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
eaabd89 cpettet 23 February 2016, 22:40:46 UTC labstore: real time application of tc setup This is intended to be merged post careful manual post staggered rollout of 272891 Change-Id: Id3892fc74cb42cbb92622eebf783979ce119d70d 23 February 2016, 22:43:40 UTC
5498ae7 cpettet 23 February 2016, 21:57:56 UTC labstore: stage tc script and enable on boot We have been piloting this configuration for a few weeks to limit the variable load exposure of labstore10* in tools. Labstore1001 especially has difficulty sustaining the extreme bursty load that any operation can trigger. These shaping values are based on data collected from the nfsiostat diamond collector deployed a few weeks ago. This will all most likely require tweaking over time. Change-Id: I19814c3da53f78666a8877937353f3987a904501 23 February 2016, 22:40:34 UTC
400d77b dzahn 19 February 2016, 05:21:14 UTC role/mail: split file, move to module/role/ - split role class file into one file per class - move into module/role/ into autoloader layout Change-Id: I21ea06953859a38dc105e0d4e3f012c04dfb949c 23 February 2016, 22:15:21 UTC
6f9875d dzahn 23 February 2016, 04:30:11 UTC backup: move roles to modules/role/ Change-Id: I39e8d35f8a19b6aa23ec681cc5efcb63e91292eb 23 February 2016, 22:05:20 UTC
22f1c53 dzahn 23 February 2016, 06:29:54 UTC ferm: fix "not documented" warnings WARNING: defined type not documented WARNING: class not documented Change-Id: I9f4dc205e42dd4e10ecfb89bda05e1b4259faa3f 23 February 2016, 22:03:22 UTC
c05cadc dzahn 23 February 2016, 04:23:07 UTC phabricator: move roles to module/role/ Change-Id: I2dd111d9c806a7a7c8abd434814fc22c4ea0e055 23 February 2016, 22:00:05 UTC
ed6e692 dzahn 23 February 2016, 04:04:40 UTC fix whitespace-related lint issues Fix a lot of lint warnigns due to incorrectly aligned arrows. Change-Id: Ib062d7142a79a8cb6bb3abd86fd7499aa3edad03 23 February 2016, 21:48:44 UTC
74e5625 Andrew Otto 18 February 2016, 22:19:07 UTC Remove $::realm conditionals around varnishkafka for labs webrequests Bug: T127369 Change-Id: I1bf5466549d97201d2aa30eb9b45db9e7f6fe793 23 February 2016, 21:27:23 UTC
8e58336 Andrew Otto 23 February 2016, 20:55:16 UTC Fix typo in my.cnf Change-Id: I1067ed9acdbd40fad403f27ff644ac46c3260eae 23 February 2016, 20:55:16 UTC
8e563da Andrew Otto 23 February 2016, 20:29:10 UTC Analytics MySQL Meta instance: 4G innodb_buffer_pool_size 4 instances Also add NOTE about WF_JOBS.created_time index Bug: T119646 Change-Id: I02be553922a40cbe427ad7c59ebdbb8022f13fac 23 February 2016, 20:31:44 UTC
eb93455 Andrew Otto 23 February 2016, 20:03:58 UTC Increase analytics mysql meta innodb_buffer_pool_size to 1G, query_cache_size to 16M Bug: T119646 Change-Id: I9013913cc9d99e704224981e592199af5341c726 23 February 2016, 20:05:12 UTC
3b5f1db Brandon Black 23 February 2016, 18:47:08 UTC 2layer: remove dead nodes storage_size Bug: T125486 Change-Id: Ief46d18401ac05f5488f0ac527d415af732d4865 23 February 2016, 18:47:45 UTC
03d61f5 andrewbogott 20 February 2016, 11:48:04 UTC designate: Set pool_target master to a public ip rather than 127.0.0.1 This shouldn't matter much, but in the long run it'll be important to have this ip match the one in the pdns domain record (as comment explains). Change-Id: I475146f37384cb7d420f535a1c6091e410ea3c4a 23 February 2016, 18:43:16 UTC
