Skip to main content

Browse the archive

https://github.com/wikimedia/operations-puppet
12 April 2024, 10:39:33 UTC
sort by:
Revision Author Date Message Commit Date
223a123 Giuseppe Lavagetto 14 August 2015, 09:10:35 UTC puppet_compiler: refresh module/role Since we rebuilt the puppet compiler, we refresh the puppet module as well. Bug: T96802 Change-Id: Ie3fc79eec44667f3bee4f5cbb9130f3826e94964 14 August 2015, 09:19:58 UTC
77787db Alexandros Kosiaris 13 August 2015, 16:43:35 UTC maps: Add ALTERs to ensure passwords CREATE USER will fail if the users already exist, so use an ALTER to ensure consistent passwords Change-Id: Ibc8c3ed5674dcb695a38031abaeb7b7034dfb629 14 August 2015, 07:38:15 UTC
6340b5d Marko Obrovac 13 August 2015, 21:05:17 UTC Tilerator: start ncpu / 2 workers By default, service::node sets the number of workers to the number of logical processors. Since Tilerator currently overloads Postres, start only half as many workers. Bug: T108974 Change-Id: I9a87e5f62d93dad7f1f05e82b89782cd200bb913 14 August 2015, 07:30:26 UTC
7d36917 Marko Obrovac 13 August 2015, 23:24:54 UTC MobileApps: Do not use the proxy to issue requests The MobileApps service issues requests only to internally-available services - the MW Action API and RESTBase - so it doesn't need a proxy at all. In fact, its proxy shouldn't be set, as placing external requests is a potential security risk. Change-Id: Ic9a44c295bf8a06242d10a9b753125bd32c978c6 14 August 2015, 07:22:52 UTC
60148cf Marko Obrovac 13 August 2015, 17:34:53 UTC service::node: Make the number of workers to start configurable Some services, like tilerator, need to be able to specify the number of service workers to run. This patch allows its configuration via the $no_workers param added to the service::node define. Bug: T108888 Change-Id: Ib0429ae8466ab712ae6927917cba98bcc88fad3b 14 August 2015, 07:20:59 UTC
a4ce1ca Ori Livneh 14 August 2015, 04:38:43 UTC base: ensure => absent on 'command-not-found' The sole purpose of this package is to punish you for your typos by adding extra latency. Change-Id: I789644c8da4ff374918e9ff98238bb13c237ddfc 14 August 2015, 04:40:26 UTC
a5edccc andrewbogott 01 August 2015, 08:54:53 UTC Update archive-project-volumes to support our new NFS setup. Also move to python3 Bug: T104857 Change-Id: I2ed0c23b778c6d6cd9754a64e6c45e8cf7411ba3 13 August 2015, 21:12:11 UTC
3633128 Ori Livneh 13 August 2015, 20:38:22 UTC xenon-generate-svg: fix handling of reversed flame graphs Change-Id: I02c78fc6e53e84723f91618c8e2aa754a5b82a88 13 August 2015, 20:38:41 UTC
ed104d0 dzahn 13 August 2015, 19:25:15 UTC fermium: adjust rsync import dir, ensure package adjust the rsync import path to use ./import/ to separate it from existing files in /var/lib/mailman ensure the mailman package is installed Bug:T108073 Change-Id: Icf163e13e9902d57a3898375e5df650f20a803d2 13 August 2015, 20:22:17 UTC
4346972 Ori Livneh 13 August 2015, 19:59:18 UTC xenon: fix for xenon-generate-svgs Change-Id: I1996d8d4bc8b88204594a1c3dc99b1a644787262 13 August 2015, 19:59:54 UTC
0f4c267 dzahn 13 August 2015, 19:09:36 UTC fermium: re-include rsync server removed by accident in a manual rebase, this include still needs to be here Change-Id: I7cef046e75f2abc63b281d0278c4b8f517091142 13 August 2015, 19:09:36 UTC
9cb2ee8 dzahn 13 August 2015, 18:33:16 UTC fermium: ferm rule to allow rsync from sodium Allow rsyncing from sodium (lists) to fermium (staging for list migration). Bug:T108073 Change-Id: Iac06805a04d8267efa0fb10d20845d807841aa8d 13 August 2015, 18:40:31 UTC
5f40846 andrewbogott 11 August 2015, 16:59:50 UTC Remove network_host setting. This hasn't been implemented in Nova in years, and even when it was it wasn't used for our setup. https://review.openstack.org/#/c/19909/ Change-Id: I7b284cd8d7e80f4c75b6912c4a53c357696d45a8 13 August 2015, 18:27:13 UTC
