[01:44:46] 10Analytics-Kanban, 10Patch-For-Review: Replace references to dbstore1002 by db1047 in reportupdater jobs - https://phabricator.wikimedia.org/T176639#3632150 (10Nuria) Industry terms such as master/slave are unambiguous, they allow us to express simply concepts that otherwise will take quite a few words. If... [04:40:53] joal: ottomata: The hive roles are in reality applied in both notebook1001 and 2, yuvi removed it from 1001 puppet at some point so we can have separation, but we never got around to actually purging everything from it, I should probably apply it back in 1001 [04:41:14] there is no docker stuff in SWAP [04:41:37] each user's notebook is a systemd unit [04:41:58] uses https://github.com/jupyterhub/systemdspawner [05:01:06] fixed https://gerrit.wikimedia.org/r/#/c/380914/ [05:30:33] https://www.irccloud.com/pastebin/7EdrryVU/ [05:30:58] afaict, the connection succeeds at the tcp level, but things like curl/requests.get at the application fail [05:31:05] both from cmd line and from notebook [05:31:25] notebooks don't have any special networking in them, and also have the right proxy settings [05:31:55] from cmd line [05:31:58] https://www.irccloud.com/pastebin/zyXMsyb7/ [05:32:28] sorry for wall of text, off to sleep now :) ping me if you figure out what's up! [05:39:15] 10Analytics-Cluster, 10Analytics-Kanban, 10Patch-For-Review, 10User-Elukey: Audit users and account expiry dates for stat boxes - https://phabricator.wikimedia.org/T170878#3638481 (10elukey) No answer from @dchen and @dduvall, but since they are WMF staff and they might be on vacation I am not inclined to... [05:39:48] 10Analytics-Cluster, 10Analytics-Kanban, 10Patch-For-Review, 10User-Elukey: Audit users and account expiry dates for stat boxes - https://phabricator.wikimedia.org/T170878#3638482 (10elukey) [06:41:45] hello team, I am going to be afk for an hour (hopefully less) in about 20mins for some errands [06:41:52] will be reachable by phone of course [06:50:29] 10Analytics-Cluster, 10Analytics-Kanban, 10Patch-For-Review, 10User-Elukey: Audit users and account expiry dates for stat boxes - https://phabricator.wikimedia.org/T170878#3638532 (10elukey) >>! In T170878#3631676, @GoranSMilovanovic wrote: > Hi, I need continuing access to the stat boxes beyond the set ex... [06:50:45] 10Analytics-Cluster, 10Analytics-Kanban, 10Patch-For-Review, 10User-Elukey: Audit users and account expiry dates for stat boxes - https://phabricator.wikimedia.org/T170878#3638533 (10elukey) [10:55:14] * elukey lunch! [11:37:29] PROBLEM - HDFS active Namenode JVM Heap usage on analytics1001 is CRITICAL: CRITICAL: 61.02% of data above the critical threshold [3891.2] [11:38:30] RECOVERY - HDFS active Namenode JVM Heap usage on analytics1001 is OK: OK: Less than 60.00% above the threshold [3686.4] [11:47:39] PROBLEM - HDFS active Namenode JVM Heap usage on analytics1001 is CRITICAL: CRITICAL: 61.02% of data above the critical threshold [3891.2] [11:50:29] joal: o/ - is there a way to see the full hive query running from yarn ? (or the cli) [11:50:38] elukey: Yessr [11:50:44] I can see a huge query running but not its details [11:50:45] elukey: tricky but possible :) [11:50:47] ahhaah [11:50:55] elukey: got to the job UI [11:51:22] elukey: not the application master, the job (https://yarn.wikimedia.org/proxy/application_1504006918778_108183/mapreduce/job/job_1504006918778_108183 [11:51:42] yep yep [11:51:59] elukey: on the left, find "Configuration", then in search box type: "hive.query" - You have it [11:52:43] ahhh it is paginated! I got that far but didn't see anything useful [11:52:45] thanksss [11:53:06] using search is cool in those ui [11:53:27] from wmf.webrequest where year = 2017 and month = 9 [11:53:31] grrrr [11:53:47] yes ... Particularly