71a1fa5 cmjohnson 23 February 2016, 18:36:23 UTC Adding prseed and dhcp entries for test servers wmf4727-test. This is test server for raid controller card and will removed once testing is complete. Change-Id: I127eb112874112af9b3b67299774ca8c88ef71ca 23 February 2016, 18:38:59 UTC
c310a00 Brandon Black 23 February 2016, 14:05:16 UTC cache_misc: remove cp1057, cp1070 Bug: T125486 Change-Id: I2a2e2f8e90f71a35359799397e9b50a9e792ca3b 23 February 2016, 18:35:41 UTC
dd0a63f Andrew Otto 23 February 2016, 18:32:04 UTC Add analytics1015.yaml for analytics-admins Also remove some .yaml files for non existant hosts Bug: T119646 Change-Id: Id4a2df1b9b72b4e1cb3a3e3ce6253f6052521c43 23 February 2016, 18:33:17 UTC
9a3d3d6 Joseph Allemandou 23 February 2016, 18:13:14 UTC Update hiera heapsize for hive and oozie servers Bump HiveServer2 heapsize from 1024 to 6144 following Cloudera recommandation for ~20 users. Bump hive metastore heapsize from 256 to 10240 following Cloudera recommandation for ~20 users. Bump oozie server heapsize from 1024 to 2048 to improve response time. Bug: T110090 Change-Id: I4562ab426b2b82290914f0a3baf8fd05e3dd72a1 23 February 2016, 18:26:03 UTC
15e8c9e RobH 23 December 2015, 19:35:48 UTC new dumps.wikimedia.org certificate (renewal replacement) new dumps.wikimedia.org.crt file, the old cert expires on 2016-02-26. This certificate cannot simply merge, but must roll into place at the same time as the private key file is updated. Please see the task for details and do not merge without full understanding of other updates needed. T122321 Change-Id: I37a4e68f0ccacb100ed2ce658a3b5dacfdcd5ec7 23 February 2016, 18:17:11 UTC
8ba0dcf Andrew Otto 23 February 2016, 17:59:36 UTC Add accidentally uncommitted hiera cdh users.yaml Bug: T119646 Change-Id: I2bfbbfb04c0727ff530a131f0998b387b5dd4adb 23 February 2016, 17:59:36 UTC
9ceb29a andrewbogott 20 February 2016, 10:48:10 UTC Remove a couple of obsolete designate settings that no longer have any effect. Change-Id: I059bd73cf4035bdbee56df2e16ae13509c5ab610 23 February 2016, 17:50:30 UTC
6f50f4e Brandon Black 23 February 2016, 14:03:14 UTC cp10[56]1: upload->misc Bug: T125486 Change-Id: I93d871288a519b3939746b2af217c5379705c617 23 February 2016, 17:37:45 UTC
65e7a48 andrewbogott 20 February 2016, 10:46:09 UTC Change master=no for labs pdns We want to accept axfr from mdns but not send out updates from pdns. Bug: T124680 Change-Id: Ief70990861cbf7dc9bbc923591bf42e497d74af0 23 February 2016, 17:27:38 UTC
18b9aff Filippo Giunchedi 23 February 2016, 15:38:08 UTC cassandra: add restbase1008-b instance Change-Id: I5fd40d942cb67b7c99ca0ce914587038f97f763d 23 February 2016, 17:27:02 UTC
7e86158 Andrew Bogott 23 February 2016, 17:04:43 UTC Revert "Updates to designate/mdns/pdns setup for Labs internal dns" this doesn't work. I'll break it into pieces, I guess. This reverts commit a9600be4c879fed6d2cce576e2d9e7e5d32e9067. Change-Id: Icaa1a7f6b86d125f5702d8ffeb2bc59b7ac9305c 23 February 2016, 17:04:54 UTC
163f5ac Andrew Otto 23 February 2016, 16:53:59 UTC Include analytics_cluster* database classes on analytics1015 These ensure hive and oozie mysql databases, users, and libs are in place so that hive-metastore and oozie-server can talk to mysql here. Change-Id: If6650d3f1426536c714f61a57a768b64c6dcc4a9 23 February 2016, 16:54:52 UTC