676da78 RobH 13 August 2015, 18:20:03 UTC grant services team full icinga permissions discussed on task, giving services group (,eevans,mobrovac,gwicke,ppchelko) full icinga permissions T105228 Change-Id: I67d2426ff837c74fe163a8b62f80414d07b2cbbb 13 August 2015, 18:22:13 UTC
6540943 Bryan Davis 11 August 2015, 21:42:15 UTC logstash: normalize "level" fields across log types We inject logs from several different sources, each of which have slightly different conventions for naming things like the event severity level. Apply a Logstash filter to attempt to normalize this naming for some of the more common input types. This normalization is intended to make comparison of events from multiple origins easier in dashboards. Change-Id: Ief0b9631bc0eb59f2bea30c6aa2e639bf27debde 13 August 2015, 18:02:42 UTC
780454d dzahn 12 August 2015, 23:04:52 UTC fermium: add rsync server to sync from sodium Install an rsync server on host fermium and let sodium connect to it to be able to push over mailman list archives for migration testing. (without having agent forwarding) Change-Id: I0b5fb7fffbed0359c1d15755feebba8f86e82087 13 August 2015, 17:55:02 UTC
9914ba4 Ori Livneh 13 August 2015, 17:46:35 UTC xenon: fix typo introduced in I1c228ea75 Change-Id: I4b1e4542e679ce6d8b76ada2b7b86494972d94d5 13 August 2015, 17:46:35 UTC
d43a930 Ori Livneh 13 August 2015, 17:40:25 UTC xenon: small tweaks to code clarity and variable names Change-Id: I1c228ea754202a7e5e1c1c4b59731b5e41446126 13 August 2015, 17:40:55 UTC
2b51312 Marko Obrovac 13 August 2015, 17:06:49 UTC MobileApps: Add the port to the restbase_uri config Change-Id: Iaafc5c1591c78705dc3083c77bc01b00b4ca8cef 13 August 2015, 17:35:02 UTC
a48d2eb Marcel Ruiz Forns 11 August 2015, 18:13:49 UTC Change percentage in EventLogging validation alert To avoid false positives for eventlogging_difference_raw_validated, this change raises the number of data points over the threshold needed to trigger the alert from 3/15 to 4/15. It also adds the modifier 'absolute' to the alarm's metric, to also raise alerts in the unlikely case the number of valid events is significantly higher than the number of raw events. Bug: T108339 Change-Id: I64b2091be24a3b2531586cf5ea2ba775ef668d1a 13 August 2015, 17:27:11 UTC
1afb522 Moritz Mühlenhoff 06 August 2015, 13:43:35 UTC Add ferm rules for spark/jmxtrans Change-Id: I34cb06e7d24ccbeffdb6ef7c1610763d38658c4a 13 August 2015, 16:53:07 UTC
c6166a5 Alexandros Kosiaris 13 August 2015, 16:34:52 UTC maps: Fix typo in maps.cql.erb Missing an 'n' character in cassandra Change-Id: I46a2729a2f6481d9e91f4c26b7f7adf92f5fb1cd 13 August 2015, 16:34:52 UTC
33f26ff dzahn 12 August 2015, 00:46:23 UTC elasticsearch: add cluster hosts to hiera Add the (currently eqiad) server hostnames of all ElasticSearch servers to hiera so that I7fe52ad7745cd35250 can use them to create firewall rules. Bug:T104962 Change-Id: Iefd296d03b7949e3892ea59a7d1dae7b06b82045 13 August 2015, 15:56:02 UTC
c6ab4d3 Alexandros Kosiaris 13 August 2015, 14:27:12 UTC mobileapps: checkout_submodules = true Otherwise src does not get checked out Change-Id: I28749f908c5946f6dd899b0f2961794d73796499 13 August 2015, 14:29:59 UTC
941ce35 Alexandros Kosiaris 13 August 2015, 14:20:54 UTC maps: Add redis to the cluster Bug: T107813 Change-Id: Ic519c536942cd5ed36875a24d62692b7c0b1b5f9 13 August 2015, 14:28:06 UTC
ec6665a Madhumitha Viswanathan 12 August 2015, 21:29:27 UTC [WIP] eventlogging: Add statsd_host param to the mysql consumer url See https://gerrit.wikimedia.org/r/#/c/229631/ for details. Bug: T105935 Change-Id: Ic1bb3599e6e567f4a6bd181ea2cbdd6ca4d3cdf0 13 August 2015, 14:17:24 UTC