with limit 10 [11:53:52] this is insane [11:54:36] elukey: we need to teach people about data size is all [11:55:00] By the way elukey - I think for yesterday queries from gwicke we could have suggested samplin [11:55:07] It would have worked correctly [11:55:18] sure, but even querying in small batches [11:55:43] elukey: Do we have an idea who is phuedx? [11:55:53] I was about to ask, lemme check [11:58:39] RECOVERY - HDFS active Namenode JVM Heap usage on analytics1001 is OK: OK: Less than 60.00% above the threshold [3686.4] [11:59:26] he is joining the chan [12:00:47] o. [12:00:49] *o/ [12:00:59] \o [12:01:19] Sorry to bother phuedx but we should have a talk abou cluster resource usage :) [12:01:30] firstly, sorry y'all -- i just opened up the query to a month without thinking [12:01:33] joal: shoot [12:01:51] phuedx: 1 month of webrequesr is ~35Tb [12:02:00] ... [12:02:16] So ... requesting that amount of data doesn't go unnoticed [12:02:20] * phuedx resists the urge not to swear [12:03:34] :D [12:03:48] phuedx: 3 things to usually think of when querying webrequest: What time-span (shorter == smaller), what source (text, upload, both), is sampling an option [12:04:01] holy hell, what did i take down? :( [12:04:35] My assumption is that, since you're after specific url patterns, sampling might not be an option (I might be wrong depending of the frequency of your pattern) [12:04:51] nothing is down! Don't worry, just a bit of business here and there :) [12:05:28] Then I think you're only after webrequest_source = 'text', not 'upload' - This will reduce data size to read by ~2 [12:05:29] PROBLEM - Hadoop HistoryServer on analytics1001 is CRITICAL: PROCS CRITICAL: 0 processes with command name java, args org.apache.hadoop.mapreduce.v2.hs.JobHistoryServer [12:06:00] Finally doing it on an hour first, then maybe a day if you don't have what you want, then possibly a week if again, and finally come to us before going for the month one :) [12:06:17] phuedx: as elukey said, nothing major, just resource shring good practices :) [12:06:47] phuedx: We can't expect you to know if we have not told :) [12:06:57] joal: thinking about it more, i think two weeks should've been enough [12:07:12] joal: otoh it's nice when sensible folk exercise a bit of common sense ;) [12:07:23] phuedx: going incremental here makes sense - What if you'd have had enough with a day ? [12:08:01] phuedx: This is the burden of magical clstering tools - everything seems feasible :) [12:08:02] yeah, you're right [12:08:09] lol [12:08:39] thanks for the points both [12:08:42] *pointers [12:16:42] joal: I am wondering if we should consider raising up the xmx values [12:16:48] on an1001's daemons [12:16:48] hm [12:17:00] elukey: I think we should wait for deletion time to come [12:17:21] elukey: problem is HDFS namenode - too busy because of too many data to handle I think [12:17:35] elukey: When we'll have clean up, we should be better (hopefully) [12:17:50] elukey: But otherwise, I'll follow you blindly if you think [12:19:06] joal: case in point -- same order of results but lower numbers for the hour query! [12:19:13] lesson learned [12:19:38] thanks phuedx! [12:19:42] joal: https://grafana.wikimedia.org/dashboard/db/analytics-hadoop?panelId=4&fullscreen&orgId=1&from=now-6M&to=now [12:19:45] phuedx: webreques is so big that usually getting a day (sometimes an hour) is representative [12:20:02] i just had no idea of the size of the data that we're storing [12:20:11] what's the total size of the cluster? [12:20:17] phuedx: and we're just talking 1 month here ;) [12:20:27] We have 1.5Pb I think [12:20:33] wowzers [12:20:37] https://grafana.wikimedia.org/dashboard/db/analytics-hadoop?panelId=25&fullscreen&orgId=1&from=now-6M&to=now [12:20:41] phuedx: --^ [12:20:58] data