9cb46e1 Jaime Crespo 19 February 2016, 19:04:17 UTC Small formatting fixes for replication lag check * Restrict output to 2 decimals * Check if lag is "" in addition to NULL (e.g. due to permission denied) Bug: T114752 Change-Id: I3b088833807be8b2ca095ee67d6934e358672a49 23 February 2016, 16:51:16 UTC
a9600be andrewbogott 19 February 2016, 04:41:46 UTC Updates to designate/mdns/pdns setup for Labs internal dns These changes are in response to a long conversation with a designate developer, Kiall MacInnes: - Remove a couple of obsolete settings that no longer have any effect. - Turn off 'master' for pdns servers. They should listen to mdns for updates but not broadcast updates themselves. - Add a second pool_nameserver representing the secondary dns server (holmium) That last change /may/ resolve confusion in interaction between pdns and mdns. It also creates a clear path forward for splitting out the two pdns servers onto separate dbs, which I have more faith in as a proper fix for the associated bug. Bug: T124680 Change-Id: I454118b9a0c506fce934df4ef9616358bdfe864e 23 February 2016, 16:38:19 UTC
2888664 Andrew Otto 23 February 2016, 16:35:21 UTC Use proper flume package naems in reprepro updates Bug: T119646 Change-Id: I5a337c198e565f5d7ed59cf660e465166f159849 23 February 2016, 16:35:21 UTC
fcfca78 Andrew Otto 23 February 2016, 16:31:03 UTC spark-core now depends on flume, mirror it from CDH to our apt Bug: T119646 Change-Id: I739c9501068f9658333907bbb8290de803522122 23 February 2016, 16:31:03 UTC
102a285 Andrew Otto 22 February 2016, 21:26:03 UTC Move Hive and Oozie to analytics1015 Bug: T110090 Change-Id: Ifcabdcfe9d2d4213d28baa9ed16f35bd0f150c3c 23 February 2016, 16:21:00 UTC
a609028 Andrew Otto 22 February 2016, 21:35:15 UTC Make MySQL instance on analytics1015 the master Bug: T110090 Change-Id: Ief6dce62699eed08014bc379bbd77057555161b0 23 February 2016, 16:16:08 UTC
d5e80ba Moritz Muehlenhoff 23 February 2016, 15:44:37 UTC Move role declaration to the top of the site.pp entry to fix Hiera lookup See Puppet_Hiera wikitech page for details. Also remove outdated comment Change-Id: Ib288ad9bd783c55a2ee1026e155efaa042d91300 23 February 2016, 15:44:37 UTC
0292704 Moritz Muehlenhoff 23 February 2016, 13:16:21 UTC Also use the canary grain in the debdeploy group for redis/eqiad Change-Id: Ibd8d961edaeffbd768ec14dcd040211e49d0687c 23 February 2016, 15:40:03 UTC
9971726 Filippo Giunchedi 23 February 2016, 15:34:33 UTC cassandra: add restbase100[89]-b to seeds Change-Id: Idf015cff5dc7ce55f3300b8532a313e35cd2331c 23 February 2016, 15:35:55 UTC
e482acf Moritz Muehlenhoff 23 February 2016, 13:14:45 UTC Also use logstash grain for logstash100[4-6] Change-Id: I38341c39933ae81fd4cf201557a13db14dec725e 23 February 2016, 15:31:34 UTC
658abbe Moritz Muehlenhoff 23 February 2016, 13:10:09 UTC Also split video scaler in eqiad/codfw-specific grains as already done for the other mediawiki grains used in debdeploy Change-Id: I72e5ef0bb61d37ee2cf914a8435aa8075a8d963c 23 February 2016, 15:08:58 UTC
88b82a0 Eric Evans 22 February 2016, 22:01:38 UTC disable package-installed initscript This replaces the initscript (a Debian conffile) with a no-op version. Bug: T127365 Change-Id: Id075ca0a0a8166d5c0e18903b66ab977d79b06cf 23 February 2016, 15:00:32 UTC
b2663de Brandon Black 20 February 2016, 17:36:36 UTC cache_misc: call recv_purge like others Nothing happens to be sending HTCP to cache_misc, and it doesn't yet run vhtcpd either, but it doesn't hurt anything to fix this deviation from the norm, and may come in handy down the road anyways... Bug: T127481 Change-Id: I5269a2400d64e75346e1ca4995f981b6ba5306cf 23 February 2016, 15:00:08 UTC