7fc87cb Brandon Black 13 August 2015, 13:25:16 UTC Remove mwuser cookie exception Some quick samples of logs confirm expectation that virtually nobody has failed to restart their browser since this was fixed over two months ago, so we can drop the extra complexity now. Change-Id: Id337e96e70bbe6379049cc9469e238830258cccb 13 August 2015, 13:45:36 UTC
1d3b5f0 Alexandros Kosiaris 13 August 2015, 12:20:14 UTC new_wmf_service: Use shallow copies for group derivation new_wmf_service erroneously used assignment instead of a copy operation with the result being data missing from the end result. Use the copy module and a shallow copy operations (note deep copy is not needed) to get the data into the temporary structure. All of this is due to the need we have to delete some groups temporarily. While at it fix the regexp for detecting the one line description of the service Change-Id: Ia527adc3716ac3fd60a7ced563ce69ce9e326b18 13 August 2015, 13:29:23 UTC
d04aa05 Alexandros Kosiaris 13 August 2015, 13:23:55 UTC Include role mobileapps in role scb Change-Id: I1e922fa9d2bb56ab08879ce5021457ff5451b97c 13 August 2015, 13:23:55 UTC
2944871 Giuseppe Lavagetto 05 August 2015, 14:56:29 UTC wikidata query: add misc-web configuration Domain query.wikidata.org gets routed to the wdqs backends directly from misc-web, in the same way logstash does - we use a random director because requests are stateless. Bug: T107602 Change-Id: I87ba959927d6d986a26cd465e9334b12c866afd5 13 August 2015, 13:19:00 UTC
55a1fd5 Alexandros Kosiaris 13 August 2015, 12:46:40 UTC Assign mobileapps service to scb cluster Bug: T105538 Change-Id: Ie56eabdffaa3434f8add07eee996b84f84b1f769 13 August 2015, 13:18:29 UTC
6a83090 Alexandros Kosiaris 13 August 2015, 13:14:38 UTC Make nikerabbits' realname have only ASCII characters Followup for dc7db81 Change-Id: I2a508c6dea30aa01f31448428321168460d27bfe 13 August 2015, 13:14:38 UTC
dc7db81 Alexandros Kosiaris 13 August 2015, 13:09:10 UTC Remove UTF8 character from nikerabbit realname Getting some puppet errors on SCB about: Error: Could not convert change 'comment' to string: incompatible character encodings: UTF-8 and ASCII-8BIT Remove the UTF8 character to fix that Change-Id: I2ff4ef3c7f8575305ed0fc2c5f0e4ac75f8094bf 13 August 2015, 13:09:10 UTC
48d5d32 Alexandros Kosiaris 11 August 2015, 15:26:32 UTC Introducing mobileapps role and puppet module Bug: T105538 Change-Id: I4cbc5a3bc1b96332d06db4b8a7b7e1f1f9ed95f6 13 August 2015, 12:07:01 UTC
d19183d Giuseppe Lavagetto 13 August 2015, 11:29:36 UTC ganglia::monitor::aggregator: fix upstart (take 2) We fix the ganglia-aggregator-monitor script to run as a real long-running job instead of a task, so that we can truly manage all the instances via it. Also, turns out that when running as aggregators, expect fork doesn't work with daemonize = yes, so we do run gmond in the foreground in this case (which is preferred from upstart anyways, and we remove the "expect fork" from the script, which was a hack anyways. Note that with the default config (the one we use for ganglia-monitor), expect fork works correctly, so something is still there to be understood. Change-Id: I829eea6deff04e84548313c21f0f28827014c641 13 August 2015, 11:39:19 UTC
af1478d Alexandros Kosiaris 13 August 2015, 10:35:06 UTC maps: GRANT on all KEYSPACES instead of specific ones It is quite possible we will be using multiple keyspaces instead of a single one. So GRANT on all keyspaces instead Change-Id: Id11f85c94597a0b554c0198a221e379bb273069c 13 August 2015, 10:35:06 UTC