is also replicated 3 times across nodes [12:20:58] Well actually 1.7 [12:35:24] joal: I'd have a plan to propose, let me know if it makes sense or not :) [12:35:47] from https://grafana.wikimedia.org/dashboard/file/server-board.json?var-server=analytics1001&refresh=1m&orgId=1 the hadoop master hosts seem to have a ton of resources not utilized [12:36:19] so I would try a Xmx6G for the namenodes in this way: [12:36:46] 1) restart the namenode in standby with the new Xmx (an1002) and watch its behavior [12:37:04] 2) if all is good, manually failover an1001's namenode to 1002 [12:37:12] observe its behavior [12:37:21] if all is good, make the change to 1001 [12:37:23] otherwise rollback [12:38:57] rationale: [12:39:20] 1) raising xmx to 6g from the actual 4g should not be a big deal for the GC [12:40:05] 2) for some reason we are using almost all the heap size for the namenode and unusual old gen gc collections patterns started to show up [12:40:30] 3) I'd prefer to deal with this issue in a controlled environment rather than getting a page for Java OOM [12:52:42] elukey: works for me [12:53:07] elukey: I can watch behavior and metricswhile you do [12:53:15] super, doing it [12:57:17] joal: restarting 1002 now [13:01:06] so far so good, going to wait ~15 min and then I'll do the failover [13:09:43] elukey: is historyserver death normal? [13:11:23] joal: the one that happened one hour ago? [13:12:27] java.lang.OutOfMemoryError: GC overhead limit exceeded [13:12:27] Halting due to Out Of Memory Error... [13:12:27] 17/09/27 12:02:19 FATAL yarn.YarnUncaughtExceptionHandler: Thread Thread[IPC Server idle connection scanner for port 10033,5,main] thre [13:12:30] w an Error. Shutting down now... [13:12:33] java.lang.OutOfMemoryError: GC overhead limit exceeded [13:12:35] wonderful [13:12:43] :( [13:13:27] ah snap it died and didn't come back??? [13:14:38] !log restart mapreduce history server on analytics1001 after crash (java.lang.OutOfMemoryError: GC overhead limit exceeded) [13:14:48] Logged the message at https://www.mediawiki.org/wiki/Analytics/Server_Admin_Log [13:14:51] RECOVERY - Hadoop HistoryServer on analytics1001 is OK: PROCS OK: 1 process with command name java, args org.apache.hadoop.mapreduce.v2.hs.JobHistoryServer [13:15:09] what the hell I didn't realize that [13:15:10] sigh [13:15:13] * elukey cries in a corner [13:17:15] !log manual failover of HDFS namenode from an1001 to an1002 to test 6G max heap size [13:17:21] Logged the message at https://www.mediawiki.org/wiki/Analytics/Server_Admin_Log [13:18:44] * joal hugs elukey [13:22:48] * elukey hugs joal [13:23:08] metrics looks good, will wait 10/15 mins and then I'll restart 1001 and failover to it again [13:40:29] !log manual failover of HDFS namenode from an1002 to an1001 [13:40:33] Logged the message at https://www.mediawiki.org/wiki/Analytics/Server_Admin_Log [13:40:35] and we are done [13:44:47] 10Analytics-Kanban, 10Analytics-Wikistats, 10Patch-For-Review: Implement some example metrics as Druid queries - https://phabricator.wikimedia.org/T170882#3446060 (10Milimetric) [13:44:53] 10Analytics-Kanban, 10Analytics-Wikistats, 10Patch-For-Review: Implement some example metrics as Druid queries - https://phabricator.wikimedia.org/T170882#3446060 (10Milimetric) [13:45:12] 10Analytics-Kanban, 10Analytics-Wikistats: Implement Topic Selector Widget - https://phabricator.wikimedia.org/T167676#3639640 (10Milimetric) a:05mforns>03Milimetric [13:45:22] (03PS1) 10Addshore: instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380974 (https://phabricator.wikimedia.org/T176577) [13:46:15] (03PS1) 10Addshore: instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380975 (https://phabricator.wikimedia.org/T176577) [13:46:19] (03CR) 10Addshore: [C: 032] instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380975 (https://phabricator.wikimedia.org/T176577) (owner: 10Addshore) [13:46:23] (03CR) 10Addshore: [C: 032] instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380974 (https://phabricator.wikimedia.org/T176577) (owner: 10Addshore) [13:46:28] (03Merged) 10jenkins-bot: instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380975 (https://phabricator.wikimedia.org/T176577) (owner: 10Addshore) [13:46:31] (03Merged) 10jenkins-bot: instanceof.php talk to wdqs1003:8888 directly for now [analytics/wmde/scripts] - 10https://gerrit.wikimedia.org/r/380974 (https://phabricator.wikimedia.org/T176577) (owner: 10Addshore) [13:46:55] if anyone ever gets tired of the spam from the wmde repos I can removethem from this channel :) [13:49:35] no issue at all :) [13:53:24] coolio! [14:01:40] yooo where's everybody? a-team [14:01:50] Hi fdans :) [14:02:01] oh hold on [14:02:05] I'm an idiot [14:02:14] ahahaha [14:02:15] You're not an idiot, you're just a bit early :) [14:02:30] my laptop is still on spanish time, I'm in London [14:02:33] jeeeesus [14:03:10] actually this is great, I can buy a power adaptor [14:03:12] fdans: Since you're in london I might ask for a favor :) [14:03:20] joal: tell me [14:03:47] There is this tea brand my wife loves and we don't find in france [14:04:08] hahah of course [14:04:22] fdans: if you have some free time as a few bucks I could borrow, maybe a little posting could be done? [14:04:25] :D [14:04:46] joal: consider it done, what brand is it? [14:05:02] pass me your address by emilio [14:05:58] hm fdans - I miss a reference here I think - emilio? [14:06:24] it's just a stupid spanish 90s way of saying "email" [14:06:58] Ah ! [14:07:05] Yes, will do [14:12:37] sent fdans - Let me know if here's any info missing [14:16:01] joal: ohh, there's one shop in Piccadilly, I can definitely drop by and send you the stash [14:33:10] joal: I am doing a delicate ops task now, do you mind if we skip ops sync? [14:33:17] elukey: ok ! [14:42:47] joal: I am very slowly restarting the node managers on our workers, since the heap size is used a lot and we haven't been seeing a jvm updates in ages [14:42:54] it should be a transparent op [14:43:08] (2 hosts every 30 seconds) [14:43:19] 10Analytics-Tech-community-metrics, 10Developer-Relations (Jul-Sep 2017): Automatically sync mediawiki-identities/wikimedia-affiliations.json DB dump file with the data available on wikimedia.biterg.io - https://phabricator.wikimedia.org/T157898#3639876 (10Aklapper) Not sure yet https://github.com/grimoirelab/... [14:45:31] !log rolling restart of all the Yarn nodemanager daemons on analytics1028-1068 (ease heap consumption pressure, seamless restart) [14:45:40] Logged the message at https://www.mediawiki.org/wiki/Analytics/Server_Admin_Log [14:47:59] (quick coffee before standup) [14:59:27] a-team: was going to join standup but i think my wifi is kaput [15:00:35] milimetric: i think i got it one sec [15:10:53] i’m so sorry a-team my laptop had a kernel panic and now it’s taking ages to reboot [15:10:54] (writing this from phone) [15:18:02] !log Kill/restart stuck jobs [15:18:06] Logged the message at https://www.mediawiki.org/wiki/Analytics/Server_Admin_Log [15:22:01] o/ what are the possible values of wmf.webrequest.agent_type -- i'd like to filter out bots/spiders from a hive query [15:22:18] couldn't see them documented on https://wikitech.wikimedia.org/wiki/Analytics/Data_Lake/Traffic/Webrequest [15:25:07] joal: also, reading the scrollback, if you need any london-based wares bringing over to the all hands, then just lmk :) [15:25:13] phuedx: anything not agent_type="user" is a bot [15:25:27] nuria_: thanks [15:25:40] phuedx: now, we estimate that maybe 5% of user marked traffiq are also