2680c5b Brandon Black 20 February 2016, 17:27:10 UTC Remove sub vcl_foo from sub-includes Functional no-op - remove 'sub vcl_foo' entrypoint defs from -common templates and zero.inc.vcl.erb, so that they only exist in the topmost includefiles (-frontend and -backend) Bug: T127481 Change-Id: I5a5d52f47ebc7628575f7b6b6b1771952fbf3820 23 February 2016, 14:56:49 UTC
d316319 Moritz Muehlenhoff 23 February 2016, 12:59:01 UTC Add salt grains for main kafka brokers and use in debdeploy Previously only the analytics kafka brokers were wired up in debdeploy. Change-Id: I9b7a990d4b2fd27a338c209f203323f19bae5bd2 23 February 2016, 14:40:55 UTC
27dd8c9 Brandon Black 23 February 2016, 13:42:05 UTC cache_misc: remove cp1056, cp1069 Bug: T125486 Change-Id: Ife86260a4bacd3e7dc1e30a3ecb50a4b1410ef72 23 February 2016, 14:38:10 UTC
8eef94b Brandon Black 20 February 2016, 15:12:48 UTC upload/misc VCL: remove t2-be bypass trick Only these two clusters had this new hack in place, which allowed skipping from t2-frontend to t1-backend (bypassing the normal route through t2-backend) when the traffic was known to be "pass" before requesting from the backend (explicit return (pass) or hit-for-pass object). The longer-term goal for this (after we have TLS to the applayer) is to go ahead and skip directly to the application layer in these situations. The current intermediate state of just skipping one cache layer was only a small optimization, and unfortunately this optimization makes related pressing work in T127481 more-complicated. Bug: T127481 Change-Id: Iac89fe9964c034acfaf63d19a860e6809100dab5 23 February 2016, 14:28:53 UTC
d40520a Moritz Muehlenhoff 23 February 2016, 12:53:23 UTC Add salt grain for alsafi and use it in debdeploy Change-Id: I6529c31cde0503eab34ede0b07f963dcfabc7c97 23 February 2016, 14:23:12 UTC
cf93e73 Moritz Muehlenhoff 23 February 2016, 12:46:29 UTC Assign salt grain for zuul::merger role and use in debdeploy Change-Id: Ia3825a5208a068c8c179ff502670c374a4ed69ce 23 February 2016, 14:05:43 UTC
bf91fbf Moritz Muehlenhoff 23 February 2016, 12:42:19 UTC Assign salt grains for debdeploy for salt masters Previously we had only one (palladium), so the grain was assigned to the puppetmaster role, now we have two separate systems, so use a separate grain and wire it up in debdeploy. Change-Id: I8b5c8c24f39899085e8c96e1a7189d9cebb28eb6 23 February 2016, 13:57:46 UTC
fe06683 Guillaume Lederrey 23 February 2016, 13:19:06 UTC Upgrade elastic search to 1.7.5 Updated the reprepro "updates" file entry for elasticsearch to the current 1.7 branch. Bug: T122697 Change-Id: I7fb1d0bd2650cf85cc59e3370c9bf6ae292ecaeb 23 February 2016, 13:19:06 UTC
a8d553a Brandon Black 23 February 2016, 12:58:46 UTC torrus: remove cache_parsoid refs Change-Id: I6f1231a23c25939476078f3b3d8088cd3d79eba6 23 February 2016, 12:59:15 UTC
eb5f9e6 Jcrespo 23 February 2016, 12:06:40 UTC Fix typo s/admin/admins/ on MariaDB's icinga config Bug: T127821 Change-Id: Ie5aea5d91304d92b77061f4de1107bb0c694f050 23 February 2016, 12:19:38 UTC
6d215ed Brandon Black 21 February 2016, 14:43:27 UTC decom cache_parsoid This removes all of the cache_parsoid configuration from our repo, and moves the 2x eqiad cache_parsoid machines to the misc cluster as part of the eqiad cache cluster rearrangements. Bug: T110472 Bug: T125486 Change-Id: Ie30f04f256831617f13bcd4b0ecba7af8727945e 23 February 2016, 12:05:38 UTC
