[00:21:02] (CR) Bmansurov: Update limn graphs (untested) (1 comment) [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/199162 (https://phabricator.wikimedia.org/T93690) (owner: Jdlrobson) [01:08:36] nuria, do you know how often events get put into the DB? [01:27:51] Krenair: constantly as they come in [01:27:58] ok [01:28:08] I think eventlogging in deployment-prep is broken [01:28:15] nothing is going into all-events.log [01:28:15] Krenair: there is a couple secs buffer but that's just about the delay (seconds) [01:28:22] mysql> select max(timestamp) from Edit_11448630; [01:28:22] +----------------+ [01:28:22] | max(timestamp) | [01:28:22] +----------------+ [01:28:23] | 20150331234226 | [01:28:24] +----------------+ [01:28:25] 1 row in set (0.00 sec) [01:28:58] Krenair: puf, labs was totally broken yesterday so it wouldn't be surprising that it is, let me see if i can resurrect it [01:30:01] -rwxrwxrwx 1 eventlogging eventlogging 0 Mar 31 06:49 all-events.log [01:32:43] nuria, /etc/eventlogging.d/consumers/all-events-log said file:///all-events.log instead of file:///srv/log/eventlogging/all-events.log [01:32:57] fixed that, restarted eventloggingctl... no luck yet [01:34:12] uh... hm. it was showing as stopped. [01:34:12] let me restart and make sure all configs are ok [01:34:14] ok [01:34:21] Krenair: ya, cause i just stopped it [01:34:26] oh, sorry [01:34:44] ok [01:35:54] Krenair: i see events into the client side event log [01:36:59] yes [01:37:01] but not server or all [01:37:19] Krenair: ya, let me make sure it has the latest vs of the code deployed [01:41:53] Krenair: puppet needs to update the config files i think, but where is the puppet depot this machine pulls from? [01:43:02] Krenair: arg , we are going to run into this issue that is broken puppet on labs ... [01:43:12] deployment-salt, I think? [01:45:26] Krenair: let me try to re-enable puppet, rerun and see what i get [01:46:11] ok [01:50:31] Krenair: ok, i re-run puppet and brought machine up-todate [01:53:06] Krenair: i think we are good now [01:53:20] Krenair: i see events on all-events [01:53:29] yep [01:54:04] Krenair: ok, let me know of any issues, will be here for a bit more [01:54:56] waiting for them to be fed into the db [01:55:47] select max(timestamp) from Edit_11448630; still returns 20150331234226 :/ [02:33:28] nuria, still nothing in the DB [02:33:53] Krenair: but events are in all-events.log right? [02:34:06] looks like it [02:35:07] Krenair: ahhh, i bet is the batch size... [02:38:23] Krenair: try to send a number of events now, batch size is 400 [02:38:36] which is no good for labs (that is something we need to fix) [02:48:31] nuria, still nothing in the db... [02:48:45] 20150331234226 is max(timestamp) for that table. [03:09:56] milimetric: you've small mail, when you've time. [03:10:39] kart_: hm? [03:10:55] I don't see anything in my inbox [03:15:03] milimetric: can you see if language limn is not updated yet? [03:15:14] I did as per README/instructions. [03:15:47] milimetric: http://language-reportcard.wmflabs.org/ [03:16:39] taking a look [03:19:40] kart_: it looks like you added bgwiki frwiki mkwiki shwiki slwiki and these all are calculating now, I'm not sure what the problem is [03:26:24] milimetric: yes. I added last 5. [03:26:32] but not showing in graph [03:26:49] kart_: they're in the graph, there's just one point though since you just added them [03:27:00] so you have to hover all the way to the right and you get one data point for them [03:28:30] milimetric: oh. Thanks! [03:28:36] milimetric: sorry for noise. [03:28:41] kart_: for the record, I did nothing :) [03:28:45] so thank yourself [10:19:40] g'morning [14:39:02] (PS4) Mforns: Add support for wiki explosion and others. [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 [14:40:32] (CR) Mforns: "Patch 4 implements frequency vs. granularity independence." [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 (owner: Mforns) [14:52:55] (PS3) Mforns: Adding wikitext analysis to queries [analytics/limn-edit-data] - https://gerrit.wikimedia.org/r/200159 (owner: Milimetric) [15:10:36] kevinator: wanna have a look at vision, slides 11/12 ? [15:56:37] (PS3) Ottomata: Add a simple hacky script to print out ADD PARTITION statements for a webrequest table [analytics/refinery] - https://gerrit.wikimedia.org/r/198759 [15:56:43] (CR) Ottomata: [C: 2 V: 2] Add a simple hacky script to print out ADD PARTITION statements for a webrequest table [analytics/refinery] - https://gerrit.wikimedia.org/r/198759 (owner: Ottomata) [16:02:25] (PS3) Ottomata: Add oozie util workflow to launch spark jobs [analytics/refinery] - https://gerrit.wikimedia.org/r/201009 (https://phabricator.wikimedia.org/T94596) [16:45:07] (CR) Joal: "Shouldn't we provide parameters for driver and executors memory setup? That's one of the cool