[14:18:00] o/ [15:04:00] Analytics-Cluster, Datasets-Archiving, Datasets-Webstatscollector: Mediacounts missing top1000 files after 2016-01-01: rsync fails - https://phabricator.wikimedia.org/T122864#1924165 (Nemo_bis) [15:20:07] Analytics-Backlog: MobileWikiAppDailyStats should not count Googlebot - https://phabricator.wikimedia.org/T117631#1924212 (Tbayer) BTW, it seems like we could test this ourselves using the "Fetch as Google" feature: https://support.google.com/webmasters/answer/6178089 [15:52:05] o/ [16:28:32] so lonelyyyyy.... [16:28:47] nuria: are we doing tasking today? Given that Joseph, Andrew and Marcel (?) May not be around [16:29:12] Awww [16:29:27] madhuvishy: if it is just both of us we will skip it [16:30:28] Dan and Luca are around [16:32:59] madhuvishy: let's see at standup then, cause if we are at least 3 I think we should have tasking [16:33:04] or *some tasking [16:33:09] Okay [16:36:47] (CR) Nuria: Adding readme with install instructions (1 comment) [analytics/limn-analytics-data] - https://gerrit.wikimedia.org/r/261592 (https://phabricator.wikimedia.org/T122626) (owner: Nuria) [16:38:53] (PS3) Nuria: Adding readme with install instructions [analytics/limn-analytics-data] - https://gerrit.wikimedia.org/r/261592 (https://phabricator.wikimedia.org/T122626) [16:41:26] (PS4) Nuria: Adding readme with install instructions [analytics/limn-analytics-data] - https://gerrit.wikimedia.org/r/261592 (https://phabricator.wikimedia.org/T122626) [16:48:43] hi nuria [16:48:50] milimetric: holaaa [16:48:59] I don't know much except what I learned from Joseph [16:49:09] wanna batcave? [16:49:39] (re: your last email with the pageview_hourly job failing) [16:51:06] here I am (at the office!) [16:54:33] Analytics-Cluster, Analytics-Kanban: Research whether no cookie header numbers improve Last access uniques {bear} [13 pts] - https://phabricator.wikimedia.org/T115350#1924383 (Nuria) [16:54:35] Analytics-Cluster, Analytics-Kanban: Peer review (with research) of methodology of last access calculations - https://phabricator.wikimedia.org/T121534#1924381 (Nuria) Open>Resolved [17:09:10] Hey ! How was the snow (for those who went ?) [17:20:08] Analytics-Cluster, Analytics-Kanban, Epic: {bear} Last Access Counts - https://phabricator.wikimedia.org/T88647#1924466 (Nuria) [17:23:15] milimetric: I assigned this one to you: https://phabricator.wikimedia.org/T98058 [17:23:28] milimetric: but if you create a new one it can be closed i guess [17:39:19] I am having trouble logging into stat1002. Would someone be able to help me? [17:40:46] Analytics, ContentTranslation-Analytics, MediaWiki-extensions-ContentTranslation, Ops-Access-Requests, and 2 others: access for amire80 to stat1002.eqiad.wmnet - https://phabricator.wikimedia.org/T122524#1924610 (Nuria) I think either 1003 or 1002 work to access analytics slaves thus this ticket... [17:42:45] Analytics-Kanban: Gather phabricator metrics for quaterly review [3] - https://phabricator.wikimedia.org/T122333#1924630 (Nuria) [17:44:58] Analytics-Kanban: Put quaterly review deck together - https://phabricator.wikimedia.org/T122334#1924658 (Nuria) a:Nuria [17:52:19] EGalvez_: what error are you getting? [17:52:57] Permission denied (publickey) / ssh_exchange_identification: Connection closed by remote host [17:53:12] typing in ssh stat1002.eqiad.wmnet on terminal [17:53:23] EGalvez_: you do not have permits then [17:53:32] weird.. I had permission before.. [17:53:32] EGalvez_: did you ever filed a request for access? [17:53:36] I did [17:53:54] EGalvez_: how long ago? [17:54:31] EGalvez_: you might need to request access again: https://wikitech.wikimedia.org/wiki/Requesting_shell_access [17:54:33] last quarter [17:54:49] EGalvez_: what is your user [17:54:50] ? [17:54:53] Chedasaurus [17:55:00] ssh @stat1002.eqiad.wmnet [17:55:00] https://phabricator.wikimedia.org/T113302 [17:55:04] try that [17:55:24] EGalvez_: make sure you are using the user with which you have access [17:57:23] @nuria I'm getting: ssh: Could not resolve hostname stat1002.eqiad.net: nodename nor servname provided, or