6fe1fb9 Filippo Giunchedi 23 February 2016, 10:09:32 UTC uwsgi: don't declare uwsgi-startup service base::service_unit by default turns ensure => 'present' into ensure => 'running' for the underlying service. The magic incantation to use is declare_service => false. Bug: T127684 Change-Id: Ia50b7d08a4e8e533166bc7b1a067beface07bef6 23 February 2016, 10:12:34 UTC
0724de4 Eric Evans 22 February 2016, 17:12:27 UTC alert only on external (parity w/ dashboards) Threshold descriptions are also updated to more closely match the verbiage used on the corresponding dashboards, and the links are made direct to the fullscreen graph of the applicable metric. Change-Id: I7f54331d59ab71bfff7b937df874b5789e418bd2 23 February 2016, 09:44:26 UTC
afc9691 Jcrespo 22 February 2016, 15:38:23 UTC Non critical DBA pages should not send an sms to the DBA group When $is_critical is false, it should only show it on IRC (admin contact group). For example, for non-active datacenter replication. Should it send, however, an email? That may need creating a new, separate dba contact group. Change-Id: I7935fc0fd31c92766732d99b398c99ba165a28d0 23 February 2016, 09:35:47 UTC
0659526 Giuseppe Lavagetto 17 February 2016, 10:48:43 UTC ipresolve: add PTR resolution, tests Since it might happen that applying a reverse resolution can be useful, now 'ptr' is accepted as a type. Tests are added as well. Change-Id: Ic13886f6459146e701c864e987a75a7dd8e3878d 23 February 2016, 09:16:22 UTC
c97a6b6 Giuseppe Lavagetto 23 February 2016, 08:41:37 UTC wmflib: fix failing test Change-Id: Icc9fe24f5ae90e59ec26714473b9f6a2ecefe7df 23 February 2016, 08:41:37 UTC
879efb0 Moritz Muehlenhoff 22 February 2016, 17:10:41 UTC Remove outdated comment Change-Id: Ia9e2071df1027c9797b9ab49098e54da000e283d 23 February 2016, 07:59:18 UTC
6687658 Ori Livneh 23 February 2016, 05:42:40 UTC Add a comment about manually-created MongoDB indexes In lieu of proper Puppetization, add a comment which explains how to create the indexes XHGui needs on the MongoDB instance it is using. Change-Id: I88fe13fbd7066b00b0eacacf845facdf3362e2d9 23 February 2016, 05:46:21 UTC
8db4d80 Ori Livneh 22 February 2016, 02:19:25 UTC gdash: decom https://gdash.wikimedia.org/ is a static mirror of the dashboarding webapp previously hosted at that address. In October 2015, we deprecated gdash in favor of grafana. gdash has carried a deprecation notice on the top of each page since then. We can now remove it. /var/www/gdash.wikimedia.org will need to be wiped manually on graphite2001 & krypton Bug: T104365 Change-Id: I48150a5c906299a6d0c51d661dbde62d201ef25a 23 February 2016, 04:48:47 UTC
2c156d9 Brandon Black 21 February 2016, 14:32:35 UTC parsoidcache: remove from LVS Bug: T110472 Change-Id: Id82a0e286801339703f0743436d2c3617adc4a1f 23 February 2016, 03:46:15 UTC
ae43f53 Tim Landscheidt 11 February 2016, 19:08:21 UTC quarry: Move role classes to module role Change-Id: Idd7f078180d8d743de5f8602647b199421476ec0 22 February 2016, 23:08:56 UTC