bots so be aware [15:25:49] phuedx: do you know http://pivot.wikimedia.org [15:26:05] phuedx: that might be a better way to explore data and filter if you have not looked ta this before [15:26:11] *at this dataset [15:26:41] nuria_: that's awesome! thanks [15:26:41] phuedx: depends what you are looking for of course [15:27:23] phuedx: let us know if we can help , do you have a ticket you are working on? [15:28:56] nuria_: mibad i now see "user or spider" in the column description [15:29:05] phuedx: ya, np [15:29:09] nuria_: one sec, i'll dig out the query that i'm writing [15:29:35] phuedx: ok, what is what you are trying to dig deeper on? [15:30:14] nuria_: context: https://phabricator.wikimedia.org/T176467 [15:30:46] we're interested in the browsers that are being used to download pdfs [15:31:34] phuedx: as joal mentioned restricting your queries to 1 day and "upload" will help [15:31:41] yup :) [15:31:43] heh [15:32:12] phuedx: for a 1st exploration [15:32:29] phuedx: we have a dataset in pivot that actually will help you [15:32:44] ! [15:33:00] phuedx: it is sampled and used for ops (what means it will under report small usage browsers but it should be plenty) [15:33:11] phuedx: i think that one contains upload and text both [15:33:23] nuria_: good call on using pivot for exploration :) [15:33:43] phuedx: last 7 days of webrequest data: https://pivot.wikimedia.org/#webrequest/totals/2/EQUQLgxg9AqgKgYWAGgN7APYAdgC5gQAWAhgJYB2KwApgB5YBO1Azs6RpbutnsEwGZVyxALbVeAfQlhSY4AF9kwYhBkc86FWs7AKVOoxZt1XTDnxEylJQaat2nbub7VBS4XPwiFSrQ43Kqv74MmIASsTkAObiSgAmAK4MxNq8AAoAjAAiVMxg1OYAtBnypYoA2gC6yOQJADZ1SoSkYMwo5cDNrcDVVTX1dUA [15:39:54] joal: staff? [15:47:28] a-team: now google kicked me out [15:47:39] want us to call your phone? [15:47:41] milimetric: can you call me, this is last try [16:04:19] (03CR) 10Nuria: [V: 032 C: 032] Add oozie jobs loading druid monthly uniques [analytics/refinery] - 10https://gerrit.wikimedia.org/r/348052 (https://phabricator.wikimedia.org/T159471) (owner: 10Joal) [16:05:29] * elukey off! [16:05:30] my internet's sporadic, a-team, I'm on my phone hotspot, just call me if you have any emergency and I'm not responding [16:05:33] o/ [16:14:10] 10Analytics-Kanban, 10Analytics-Wikistats: Pageview retrieval does not work if one of the fails requests - https://phabricator.wikimedia.org/T176261#3640155 (10Nuria) a:03mforns [16:14:29] mforns: I have assigned you a bugfix on wikistats: https://phabricator.wikimedia.org/T176261 [16:14:44] 10Analytics-Kanban, 10Analytics-Wikistats: Pageview retrieval does not work if one of the fails requests - https://phabricator.wikimedia.org/T176261#3619243 (10Nuria) [16:16:32] 10Analytics-Kanban, 10Analytics-Wikistats: Wikistats metrics should link to corresponding page in meta - https://phabricator.wikimedia.org/T176241#3640159 (10Nuria) a:03fdans [16:16:52] fdans: i have also assigned you a small wikistats task [16:17:32] nuria_: that makes sense [16:17:57] fdans: ok, you can take a look at dashiki config , for ex: [16:19:34] fdans: we still have to deploy the changes for copy for "unique devices " to be "monthly unique devices" [16:19:36] right? [16:20:19] nuria_: sorry, my memory is betraying me on that one [16:20:32] do we have a task for it? [16:20:44] fdans: https://phabricator.wikimedia.org/T176240 [16:21:17] ooooo you know what happens nuria_ ? [16:21:28] fdans: aham [16:21:32] I thought this task was about specifying the current granularity [16:21:38] not about this specific metric [16:23:11] fdans: well there might be two things: 1) if the dashboard displays by default "monthly unique devices" the copy should say that [16:23:36] fdans: i think this is where we stand now as there are no breakdowns possible for that metric on time granularity on ui [16:23:54] 10Analytics: Making geowiki data public - https://phabricator.wikimedia.org/T131280#3640172 (10Ijon) Thank you, @Nuria, for the clear update. Do we have a solid solution for the concerns your team has had about leaking a user's country due to insufficient bucketing? I'd like some assurance that come Q3, the fe... [16:24:00] fdans: but also 2) add breakdowns daily and monthly for metric (that should be another task) [16:25:01] nuria_: I'm not sure I follow the second point... do you have a minute to batcave? [16:25:09] oh nvm your connection problems [16:25:32] fdans: i have a meeting with dan but no worries teh bug is about fixing 1) [16:25:42] fdans: is that one more clear? [16:26:36] nuria_: ok, yeah that's no problem, the only thing is where does the daily granularity that we're offering for 1 and 3 month time range, stand here [16:28:12] fdans: that metric cannot be reported on 3 month granularity though [16:28:21] fdans: can only be reported monthly [16:29:24] soooo... why does the api have a daily option? [16:29:56] oh, right, it's not about that, it just doesn't make sense to report it [16:30:51] 10Analytics: Making geowiki data public - https://phabricator.wikimedia.org/T131280#3640202 (10Nuria) @Ijon think after our implementation of editing metrics that we hope to launch beginning on Q2 as part of the new wikistats 2.0 we are going to have more options as to how to compute this data with a level of... [16:31:05] fdans: ok, there are two options: daily and monthly [16:31:12] fdans: sorry it was not clear [16:31:36] fdans: but monthly option cannnot be aggregated on a 3 month timeframe [16:32:24] fdans: it can be reported distinctively for 3 months in either monthly (3 values) or daily (90 values) time splits [16:32:29] fdans: makes sense? [16:32:32] oh were not giving the user that option currently [16:32:40] (agg by three months) [16:33:33] https://usercontent.irccloud-cdn.com/file/N64nmNhY/Screen%20Shot%202017-09-27%20at%2017.33.13.png [16:33:55] this is what we have for three months, on a daily granularity [16:34:29] that's why for me there's value on specifying the granularity that we're asking to the API, and add it as a subtitle [16:35:07] but maybe not manually prepending "monthly" to the title of the metric, since we're also showing daily [16:35:51] fdans: on meeting, talk later [16:36:11] *can talk later [16:36:19] yes, sorry nuria_ :) [16:36:52] fdans: np! [16:53:24] 10Analytics: Making geowiki data public - https://phabricator.wikimedia.org/T131280#3640292 (10Ijon) Thank you for the quick response, @Nuria. It sounds as though there **is** a worthwhile discussion to be had on //whether and how// we can achieve the statistics requested in this ticket. Unless I misunderstood... [17:16:31] fdans: back [17:17:25] fdans: i see, that is not aggregation, i understand, that seems correct (add it as subtitle) [17:18:16] nuria_: that’s great, I already have that change done, will push it in a bit [17:18:26] fdans: ok [17:18:50] milimetric: can you triple check my comment here makes sense: https://phabricator.wikimedia.org/T131280 [17:30:31] 10Analytics, 10Operations: rack/setup/install furud.codfw.wmnet - https://phabricator.wikimedia.org/T176506#3640394 (10RobH) a:05RobH>03faidon [17:31:13] 10Analytics, 10Operations: rack/setup/install furud.codfw.wmnet - https://phabricator.wikimedia.org/T176506#3627866 (10RobH) Ok, the two MD1200 disk shelves detect, and they are current not configured in any raid array. The OS is installed and calling into puppet, but is set to role spare for now. Assigned t... [17:38:28] 10Analytics: Add action api counts to graphite-restbase job - https://phabricator.wikimedia.org/T176785#3636597 (10GWicke) This metric would perfectly complement the REST equivalent in