8f23dc5 Giuseppe Lavagetto 03 August 2015, 12:34:43 UTC ganglia-monitor-aggregator: fix upstart script At the moment, the ganglia-monitor-aggregator-instance upstart jobs report "unkown instance" for all running aggregators, but the corresponding gmond process is indeed running. The reason for that is that gmond does demonize - so we need "expect fork" here. Note that a bug on the package (https://bugs.launchpad.net/ubuntu/+source/ganglia/+bug/1387853) suggests to run gmond in the foreground instead and may have a point in general. This small fix will however be enough to fix our current situation. Also, we modify the parent script to sleep infinity once it's launched, and a pre-stop script, so that it is indeed possible to manage the gmond instances with a single command. Change-Id: I861ff238c59b46e1d95cdd8be91bf950436f8112 13 August 2015, 10:33:53 UTC
627e091 Yuri Astrakhan 10 August 2015, 14:27:31 UTC maps: Add usernames/passwords to kartotherian config In order for the kartotherian service to connect to DB & cassandra provide it with the passwords in the generated service config. Bug: T108610 Change-Id: Ib18328e86f15ddf5505bcb0b443b02f63c5b95b8 13 August 2015, 09:57:18 UTC
c29fa1b Eric Evans 10 August 2015, 18:08:52 UTC update cassandra-metrics-collector version This changeset depends on: https://gerrit.wikimedia.org/r/#/c/230582/ Bug: T101764 Change-Id: I41c2b09cd7cb1fadfbdc45871d5458b04f4102fe 13 August 2015, 08:29:05 UTC
8d71ae6 Ori Livneh 13 August 2015, 04:29:46 UTC xenon: use full path when calling os.path.getctime Change-Id: I01808e664ea1c09da62a1a121662c3183043719e 13 August 2015, 04:29:46 UTC
e098d8d Ori Livneh 13 August 2015, 01:48:20 UTC xenon: stop generating weekly logs; tag files with entry-point Change-Id: I5a40257796fee03145f387462764d9752a9a4a88 13 August 2015, 01:49:00 UTC
29fbd7d Ori Livneh 13 August 2015, 00:09:35 UTC xenon-generate-svgs: make flamegraph.pl invocations nice Change-Id: Ib08b286106dcae25967b452f91588fea0727b43f 13 August 2015, 00:09:35 UTC
79bf8df Andrew Bogott 12 August 2015, 22:56:04 UTC Revert "Remove network_host setting." This reverts commit 7f8735bdc01240eedcda867a6e091f05d0153ef9. Change-Id: I4cd30703b51e8f1b7cd9c3e13cffbae43ddf76b4 12 August 2015, 22:56:04 UTC
7f8735b andrewbogott 11 August 2015, 16:59:50 UTC Remove network_host setting. This hasn't been implemented in Nova in years, and even when it was it wasn't used for our setup. https://review.openstack.org/#/c/19909/ Change-Id: I6071ac275f217ff7ef0b57ee74d6b73a32eb711e 11 August 2015, 17:00:51 UTC
3294ff1 Federico Leva 12 August 2015, 21:03:35 UTC [Planet Wikimedia] Add Content Translation to the English Planet Blog on MediaWiki development. Requested at https://meta.wikimedia.org/?diff=13028813 Change-Id: I7b32812f8cdc53383c21bfff681c3a801271247f 12 August 2015, 21:06:04 UTC
26d8060 Ori Livneh 12 August 2015, 21:01:00 UTC xenon-generate-svgs: make the palette of reversed flame graphs blue Change-Id: If9de0d39b3a0ef80f2fde335c24a3fe3b2709e1b 12 August 2015, 21:01:00 UTC
d3ce329 Ori Livneh 12 August 2015, 20:57:15 UTC Make snapshot::packages compatible with trusty Change-Id: Ifd8a713a9f76d586dfa164e88ca0f9b0188c51fe 12 August 2015, 20:57:57 UTC
97168a6 dzahn 12 August 2015, 05:34:34 UTC admins: delete mailman-users group since now mailman-admins is the group name with actual permissions, has the same members and this group can't do much anyways Change-Id: Idb26e66d8fffe93e7bf24bb12f0fa16e0563ae05 12 August 2015, 20:46:41 UTC
9bdff4f Andrew Otto 12 August 2015, 19:25:34 UTC Install python-pykafka for eventlogging server Change-Id: If0eed2bfb4575a39554e1b0c5eb3fe9361da05f6 12 August 2015, 20:34:45 UTC
cb9759d dzahn 12 August 2015, 20:19:14 UTC dumps: add ferm rule for port 2049 from INTERNAL Change-Id: I4c815da49d0e406f62f1f7c49f97eb8c1ce7ff6f 12 August 2015, 20:24:00 UTC