290280f andrewbogott 20 February 2016, 06:21:12 UTC designate: Open firewall to axfr traffic from pdns hosts. Previously the secondary pdns server was unable to get xfr updates from designate-mdns. It stayed in sync by accident, due to sharing a commmon db with the primary pdns server. This may also have caused lockups, if the secondary was holding a lock on the database while waiting for an xfr update. Bug: T124680 Change-Id: Ie26af2f6e51e91ec3695718ee5d8e7c3bb4af861 22 February 2016, 22:45:09 UTC
99fc8c4 Antoine Musso 22 February 2016, 22:23:34 UTC Glance policy: grant manage_image_cache permission The godaddy openstack-puppet has two glance policy files, one that which allows snapshots: { "context_is_admin": "role:admin", "default": "", "manage_image_cache": "role:admin" } Another that does not: { "context_is_admin": "role:admin", "default": "", "manage_image_cache": "role:admin", "add_image": "role:admin", "modify_image": "role:admin", "delete_image": "role:admin" } References: https://github.com/godaddy/openstack-puppet/blob/master/roleprofile_modules/profile/files/glance-policy.json.snapshots https://github.com/godaddy/openstack-puppet/blob/master/roleprofile_modules/profile/files/glance-policy.json.nosnapshots So I suspect in our context we want to enable 'manage_image_cache'. Bug: T127755 Change-Id: I3f8c4c34dfc1f04064e4f68b875778bf30ce9d4a 22 February 2016, 22:36:15 UTC
a3ecb54 Tim Landscheidt 11 February 2016, 03:52:57 UTC Tools: Move role classes to module role Change-Id: I006d5eee5dc0365279e954b3c184ce20a818ef99 22 February 2016, 22:30:42 UTC
a9c808c dzahn 19 February 2016, 05:14:06 UTC wikimetrics: rename prod role, move to module/role - rename production role to ::production just like the description right below says, for consistent structure, like wikimetrics::staging - mode to module/role into autoload layout Change-Id: I3c0546bcba23b81ca636ca75efe2192ed15db0b0 22 February 2016, 22:05:05 UTC
919c26c Ori Livneh 22 February 2016, 21:19:02 UTC xhgui: enable PHP opcache Change-Id: I5fc3a30100784736bbe68ff8a2bc7c7262ea2197 22 February 2016, 21:19:46 UTC
605f76c Ori Livneh 22 February 2016, 21:09:29 UTC xhgui: increase memory limit to 512M 128M was causing requests to fail. Change-Id: I9cd7d823962845eb2fd60216407eb9e53424c5a8 22 February 2016, 21:10:21 UTC
2ef752b Andrew Otto 22 February 2016, 19:57:31 UTC Pass $service_name to service::deploy::scap in service::node otherwise it will try to restart the $repo on deploy. Change-Id: If134ed73647aa5cf2941bbb76a96ae4d14b6ee7d 22 February 2016, 19:57:31 UTC
56b9bdc Andrew Otto 22 February 2016, 18:55:12 UTC Add otto to aqs-admins Bug: T127720 T126294 Change-Id: Iebdea09a7bf7ec3b2ebf440942937a15ef45b26d 22 February 2016, 18:55:12 UTC
96801f6 Andrew Otto 22 February 2016, 18:45:08 UTC Use aqs-admins group for AQS deployment via deploy-service user - add aqs-admins to tin - allow aqs-admins to access deploy-service scap ssh key in keyholder Bug: T127720 T126294 Change-Id: I92a510496118dfdbfde5317d1a1eac58de4a328a 22 February 2016, 18:45:08 UTC
e1aa67c dzahn 22 February 2016, 18:26:23 UTC admin: add bast-only group for ppchelko ppchelko is in restbase-roots but also needs the bastion-only group to connect via bastions to restbase hosts follow-up to Bug:T126283 Change-Id: Idbfc5eb97b6f1c4cbe77efcfb94040fdca8d9380 22 February 2016, 18:34:21 UTC
937185b milimetric 22 February 2016, 18:24:59 UTC Fix bad module reference in AQS Change-Id: If4f123de1b77cdb2ad91fd7cc2b7c8edd32167a6 22 February 2016, 18:24:59 UTC
