:17] (CR) Gilles: [C: 2] Collect data for opt-in/opt-out events [analytics/multimedia] - https://gerrit.wikimedia.org/r/140659 (owner: Gergő Tisza) [10:04:20] (Merged) jenkins-bot: Add original file clicks to the TSV data generation [analytics/multimedia] - https://gerrit.wikimedia.org/r/140629 (owner: Gergő Tisza) [10:04:23] (Merged) jenkins-bot: Collect data for opt-in/opt-out events [analytics/multimedia] - https://gerrit.wikimedia.org/r/140659 (owner: Gergő Tisza) [11:40:21] (CR) Gilles: [C: 2] Add click tracking for the above-the-fold file page link [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140264 (owner: Gergő Tisza) [11:42:03] (CR) Gilles: [C: 2] Show original file view count on graph [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140630 (owner: Gergő Tisza) [11:42:32] (CR) Gilles: [V: 2] Add click tracking for the above-the-fold file page link [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140264 (owner: Gergő Tisza) [11:43:06] (CR) Gilles: [V: 2] Show original file view count on graph [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140630 (owner: Gergő Tisza) [11:43:44] (CR) Gilles: [C: 2 V: 2] Display opt-out/opt-in events [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140657 (owner: Gergő Tisza) [13:50:37] (PS1) Gilles: Add the new metrics to the graphs [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140710 [14:08:38] (CR) Gilles: [C: 2 V: 2] Add the new metrics to the graphs [analytics/multimedia/config] - https://gerrit.wikimedia.org/r/140710 (owner: Gilles) [16:04:53] tnegrin: are you joining the multimedia meeting? [16:05:21] I'm trying to attend -- I don't have good enough network it seems [16:05:48] my parents-in-law have cheap ass dsl [16:06:15] alright, I’ll relay that [16:23:28] we can’t really hear you :-/ [18:06:51] [travis-ci] wikimedia/mediawiki-extensions-EventLogging#221 (wmf/1.24wmf10 - 1268941 : Reedy): The build passed. [18:06:51] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-EventLogging/commit/126894167d2c [18:06:51] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-EventLogging/builds/27976895 [18:08:07] mwalker, 200 and 304, right? [18:08:14] I've screwed up my regexes somewhere. Oh dear. [18:09:52] Ironholds, our code reads: if squidstatus not in (0, 200, 204, 206, 304): [18:10:03] ^ So all of those are considered valid [18:11:34] mwalker, ta [19:17:39] (PS14) AndyRussG: WIP Create a cohort from campaign participations [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/126927 (owner: Awight) [19:20:14] (CR) AndyRussG: "In this patch set:" [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/126927 (owner: Awight) [21:13:36] (PS1) Milimetric: Don't show invalid cohorts on report screen [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/140823 (https://bugzilla.wikimedia.org/66290) [21:21:12] thanks for the comments qchris, you read my mind: backup test seems at risk of not being done by Tuesday :) [21:21:19] *backup task [21:22:37] I looked several times today and wanted to review them. Too bad the wrong patch set got uploaded in the European morning :-/ [21:23:31] But it's just two scripts + puppet glue. I hope we can get that done :-) [21:33:16] (PS2) Milimetric: Don't show invalid cohorts on report screen [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/140823 (https://bugzilla.wikimedia.org/66290) [21:33:18] (PS1) Milimetric: Ensure wiki cohorts work [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/140830 (https://bugzilla.wikimedia.org/66290) [23:32:41] DarTar: I'm logged into tin and looking at all.dblist, but I don't see any candidates for the unified eventlogging thing you mentioned yesterday... [23:33:16] you should connect to analytics-store [23:34:06] DarTar: that's a machine name? [23:35:19] DarTar: ok i tried to ssh through bast1001 and was denied access [23:35:19] yes [23:35:32] I should RT this? [23:37:07] no [23:37:16] do you have access to stat1002 or stat3.wikimedia.org? [23:37:30] Ironholds: what kind of road trip... [23:37:36] ssh there through bast, then mysql analytics-store.eqiad.wmnet... [23:37:37] well yes, I have stat1003 access [23:37:44] using the research username/pass [23:37:45] bingo [23:38:10] but I cannot ssh the third leg, it wants publickey auth [23:38:32] ssh -A [23:38:33] ? [23:38:47] #necessaryevil [23:38:47] yikes [23:38:52] NO [23:38:54] NO? [23:39:02] awight, as in, it won't let you ssh from stat3 to analytics-store? [23:39:15] awight@stat1003:~$ ssh analytics-store.eqiad.wmnet [23:39:18] Permission denied (publickey). [23:39:25] * Ironholds sighs [23:39:26] -A to stat3 or not [23:39:32] awight, read what I wrote ;p [23:39:39] do not -A. you are not looking to ssh to analytics-store [23:39:49] connect to stat3, through bastion [23:39:52] done [23:39:55] then connect to analytics-store through the mysql client [23:39:58] aha [23:39:59] jesus [23:40:03] * Ironholds slaps mark [23:40:09] never recommend agent forwarding! [23:40:36] https://wikitech.wikimedia.org/wiki/Server_access_responsibilities [23:40:40] thou shalt not use forwarding! [23:40:59] Ironholds: ino [23:41:26] Ironholds: when you say, research un/pw, that doesn not seem to be "research"/ [23:41:41] you don't have the research username/pass? [23:41:42] ah. [23:41:48] * Ironholds headscratches [23:42:03] pm/otr? [23:42:41] btw normal people do some butt-scratching and burger-wolfing when they stop somewhere on a road trip. [23:42:46] you two are sick. not normal. [23:42:58] helping people with access from the road. foo! [23:43:35] awight, we're in Seattle! [23:43:39] working out of J-Mo's hackspace [23:43:46] it's where he hacks people up [23:43:51] I think I saw some of the BTK victims earlier [23:43:54] * J-Mo chop chop [23:43:56] there's one mystery explained [23:51:59] Ironholds: DarTar: so, the db is "log"? [23:52:11] yup [23:52:24] oh wow. /me sees the OS blow me a kiss [23:58:32] DarTar: looks like there is some evidence that edit conflicts are not rare...