not known [17:57:53] EGalvez_: try: ssh @stat1002.eqiad.wmnet [17:58:07] EGalvez_: note your hostname is wrong [18:00:21] elukey, madhuvishy , milimetric : i am in batcave [18:02:10] Its giving me the same error - madhu emailed me a suggesting I am trying out as well right now. Will get back if it works [18:08:22] Analytics-Kanban, Analytics-Wikimetrics: Use fabric to deploy wikimetrics {dove} [13 pts] - https://phabricator.wikimedia.org/T122228#1924778 (madhuvishy) [18:12:42] Analytics-Kanban: Gather phabricator metrics for quaterly review [3 pts] - https://phabricator.wikimedia.org/T122333#1924794 (madhuvishy) [18:13:04] Nuria - it worked! Thanks for the help [18:13:34] Analytics-Kanban: Put quaterly review deck together [5] - https://phabricator.wikimedia.org/T122334#1924795 (Nuria) [18:13:47] Madhu had me add my key again using ssh-add ~/.ssh/insert your prod public key file name (in case someone else asks) [18:14:02] Analytics-Kanban: Put quaterly review deck together [5 pts] - https://phabricator.wikimedia.org/T122334#1924796 (madhuvishy) [18:16:15] Analytics-Kanban, Reading-Admin: Visualization of Browser data to substitute current reports on wikistats {lama} - https://phabricator.wikimedia.org/T118329#1924802 (madhuvishy) [18:17:04] Analytics-Kanban: Piwik beacon on prod instance should be accessible - https://phabricator.wikimedia.org/T123260#1924803 (Nuria) NEW a:Milicevic01 [18:17:14] nuria: wrong mili ;) [18:17:52] Analytics-Kanban: Piwik beacon on prod instance should be accessible - https://phabricator.wikimedia.org/T123260#1924803 (Nuria) [18:18:33] Analytics-Backlog: Move IOS team piwiki usage to production instance - https://phabricator.wikimedia.org/T123262#1924821 (Nuria) NEW [18:18:47] Analytics-Backlog: Move IOS team piwiki usage to production instance - https://phabricator.wikimedia.org/T123262#1924832 (Nuria) p:Triage>Normal [18:19:45] Analytics-Kanban: Piwik beacon on prod instance should be accessible - https://phabricator.wikimedia.org/T123260#1924834 (Nuria) a:Milicevic01>Milimetric [18:20:02] Analytics-Kanban: Piwik beacon on prod instance should be accessible [3 pts] - https://phabricator.wikimedia.org/T123260#1924836 (madhuvishy) [18:20:18] Analytics-Kanban: Piwik beacon on prod instance should be accessible [5 pts] - https://phabricator.wikimedia.org/T123260#1924803 (madhuvishy) [18:20:49] Analytics-Kanban: Add piwiki beacon to financial report website - https://phabricator.wikimedia.org/T123263#1924838 (Nuria) NEW [18:21:10] Analytics-Kanban: Add piwiki beacon to financial report website - https://phabricator.wikimedia.org/T123263#1924846 (Nuria) The code is already on gerrit https://gerrit.wikimedia.org/r/#/c/254168/ [18:22:19] Analytics-Kanban: Add piwiki beacon to financial report website - https://phabricator.wikimedia.org/T123263#1924847 (Nuria) Where is this deployed? [18:23:10] Analytics-Kanban: cassandra backfill monitoring [0 pts] {slug} - https://phabricator.wikimedia.org/T115360#1924849 (madhuvishy) [18:25:00] Analytics-Kanban: Add piwiki beacon to financial report website [5] - https://phabricator.wikimedia.org/T123263#1924852 (Nuria) [18:36:47] Analytics-Kanban: Evaluate performance of country breakdown of last access monthly unique numbers - https://phabricator.wikimedia.org/T123265#1924872 (Nuria) NEW a:Nuria [18:39:16] Analytics-Kanban: Evaluate performance of country breakdown of last access monthly unique numbers {bear} [5 pts] - https://phabricator.wikimedia.org/T123265#1924884 (madhuvishy) [18:41:01] Analytics-Backlog: Productionize last access jobs for daily and monthly calculations {bear} - https://phabricator.wikimedia.org/T122514#1924890 (madhuvishy) [18:45:04] Analytics-Kanban: Provide weekly app session metrics separately for Android and iOS - https://phabricator.wikimedia.org/T117615#1924910 (madhuvishy) [18:45:06] Analytics-Kanban: Move App session data to 7 day counts - https://phabricator.wikimedia.org/T117637#1924909 (madhuvishy) [18:48:03] Analytics-Kanban: Provide weekly app session metrics separately for Android and iOS, and move to 7 day