c75407a dzahn 09 February 2016, 05:49:06 UTC admin: add ppchelko to restbase-roots access request as in T126283, adds Petr Pchelko to restbase root users root is needed to deploy with ansible Bug:T126283 Change-Id: I1113ccf8bbabb76b191b2e0fce836c593bff5614 22 February 2016, 18:20:56 UTC
a6bb24b dzahn 19 February 2016, 06:01:08 UTC toollabs/mailrelay: fix top-scope vars without namespace Change-Id: I3644c133a7a4e304b25ac8dc55409c1b8df277d9 WARNING: top-scope variable being used without an explicit namespace on line 22 WARNING: top-scope variable being used without an explicit namespace on line 26 22 February 2016, 18:17:42 UTC
d9bb7d0 Brandon Black 22 February 2016, 17:54:40 UTC restore debdeploy hieradata for codfw parsoid This was accidentally deleted in: Ic176fb5d9b118b6c2357e76fe30aebc285a00a0c Change-Id: I39825beccb1e3815b05867eb9c6a5505af89fb25 22 February 2016, 17:54:40 UTC
a665ff8 Riccardo Coccioli 18 February 2016, 21:38:58 UTC mariadb: Add parallel gzip package Add pigz, used when transferring data beteen DBs Bug: T127385 Change-Id: I794fe9777a0e429d5398fda914b7e8cb69e06a7c 22 February 2016, 16:45:52 UTC
82eeae3 Brandon Black 21 February 2016, 14:20:48 UTC codfw: move cache_parsoid nodes to cache_upload We've never actually configured cache_parsoid in codfw for real use, and now with its impending decomission, we can reallocate these nodes to the upload cluster where they're more-useful. Bug: T110472 Change-Id: Ic176fb5d9b118b6c2357e76fe30aebc285a00a0c 22 February 2016, 16:23:50 UTC
dfaa0f2 Emanuele Rocca 20 February 2016, 12:32:09 UTC Reduce vcl_error redundancy Currently all cluster-specific vlc_error definitions call 'return (deliver)' once they're done with sub errorpage. Further, the cluster-independent version of vcl_error short-circuits on backend generated 400s and 413s to immediately deliver the page. Here we try to simplify the logic a bit by essentially including the 400/413 check in sub errorpage and always calling deliver at the end of errorpage itself. This way we can remove all the redundant 'return(deliver)' from cluster-specific versions of vcl_error. Change-Id: Ib93e1444d19ceac6279bc4c8b2343a94e20ee341 22 February 2016, 15:56:39 UTC
2d82f24 Marko Obrovac 18 February 2016, 22:23:32 UTC AQS: Separate AQS off of RESTBase For a first deploy, we tied AQS to use RESTBase's class. However, now that the restbase class simply uses service::node, there is no need to have AQQ point to RESTBase. Add analytics/aqs/deploy to Trebuchet's deployment.yaml to bootstrap aqs/deploy clone on tin Bug: T126294 Change-Id: I65758cf23926a0463e058f118e3192411d120dec 22 February 2016, 15:01:52 UTC
fd09c6f Ori Livneh 22 February 2016, 08:43:01 UTC statsv: set Restart=always in unit file Change-Id: Ib462012410d08aa90bb4ccfb0cc3ba365a0ea1ff 22 February 2016, 08:43:43 UTC
70309ce Tim Landscheidt 22 February 2016, 04:37:28 UTC Tools: Fix undefined variable in toollabs::kube2proxy Change 242439e352a261607a37542f8874a2a542030b33 removed the parameter $ensure from toollabs::kube2proxy, replaced it in its use for the /usr/local/sbin/kube2proxy file resource with the previous default "present", but left it untouched at the kube2proxy base::service_unit. Puppet apparently treats this undefined reference as "undef" and uses the default value for the $ensure parameter of base::service_unit, "present". This change makes this explicit. Change-Id: I3dd86ed1a2a48b2baa1240d9a73082643e306864 22 February 2016, 05:22:12 UTC