170bf0a Ori Livneh 12 August 2015, 20:01:47 UTC xenon: invert color scheme, not orientation Fix-up for I941660b516. Change-Id: Ia2b0530b7c628ef16d2d9ae80dd785d9ad952b7a 12 August 2015, 20:03:40 UTC
ba19af4 Ori Livneh 12 August 2015, 19:27:11 UTC Apache: set DeflateCompressionLevel to 9 The default is to use zlib's default, which is 6. In my tests, the savings were neglible, but better than nothing. Change-Id: I4d471434e4252304dce9ab850c85a585ead66a8b 12 August 2015, 20:02:31 UTC
dca9c4d Brandon Black 12 August 2015, 19:42:12 UTC 403 for maps - T108765 Change-Id: I369f5cc6ef99cf85b0750ee3d64fd2bada25c9db 12 August 2015, 19:42:39 UTC
5225603 Dzahn 12 August 2015, 19:27:45 UTC Revert "tungsten: set DHCP options to install jessie" This reverts commit 16e0b6f9fcfc201cfb47b5ad77f6442627711488. Change-Id: Ib5dea196ee9be63a5c688c9bf580a3772dce3c1b 12 August 2015, 19:29:05 UTC
454c92a Ori Livneh 12 August 2015, 19:13:11 UTC xenon: generate reversed flame graphs; prune old svgs * If an SVG file does not have a corresponding log file, prune it. * Generate a reversed flame graphs as well. Change-Id: I941660b516aecc4797253677d19acbcb14fe668d 12 August 2015, 19:14:57 UTC
7b879d8 RobH 12 August 2015, 18:57:02 UTC revoke Joel Sahleen's access contractor no longer with org, removing org user T108854 Change-Id: I822301e0b178ab395b2c752361fe4a5425a1809a 12 August 2015, 19:00:27 UTC
16e0b6f dzahn 12 August 2015, 18:47:14 UTC tungsten: set DHCP options to install jessie We want jessie nowadays. Bug:T106563 Change-Id: I0d907caecb18b333c6a5a50b447873e7a61725d8 12 August 2015, 18:49:04 UTC
87fbc9c dzahn 12 August 2015, 16:53:20 UTC tungsten: change partman recipe to use RAID 10 T106563 is for re-using this server to transcode Wikimania videos and requests RAID 10 to be used, so change the partman recipe from "db" to this existing one Bug:T106563 Change-Id: Ia16b32bc9134c78188e93fadfd4fe1169f540854 12 August 2015, 18:41:18 UTC
6d6b035 Andrew Otto 12 August 2015, 18:25:13 UTC Use monitoring::graphite_threshold for Kafka alerts Also fix $graphite_broker_key regsubst to use G flag Change-Id: I42d8db97f114443a6fc77713f4141bd60fc87ac3 12 August 2015, 18:25:38 UTC
5f783b1 andrewbogott 07 August 2015, 17:30:27 UTC Add Chris Steipp to analytics-privatedata-users Bug: T108227 Change-Id: Ib65380cf5aa2878c666e3784abce80e087e70b33 12 August 2015, 18:10:17 UTC
5ce1b64 Andrew Otto 12 August 2015, 17:59:11 UTC Update kafka module wiht ReplicaManager jmxtrans fix Change-Id: Id491c79f25aa0eff93cf0e083d884a545e12e6cd 12 August 2015, 17:59:32 UTC
8d157d4 Marius Hoch 11 August 2015, 21:22:27 UTC Add ssh key for new notebook Also uploaded to terbium (to confirm it's me for real): https://people.wikimedia.org/~hoo/tmp/wmfProdNew.pub Change-Id: I136900fb5b931c489cb69cbc07be7bef3e63d8c1 12 August 2015, 17:50:55 UTC
dd58278 Bryan Davis 07 August 2015, 19:49:47 UTC logstash: Count MediaWiki log events with statsd Count each MediaWiki logstash event by sending an increment command for the counter "logstash.rate.mediawiki.$CHANNEL.$LEVEL" (eg "logstash.rate.mediawiki.memcached.ERROR") to a statsd server. Bug: T100735 Change-Id: I4771abef68a151d08340b63f95cc556c8010416d 12 August 2015, 16:46:18 UTC
13bab62 Coren 12 August 2015, 15:45:54 UTC nrpe: simplify systemd unit status check Since the start time of units from timers is updated at start and not at end, we can rely on that alone to check for activity. This change removes the explicit test for ActiveState when the check is for 'periodic' Change-Id: I26bc086adec871ed0d5bb9bc998d25f6bc7338c1 12 August 2015, 16:20:04 UTC
0eb67ae Andrew Otto 12 August 2015, 16:13:07 UTC Use correct class graphite::monitoring_threshold for Kafka graphite based alert Change-Id: Ic2724303ae2f6aaa0bdd86285ae22ecace06f6b2 12 August 2015, 16:13:07 UTC