counts. - https://phabricator.wikimedia.org/T117615#1924933 (madhuvishy) [18:53:48] Analytics-Kanban: Provide weekly app session metrics separately for Android and iOS, and move to 7 day counts [13 pts] - https://phabricator.wikimedia.org/T117615#1924959 (madhuvishy) [18:59:20] Analytics-Kanban, operations, HTTPS: EventLogging sees too few distinct client IPs {oryx} [8 pts] - https://phabricator.wikimedia.org/T119144#1924970 (madhuvishy) [18:59:23] Analytics-Kanban, operations, HTTPS: EventLogging sees too few distinct client IPs [8] - https://phabricator.wikimedia.org/T119144#1924971 (Nuria) [19:08:17] Analytics-Kanban, operations, HTTPS: EventLogging sees too few distinct client IPs {oryx} [8 pts] - https://phabricator.wikimedia.org/T119144#1924985 (madhuvishy) [19:11:35] Analytics-Kanban: Reorganize oozie jobs to not use mobile cache webrequest_source - https://phabricator.wikimedia.org/T122651#1924988 (Nuria) There are no jobs running on mobile but we would need to change in ("mobile", "text") to In ("text") Plus start/stop jobs [19:49:02] if I have feature requests for the pageview api, where should I file those? [19:52:53] milimetric: ^? :) [19:53:16] legoktm: tag them "Analytics" in phabricator [19:53:36] but our backlog is huge, we're cleaning up all our historical stuff [19:54:22] legoktm: so if it's urgent it's best to ping one of us (me, Nuria, etc.) [19:54:42] not urgent, I'd just like localhost to be whitelisted for CORS access [19:55:02] I'm trying to develop something, and I really don't want to have to build it on labs just for CORS [19:55:45] oh legoktm, then you need to talk to the services team [19:56:02] The pageview API runs on a back-end RESTBase, with the "main" RESTBase interface in front of it [19:56:03] heh [19:56:14] so we don't control the CORS configuration [19:57:12] * legoktm hops IRC channels [19:57:15] that sounds like a really quick change though, I can ask them if you like [21:57:28] madhuvishy: I think my logic to split by countries needs work cause numbers do not match [21:57:44] madhuvishy: let me show you my code via CR on your patch Ok? [21:57:52] nuria: sure [21:58:35] (PS9) Nuria: [WIP] Daily and monthly uniques oozie jobs [analytics/refinery] - https://gerrit.wikimedia.org/r/216341 (https://phabricator.wikimedia.org/T92977) (owner: Madhuvishy) [21:59:10] madhuvishy: take a look at this one: https://gerrit.wikimedia.org/r/#/c/216341/9/oozie/last_access_uniques/monthly/last_access_uniques_monthly.hql [22:01:15] madhuvishy: ops 1 sec [22:01:53] (PS10) Nuria: [WIP] Daily and monthly uniques oozie jobs [analytics/refinery] - https://gerrit.wikimedia.org/r/216341 (https://phabricator.wikimedia.org/T92977) (owner: Madhuvishy) [22:45:45] milimetric: MobileWebSectionUsage [23:25:07] Analytics-Backlog, ContentTranslation-Analytics, MediaWiki-extensions-ContentTranslation, operations: schedule a daily run of ContentTranslation analytics scripts on terbium - https://phabricator.wikimedia.org/T122479#1926062 (Dzahn) [23:25:12] Analytics, ContentTranslation-Analytics, MediaWiki-extensions-ContentTranslation, Ops-Access-Requests, and 2 others: access for amire80 to stat1002.eqiad.wmnet - https://phabricator.wikimedia.org/T122524#1926059 (Dzahn) Open>Resolved a:Dzahn thanks for the confirmation. resolving [23:25:33] Analytics, ContentTranslation-Analytics, MediaWiki-extensions-ContentTranslation, Ops-Access-Requests, operations: access for amire80 to stat1002.eqiad.wmnet - https://phabricator.wikimedia.org/T122524#1926063 (Dzahn) [23:28:17] milimetric: nuria do we have a recommended maximum number of events for EL data that is consumed by mysql [23:28:40] madhuvishy: total? for all schemas? [23:28:45] no, per schema [23:28:53] if i was a schema owner [23:29:01] how many events would i want to send [23:29:28] madhuvishy: answer is 'as few as possible to gather the data you need" [23:30:03] nuria: :) well having some heuristics would be good i think [23:30:04] madhuvishy: which might not be very helpful [23:30:19] madhuvishy: depends on the nature of the data right? [23:30:36] madhuvishy: some data needs to be sampled