https://grafana.wikimedia.org/dashboard/db/api-summary?orgId=1, and as a result give us direct information on overall API use. [17:53:37] joal: am i correct that we do not have any kind of geo info in teh mw history reconstruction? [17:54:50] nuria_: yes, no geo is even possible without importing recent_changes table, which we don't [17:55:19] milimetric: and recent_changes only has info for anonymous edits rights? [18:25:25] (03PS1) 10GoranSMilovanovic: Debug Crosstabs ggrepel w. facet_wrap [analytics/wmde/WDCM] - 10https://gerrit.wikimedia.org/r/381058 [18:25:52] (03CR) 10GoranSMilovanovic: [V: 032 C: 032] Debug Crosstabs ggrepel w. facet_wrap [analytics/wmde/WDCM] - 10https://gerrit.wikimedia.org/r/381058 (owner: 10GoranSMilovanovic) [18:25:58] (03Merged) 10jenkins-bot: Debug Crosstabs ggrepel w. facet_wrap [analytics/wmde/WDCM] - 10https://gerrit.wikimedia.org/r/381058 (owner: 10GoranSMilovanovic) [18:38:01] Thanks milimetric for nuria_'s question [18:57:25] 10Analytics: Mount dumps on SWAP machines (notebook1001.eqiad.wmnet / notebook1002.eqiad.wmnet) - https://phabricator.wikimedia.org/T176091#3640736 (10diego) Installing sshfs would be also a good solution for this and for https://phabricator.wikimedia.org/T176093 [21:43:10] nuria_: sorry I missed your second part, recent_changes has info for all edits, anonymous or not, but only keeps the last (90?) days [21:54:16] nuria_: thanks for the review of https://gerrit.wikimedia.org/r/379441 ! You commented about remembering to “union” if other schema is created. Is that something I should worry about at this stage and try to future proof the SQL, or do we handle that when it becomes an issue? [21:59:03] 10Analytics, 10Analytics-Dashiki, 10Community-Tech, 10Patch-For-Review: Add draft namespace creations to page creation dashboard - https://phabricator.wikimedia.org/T176375#3641300 (10Nettrom) 05Open>03Resolved # Verified that the dataset of number of pages created is available in the [[ https://analy... [22:04:12] hi Nettrom, I'm glad nuria got to the review before me. Her comment is not something you have to worry about right now, just if you want to include results from multiple schemas in the future. The reportupdater won't care if you change the query, it even handles changing the select clause relatively gracefully. [22:06:06] 10Analytics-Kanban, 10Patch-For-Review: Replace references to dbstore1002 by db1047 in reportupdater jobs - https://phabricator.wikimedia.org/T176639#3641312 (10mpopov) @mforns: we specify the analytics-store hostname in our R package (the function that makes sql queries: https://github.com/wikimedia/wikimedia... [22:15:55] 10Analytics: Mount dumps on SWAP machines (notebook1001.eqiad.wmnet / notebook1002.eqiad.wmnet) - https://phabricator.wikimedia.org/T176091#3641343 (10DarTar) cc @Halfak for visibility [22:19:12] Nettrom: we can handle that later, just an fyi of the imperfection of schemas +json [22:44:24] milimetric & nuria_ : That’s what I thought, thanks for confirming that! And thanks again, got the dashboard running nicely now with all metrics avaialble. Really neat that once the dataset is available, everything could be handled on-wiki! [22:45:47] Nettrom: thanks for appreciating that despite the manual steps to make it work here and there. We all wish it was even more automated and smooth, so any thoughts on how it can be improved are welcome [23:25:09] Anybody know of a nice way to form a query for webrequests that gives all partitions between two timestamps? [23:25:24] The partitions are on YMDH [23:26:03] The approach I have is pretty messy... [23:31:19] Shilad: I adapted a function that Ellery wrote that inserts the necessary YMDH clauses based on a start and end date. Still not pretty, though. [23:31:50] would love to know if there are prettier solutions :)