baf71f5 Andrew Otto 12 August 2015, 16:04:32 UTC Update alerts and jmx for Kafka 0.8.2+ Bug: T106581 Change-Id: Ia384b0a71d0e8e85d63c1eeccbf1caa79fb47cc4 12 August 2015, 16:05:23 UTC
214308c Chad Horohoe 12 August 2015, 16:02:15 UTC Elastic: Unify default plugins_dir in /srv/deployment We set it to /srv/deployment everywhere. No install here actually uses /usr/share Change-Id: If5ef36c2bf4b072c062f49c3b0837a4a76b7de4d 12 August 2015, 16:02:15 UTC
b25de6d Chad Horohoe 28 April 2015, 16:57:07 UTC Elastic: move auto_create_index into hiera instead of role Change-Id: Ibd58670e17c65817dd36986c88f5d6e8a2f86ff8 12 August 2015, 15:54:15 UTC
cae6161 Brandon Black 11 August 2015, 16:03:13 UTC varnish: director->backends is now always an array Change-Id: I18a411e9433081abacb32ea3c10dbf4513f5862d 12 August 2015, 14:54:42 UTC
6e9ec68 Brandon Black 12 August 2015, 13:16:18 UTC tlsproxy: add deferred/backlog options deferred enables TCP_DEFER_ACCEPT, potentially reducing nginx process wakeups per conn. backlog is set to 4096 to match the somaxconn sysctl that's set from webserver::sysctl_settings, included by tlsproxy::instance. Change-Id: If276c8ea5a860ba2f8dc175283cc4bfafadcc199 12 August 2015, 13:51:01 UTC
4ab62cf Yuri Astrakhan 06 August 2015, 14:54:04 UTC Added tilerator service, granted kartotherian OSM DB read access * Cloned kartotherian settings for everything since they are practically the same * Clarified kartotherian's service description * Add tilerator to the same servers as kartotherian * Allowed kartotherian to access OSM database - we might have to get high-zoom tiles directly from the SQL instead of pre-generating them. * Editor auto-removed one trailing space in one of the files. * Changed all private hiera settings to reference the maps:: namespace * Added tilerator LVS service IP Bug: T105074 Change-Id: Ifcf4f1ef77c2d25e0fc42f0582e950cc76624cf2 12 August 2015, 13:18:46 UTC
1cd724a Alexandros Kosiaris 11 August 2015, 16:17:43 UTC Get scb up to par with sca Create relevant yaml structures for conftool, ganglia, icinga. Minor cleanup of icinga structures as well while at it. Change-Id: Icb4ef8f188da206d8047730823d7c8b775fb7f8c 12 August 2015, 12:58:49 UTC
68ee721 Giuseppe Lavagetto 12 August 2015, 07:52:08 UTC apertium: raise the open files limit Bug: T107270 Change-Id: I89848062379d0e153f4330729d6531a71f31feb3 12 August 2015, 07:53:40 UTC
969ffad Alex Monk 11 August 2015, 20:04:57 UTC Kill ee-prototype.wikipedia.beta.wmflabs.org Bug: T107397 Change-Id: Ia1f05d41d12a9357ac14c547b5ddd5105826e32b 12 August 2015, 01:29:03 UTC
7946233 dzahn 12 August 2015, 01:11:14 UTC varnish-misc: remove backend zirconium decom/reclaim Bug:T105510 Change-Id: Ib27725b8fa6b999986bc0965d3a261ff503f6d9f 12 August 2015, 01:11:14 UTC
d08aae2 Faidon Liambotis 12 August 2015, 00:53:36 UTC Add msw1-codfw and mr1-codfw to monitoring tools Specifically: - icinga - RANCID - smokeping - Torrus Change-Id: I4aea417492bc118b72854d84933311a65457eef7 12 August 2015, 00:54:40 UTC
b3124b3 dzahn 10 August 2015, 22:55:16 UTC dbtree: make compatible with Apache 2.4 Like on I4c8e0aa365fbcc6 but only dbtree Needs to be there before dbtree can move to jessie/Apache 2.4. Change-Id: I51cd69bb106f0a4a9bdd457924f20fabf90db215 11 August 2015, 23:59:00 UTC
1384797 dzahn 11 August 2015, 00:31:40 UTC OTRS: make compatible with Apache 2.4 Like on I4c8e0aa365fbcc6 but only OTRS (ticket.wm.org) Needs to be there before OTRS can move to jessie/Apache 2.4. Apache syntax changed between 2.2 and 2.4. Change-Id: I7077bad59bb782813bc5c4358f7798e82a805961 11 August 2015, 23:52:12 UTC