and combined with other data [23:30:41] other stands on its own [23:30:51] the idea is noit to think of volume but [23:30:53] *not [23:31:03] what do you want to measure? [23:31:12] nuria: no - so we have a schema that's been sending over 100 events a second [23:31:16] madhuvishy: and if needed calculate powers for those numbers [23:31:17] for the last 3 or so weeks [23:31:34] which has been tampering with our mysql replication [23:31:42] madhuvishy: that is pretty crazy, with that volume of data yiou cannot hope to look at it on mysql [23:31:45] right [23:31:59] madhuvishy: but that should be blacklisted untill flow is reduced [23:31:59] i want to know a good number for being able to look at it in mysql [23:32:13] yeah I'm talking to mobile to work that out [23:32:51] and jdlrobson was asking what's a good max events/sec to have [23:33:10] madhuvishy: I rather think from the "measurement" stand point that mysql. Let's start with "what do we want to measure?" [23:33:32] madhuvishy: if you send them all to schema for a months at end the table eventually will be usable [23:33:52] madhuvishy: so even if we given a number today it doesn't mean you can query your table in 6 months [23:34:00] makes sense? [23:34:03] nuria: mm hmm [23:34:07] okay [23:34:34] madhuvishy: so, if it is hindering operations i would 1) blacklist 2) notify team [23:34:47] 3) get together to see what are we trying to measure [23:35:11] madhuvishy: we have no option when it comes to blacklisting if the system is usable for other users [23:35:40] nuria: no i think it was a bug they introduced 3 weeks back - they din't want to collect this volume [23:35:51] they are gonna lower sampling rate [23:35:52] madhuvishy: ya, that is normally the case [23:36:13] madhuvishy: this is happen before [23:36:45] madhuvishy: so let's blacklist until sample rate is changed and maybe drop data collected at too high rate right? [23:38:09] madhuvishy: makes sense? [23:38:30] madhuvishy: i can do the blacklisting change (unless smpale rate is changed already) [23:38:32] *sample [23:38:48] nuria: yup i was checking on #wikimedia-mobile [23:38:52] we are a go to blacklist [23:38:56] madhuvishy: ok [23:39:00] dan is pushing the change [23:39:03] ok [23:40:09] nuria: https://gerrit.wikimedia.org/r/263545 [23:40:56] lol, nuria ori hates +1 as I just found out last week :) [23:41:23] madhuvishy: i am powerless when it comes to puppet [23:41:29] madhuvishy: i can only =1 [23:41:31] +1 [23:41:45] which ahem ... gives me piece of mind in a way [23:41:50] nuria: yeah [23:41:56] we'll get someone here to merge [23:43:44] YuviPanda: can you merge https://gerrit.wikimedia.org/r/#/c/263545/ [23:44:08] hey [23:44:10] sure [23:44:45] madhuvishy: do you want me to force a puppet run somewhere? [23:44:58] YuviPanda: nah, it can wait 20 minutes [23:45:03] ok [23:45:08] thanks :) [23:46:01] np [23:46:04] madhuvishy: i would say that a "good" eyeball rule is that [23:46:22] madhuvishy: let's say a table becomes unusable after 2 million records [23:46:38] madhuvishy: so 5 per sec puts you there in over 6 months [23:46:55] seems a sensible limit for data you want to get to fast [23:48:15] jdlrobson: ^ [23:48:41] thanks nuria that's super useful [23:49:07] jdlrobson: all right, let's go with that [23:49:43] nuria: we do drop data that's over 90 days old [23:50:02] so if we do that then may 10 events per second would be alright? [23:50:37] madhuvishy: that makes sense yeah, for a rule of thumb, let me write it up [23:50:39] * elukey looks at the code review [23:50:43] nuria: https://www.mediawiki.org/wiki/Reading/Web/EventLogging_best_practices#Sampling_rate [23:51:14] jdlrobson: niceeee!!!! [23:51:17] thank you [23:51:59] nuria: this is in reading web's docs :) we should have something similar in our general docs too [23:52:22] but this is super useful [23:52:55] madhuvishy: hear hear, was looking for suitable place [23:53:06] madhuvishy: the trick is that EL extension is documented on mediawiki too [23:53:12] yeah [23:54:06] madhuvishy: will do tomorrow, need to leave now, ciao [23:54:30] nuria: cool!