feature of Spark :)" [analytics/refinery] - https://gerrit.wikimedia.org/r/201009 (https://phabricator.wikimedia.org/T94596) (owner: Ottomata) [16:46:35] joal, probably! although, i' not sure how that works in yarn [16:49:30] ottomata: I think Spark asks yarn for the correct amount of resource [17:10:50] (PS3) Jdlrobson: Update limn graphs (untested) [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/199162 (https://phabricator.wikimedia.org/T93690) [17:20:23] Ironholds: Will you have some time for me later on this morning ? [17:28:23] ottomata: I'd like to copy a version of enwiki history dump on hdfs [17:28:58] sure [17:29:01] ottomata: I can do that from stat1002, and just to confirm you are happy if I put it somewhere like: /wmf/data/wikidumps/enwiki/date [17:29:07] hm. [17:29:28] I can go for my personal forlder, but I think it's better to have it global [17:29:33] wdyt ? [17:29:39] i think unless we build a process to automate doing it, and officially make this an analytics supported dataset, it might be better in your user dir [17:29:46] not sure though [17:29:50] ok perfect [17:29:52] I do that [17:30:12] That's about 1Tb to copy (3tb real, with replication) [17:36:49] be back in a bit [18:00:57] (PS5) Mforns: Add support for wiki explosion and others. [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 [18:06:23] (CR) Mforns: "Patch 5 fixes a bug in get_previous_results paths that wasn't reading the report files properly." [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 (owner: Mforns) [18:26:28] (PS4) Mforns: Adding wikitext analysis to queries [analytics/limn-edit-data] - https://gerrit.wikimedia.org/r/200159 (owner: Milimetric) [18:27:39] (CR) Mforns: "I added the configuration for the queries to execute, and tweaked the way the placeholders are used (otherwise the MySQLdb lib did fail, I" [analytics/limn-edit-data] - https://gerrit.wikimedia.org/r/200159 (owner: Milimetric) [18:39:52] nuria, yt? [18:39:59] mforns: yesssir [18:40:02] hi [18:40:13] I'm trying to execute the modified reportupdater [18:40:31] on stat1003, but as I have no sudo permits, I can not [18:41:09] tried to scp the code into my home folder but when I execute, the program can not get the research user password [18:41:17] I think [18:41:51] so, nuria I think I'm going to self-merge the changes, if that is ok to you [18:42:01] mforns: on stat1003 you shouldn't need sudo to access the db, let's see whether you have db access 1st [18:42:03] try: [18:42:10] ok [18:42:43] mysql --defaults-extra-file=/etc/mysql/conf.d/research-client.cnf --host dbstore1002.eqiad.wmnet -e "select left(timestamp,8) ts , COUNT(*) from log.NavigationTiming_10785754 where timestamp >= '20150319062613' and timestamp <='20150320062613' group by ts order by ts"; [18:42:46] for example [18:43:23] with your regular user [18:43:35] yes, this works [18:43:44] ok, so db access is no problem [18:44:19] The error I get is: MySQLdb can not connect to database ((1045, "Access denied for user 'mforns'@'10.64.36.103' (using password: NO)")) [18:44:37] but the user should be research [18:44:42] right [18:45:32] mforns: you have to pass --defaults-extra-file=/etc/mysql/conf.d/research-client.cnf for any mysql command [18:45:39] let me look at crontabs on 1003 [18:46:42] nuria, if I do it as you said, it works, but the problem is that the code I copied to my home folder (reportupdater) is being run as mforns, and can not access /a/.my.cnf.research [18:46:53] so can not access the db either, it should be run as root [18:47:00] but I can not [18:47:42] are you sure it should be run as root? I think everyone -even root- uses the research account [18:48:01] cc ottomata , holaa, yt? [18:48:15] mforns@stat1003:~/reportupdater$ less /a/.my.cnf.research [18:48:33] '/a/.my.cnf.research: Permission denied' [18:48:43] try: /etc/mysql/conf.d/research-client.cnf [18:49:06] ok, it works [18:49:15] is this the new file after password change? [18:50:11] mforns: that is the file to use [18:50:15] ok [18:50:19] hm, but it hought the /a/ one was a symlink [18:51:21] hm [18:51:24] yes, but the link points to: /etc/mysql/conf.d/stats-research-client.cnf instead of /etc/mysql/conf.d/research-client.cnf [18:51:25] weird, limn is doing this, hm. [18:52:21] ja, that one is for crons run by the stats users [18:52:22] users [18:52:23] user* [18:53:10] nuria, ok I'll point to the one you passed in the config [19:22:25] ottomata: not to be the party pooper YET AGAIN but .... [19:22:50] ottomata: I think there are additional changes to do so we can keep building the refinery code on 1002 [19:29:02] yesss? [19:29:29] ottomata: can you mvn compile refinery on 1002 [19:29:53] checking [19:29:56] with marcel's change? [19:30:01] or just master head? [19:30:11] ottomata: with marcel's change [19:31:59] nuria, you getting this