e456cd2 dzahn 11 August 2015, 23:39:25 UTC admin: mailman-admins on fermium, not just users The mailman staging host fermium should have mailman-admins, not just the mailman-users group. also see: I2a721c2fbc788382, I1f6a7016d49de has been approved Bug:T108349 Change-Id: Ica54a54eb5c1f7fc549eaaff96e033c15ed5c674 11 August 2015, 23:39:25 UTC
f58ecf1 Alexandros Kosiaris 11 August 2015, 19:10:47 UTC maps: Fix typo introduced in 5daab4f s/master/slave/ Change-Id: I7266dc7f36fc3a60258502d7dda67b9929b1552f 11 August 2015, 19:10:47 UTC
5daab4f Alexandros Kosiaris 11 August 2015, 19:02:41 UTC maps: Include tuning.conf in slaves as well Change-Id: Ie76f82f7aad31588db5150b1edb0c2ffe6a9ea83 11 August 2015, 19:03:16 UTC
02cdf66 dzahn 11 August 2015, 18:51:27 UTC zirconium: decom, rm from site.pp,DHCP,netboot zirconium can be decom'ed/reclaimed, all services have been moved Bug:T105510 Change-Id: I6eeb808e5894928976520d9721a2129ca986ccb4 11 August 2015, 18:51:27 UTC
7273e07 dzahn 11 August 2015, 18:29:16 UTC grafana: remove role from zirconium Bug:T104946 Change-Id: I539193caaf0fca2d9bc6a9ffb367d795109b18e4 11 August 2015, 18:30:14 UTC
98da387 dzahn 10 August 2015, 21:36:37 UTC misc-web: switch grafana to backend krypton Bug:T105008 Change-Id: If01ce07e512e9a71415060799e36d2216d0e8c7f 11 August 2015, 18:23:52 UTC
aa9cf27 dzahn 11 August 2015, 18:10:49 UTC grafana: needs to load Apache mod_proxy_http too We need to load all these modules on Apache 2.4 for it to work. proxy, proxy_http, proxy_balancer and lbmethod_byrequests. Before we still got "No protocol handler was valid for the URL" and a 500 error when it was trying to access the dashboards. We had 3 of the 4 above loaded, just not this one. Somehow it was loaded on zirconium on Apache 2.2 though. Bug:T105008 Change-Id: I0e8f2e7aaf1ae7e178b933998337202999afa018 11 August 2015, 18:19:19 UTC
53ff199 Coren 11 August 2015, 16:46:01 UTC Labs: put the real check interval for backups Set to 25h (that is, should run daily but has a bit of give to avoid false positive because load, etc fluctuations) Change-Id: Ib2a4501f9d1225236d2792901eb29dbbdfa7ed9f 11 August 2015, 17:30:16 UTC
ed0b44d Brandon Black 11 August 2015, 13:46:35 UTC define puppet ganglia stuff for caches @ codfw Change-Id: I0e06d07d8003169df3d4114dd62d593fe6bdabce 11 August 2015, 17:22:31 UTC
2a92801 Brandon Black 11 August 2015, 16:15:53 UTC cache_mobile: def_ttl 30d Bug: T108612 Change-Id: Ic28ff9c2d84313b43f48aa5d9bd6219d4b0be64f 11 August 2015, 17:20:49 UTC
36b680e Brandon Black 11 August 2015, 16:18:18 UTC cache_(text|upload): frontend default_ttl => 30d Bug: T108612 Change-Id: I1b57d0d69fe3d132be9aa82515ebc15f92a31cc4 11 August 2015, 17:11:18 UTC
115a02e Brandon Black 11 August 2015, 16:16:33 UTC cache_(text|upload): remove runtime_params conditional Change-Id: Ifcd565e4cd5b281a9882b114b9a11fa665678106 11 August 2015, 17:11:00 UTC
119bfae dzahn 10 August 2015, 22:41:21 UTC etherpad: make compatible with Apache 2.4 Like on I4c8e0aa365fbcc6 but only Etherpad Needs to be there before Etherpad can move to jessie. Change-Id: I8c47fe31b9dcbd65a400f9bd8010ef7f6b8ab143 11 August 2015, 16:38:34 UTC
5c59d04 Bryan Davis 11 August 2015, 16:08:26 UTC logstash: fix ldap_bindpass I tried to move ldap_bindpass into hiera in Ia042191 but the scoped lookup I used there doesn't work. It just passes an empty string to the puppet class at runtime. As a quick fix, change back to the prior logic of pulling the private password setting into scope inside the class itself. Change-Id: I2d45c21e227052cf705468fff1f02364a4686cb4 11 August 2015, 16:08:26 UTC
0097a71 Alexandros Kosiaris 11 August 2015, 15:43:07 UTC new_wmf_service.py: fix icinga monitoring Remove the hardcoded check and replace it with something that makes sense. Targeting nodejs services for now Change-Id: Id7a30769f1c615de500c97f44f4a180a71956ce7 11 August 2015, 15:46:37 UTC