one? [19:31:59] com.jolbox:bonecp:jar:0.8.0.RELEASE [19:34:11] ottomata: yess [19:34:13] hm. [19:34:18] and others, [19:34:59] ottomata: if you are in the middle of something we can fix this at a later time [19:35:57] i think i can help, hang on [19:36:09] i'm in the middle of something else not working [19:36:09] :p [19:40:11] Ironholds: you there? [19:40:38] joal, yep [19:40:44] cool :) [19:40:45] but only for the next ~5 minutes, so, warning [19:40:50] np [19:40:56] batcave for efficienvy ? [19:41:54] sure! [19:44:23] nuria: blast your .m2 and try again [19:46:42] ottomata: k [19:48:16] ottomata: working ..... [19:48:50] cool [19:50:19] ottomata: thank youuuuu [19:54:22] Lads, have a good end of day ! [19:54:33] Tomoooooorrow :) [19:56:08] joal|night: ciao ciao [20:06:46] laterrrs [20:08:04] aww maaan [20:08:53] have to use kafka 0.8.2 to use schema registry. [20:09:01] rats. i guess i'll just use confluents version in labs for now :/ [20:27:23] nuria, aren't events supposed to be going into the DB in the order which they were generated? [20:27:38] (CR) Nuria: Add oozie util workflow to launch spark jobs (1 comment) [analytics/refinery] - https://gerrit.wikimedia.org/r/201009 (https://phabricator.wikimedia.org/T94596) (owner: Ottomata) [20:28:54] Krenair: events are buffered as received, so they are inserted together as received. [20:29:09] | 31374 | 20150331142238 | ready | [20:29:09] | 31378 | 20150331142237 | init | [20:29:09] Krenair: but "received" and "created" are two different things. [20:29:10] :/ [20:29:35] Krenair: sounds like client is sending events out of sequence, look at timestamps on all-events.log [20:29:48] init comes from the server [20:29:55] that's id, timestamp, event_action [20:30:09] the timestamp is larger on the first, but ID smaller [20:31:19] Krenair: ya, that can happen as server side events come from php stack [20:31:37] Krenair: timestamp is added there (let me check) [20:32:06] Krenair: for client side events I think times are added by varnish (need to check too) [20:32:18] Krenair: actually varsnishncsa [20:32:52] Krenair: ids are pretty meaningles though, we plain removed them as of recent [20:33:05] Krenair: as we do not use them for anything [20:36:10] mysql> select max(timestamp) from Edit_11448630; [20:36:10] +----------------+ [20:36:11] | max(timestamp) | [20:36:11] +----------------+ [20:36:11] | 20150401191956 | [20:36:12] +----------------+ [20:36:13] 1 row in set (0.00 sec) [20:36:16] it's stuck again? :/ [20:36:19] it is very possible with varrnishncsa and varnishkafka to have sequence ids that are not in timestamp order, as it depends on if the timestamp or seqid is computed at the beginning or the end of the request [20:36:28] i can't rememer which happens when [20:36:38] i think the timestamp is request response time [20:36:46] and the seq id is computed at request receive time [20:36:49] or it could be vice versa [20:41:06] ottomata: but what he linked to are not uuids are ids on db, those are set upon insertion but - as i said - are not used at all [20:42:12] Krenair, ottomata: for client side events varnishncsa appens timestamps [20:44:12] Krenair, ottomata : for server side events the php onlines (however we call these machines do set the time of the event): https://github.com/wikimedia/mediawiki-extensions-EventLogging/blob/master/includes/EventLogging.php#L55 [20:50:34] Krenair: nothing has changed from yesterday in beta labs, batch size is still 5 so events are entered in batches of 5 [20:50:56] Krenair: i will be out for a bit but i can help you when i am back if you are having troubles [20:53:00] yeah, we've had a lot more events go in than 5 [20:53:19] last event in the DB is still from 19:19:56, it's now 20:53 [20:57:20] Analytics, Mobile-Web: Instrument “tags” and anonymous gather pages to track engagement with browse. - https://phabricator.wikimedia.org/T94744#1171999 (JKatzWMF) [21:00:55] Analytics-Kanban, Analytics-Wikimetrics, Community-Wikimetrics, Patch-For-Review: Utf-8 names on json reports appear as unicode code points: "\u0623\u0645\u064a\u0646" - https://phabricator.wikimedia.org/T93023#1172022 (kevinator) It looks like my JSON browser plugin, and Apple's TextEdit are messi... [21:05:11] (PS6) Mforns: Add support for wiki explosion and others. [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 [21:06:32] (CR) Mforns: "Patch 6 fixes a bug in updating the timeboxed reports for the last date" [analytics/limn-mobile-data] - https://gerrit.wikimedia.org/r/200239 (owner: Mforns) [21:44:35] (PS12) Mforns: Implement A/B comparison [analytics/dashiki] - https://gerrit.wikimedia.org/r/198169 (owner: Milimetric) [21:47:02] (PS5) Mforns: Adding wikitext analysis to queries [analytics/limn-edit-data] - https://gerrit.wikimedia.org/r/200159 (owner: Milimetric) [22:48:42] Krenair: back, let me know if you need help