42a05d0 Bryan Davis 07 August 2015, 22:31:19 UTC logstash: Enable doc_values in template mapping Add '"doc_values": true' to numeric, boolean, geo-point and not_analyzed strings in the template mapping for new Elasticsearch logstash-* indices. Doc values reduce runtime heap consumption for fielddata usage by creating pre-computed on-disk structures. We were doing this dynamic raw strings previously but the configuration syntax has changed and upstream Elasticsearch is not recommending this for all applicable field types. The next major version (2.0) of Elasticsearch will use doc_values by default. Changing this in Puppet is just a bookkeeping and new Elasticsearch cluster feature. Existing clusters will need to have their mappings updated manually via HTTP PUT to /_template/logstash to replace the current index template. This template has already been applied to the production and beta logstash clusters. Ref: <https://www.elastic.co/guide/en/elasticsearch/guide/current/doc-values.html> Bug: T74930 Change-Id: Ie8cf588f7a0a9ebdb1ba0ca337e1cf934065b3e9 11 August 2015, 15:39:27 UTC
0df444e Bryan Davis 27 July 2015, 00:26:45 UTC labs: new role::logstash::stashbot class Add a new role and configuration for use in the stashbot labs project to create instances for logging irc messages. The irc filters have special handling for messages with certain prefixes: * !log for server admin log (SAL) logging * !bash for quips logging * Remove role::logstash::ircbot which was previously only used in the beta cluster. * Refactor role::logstash to include role::logstash::elasticsearch rather than duplicating its configuration. * Update role::logstash::elasticsearch to only apply ::elasticsearch::ganglia when $::standard::has_ganglia is true. This mirrors usage in role::elasticsearch. * Refactor role::kibana to allow fine grained configuration via hiera. Change-Id: Ia04219138f3dab667d7c2a667994681aeeabc563 11 August 2015, 15:39:06 UTC
f290715 Bryan Davis 25 July 2015, 16:01:44 UTC Update configuration for logstash 1.5.3 * Remove obsolete logstash-contrib package * Setup trebuchet package for distributing logstash plugins * Use require_package() to clean up dependencies * Switch from deprecated elasticsearch_http to elasticsearch output * Change logstash::output::elasticsearch from class to define * Reuse logstash::output::elasticsearch for apifeatureusage output Bug: T99735 Change-Id: I24c049139f7ff4b70a4504eea4e4d49d40e79827 11 August 2015, 15:38:37 UTC
36e14a2 Alexandros Kosiaris 11 August 2015, 15:20:03 UTC new_wmf_service: fix bug with wrong function name it's setup_conftool_data(). Also the function needs to return True to signal successful termination Change-Id: I7b269d513e065e4fad6f4df9943b06f0112bab10 11 August 2015, 15:35:57 UTC
f137803 Alexandros Kosiaris 11 August 2015, 15:01:46 UTC postgres: stream WALs while doing pg_basebackup This really doesn't matter in the new server scenario, but when trying to get an already long running server resynced to its slaves, WAL needs to copied over as well. Also opt for the streaming mechanism as it is less susceptible to race conditions Change-Id: Ia1d8104e080deac1fdeefb3f59385ed8783a6285 11 August 2015, 15:30:42 UTC
ddc277d Coren 10 August 2015, 16:25:37 UTC labstore: add timers for backups Bug: T106474 Change-Id: Ie10622ba825f620240be654798dda763a68129b7 11 August 2015, 15:16:52 UTC
8ef7b6d Alexandros Kosiaris 11 August 2015, 14:53:03 UTC lvs: remove the old unused osm lvs_service Cruft leftover from 15bdf16 where osm IPs were removed Change-Id: Ie2032e3b0832323cc3cdb21355d86f7b73df2c21 11 August 2015, 14:57:05 UTC
back to top