a6a74d5 Ori Livneh 22 February 2016, 04:10:31 UTC xhgui: require auth only for POST; allow anon GETs With Ie4c87619 and I219c0901 in place, sensitive data is not logged to xhgui, so it is OK to make the site world-readable. It should in theory be fine to allow POST as well (for searching, adding function watches, and creating custom views), but that increases the attack surface substantially without providing a ton of benefit, so keep that locked down. Change-Id: I754f424a0fd8e4e5e1e2256b9a759684321c4dd1 22 February 2016, 04:15:32 UTC
0b20964 Ori Livneh 20 February 2016, 21:31:00 UTC xhgui: add trailing slash to ProxyPass target URL ...to ensure that it is covered by the auth directives in <Location /> for the xhgui vhost. Change-Id: I4aed60c9158462f12697226bb14edfcf3ba00ca7 20 February 2016, 21:32:13 UTC
5795841 Ori Livneh 20 February 2016, 21:05:31 UTC xhgui: require auth for entire vhost Require authentication for the entire vhost by moving the `require ldap-group` directives to a <Location /> block. Change-Id: I6686903c901fe38cf59e7a22ce5b3b194e817c4b 20 February 2016, 21:08:35 UTC
ec2ef5e Alex Monk 20 February 2016, 17:47:11 UTC Fix pep8 compliance in modules/base/files/labs/puppetalert.py, use HTTPS Change-Id: I48122c476abe5b4934db5ba75c1d88aeb13c8274 20 February 2016, 20:47:02 UTC
8d46751 Ori Livneh 20 February 2016, 18:25:28 UTC Correct Wikitech URL in labs/puppetalert.py Change-Id: Iefdebb94f93e41c73282bb37554fc24d40893094 20 February 2016, 18:25:28 UTC
848c9fd Ori Livneh 20 February 2016, 11:21:52 UTC xhgui: require ops / nda / wmf ldap group The UI contains things like cookie headers. Change-Id: Ifda975fd3f473ab41dc1f9fe9be1291e03a70193 20 February 2016, 11:21:52 UTC
5ac2891 Ori Livneh 20 February 2016, 11:15:03 UTC Fix-up for Idc3065856: fix xhgui apache config I basically brute-forced this until it worked. I hate Apache sometime. Change-Id: Ib2cc3cb925d7923ff85d720e76a5f30b6ab3ab88 20 February 2016, 11:16:02 UTC
187a77d Ori Livneh 20 February 2016, 10:06:06 UTC role::xhgui: require php5-mongo; provision cache dir Follows Idc3065856. Change-Id: I33c5d3e1db2fceef22850c71aed5231ef1e21d09 20 February 2016, 10:06:06 UTC
7858b61 Ori Livneh 20 February 2016, 09:58:22 UTC Fix-up for Idc3065856: correct xhgui repo name Change-Id: Ie8f432ddad53a3220d046600deb39993dde631fa 20 February 2016, 09:58:22 UTC
b67b5f1 Ori Livneh 20 February 2016, 08:01:55 UTC Complete xhgui role - Deploy xhgui with Git. - Serve xhgui via Apache. - Proxy performance.wikimedia.org/xhgui to xhgui vhost. Change-Id: Idc306585643229439a0a992d160e6c114efe11b3 20 February 2016, 09:52:13 UTC
d0c9091 dzahn 19 February 2016, 05:29:14 UTC role/lists: rename prod role, move to module/role - move role classes to module/role - rename ::server class to match description and to fit into this scheme avoiding the limitation of the role keyword - move into autoloader layout - minimal docs / lint-fix Change-Id: I4834b14053c5f799b3df4514188fbbec0d9a806e 19 February 2016, 23:53:59 UTC
3f81363 Tim Landscheidt 19 February 2016, 04:50:32 UTC role: Align arrows in manifests Change-Id: I7f2777652de46349e175ab0e8c68a5a840e866d6 19 February 2016, 23:13:12 UTC
1065c1f Ori Livneh 19 February 2016, 22:25:46 UTC Add role::xhgui; apply on hafnium For now, this provisions a MongoDB instance. This is what XHGUI uses to store profiling data. The app servers will be configured to write profiling data to this MongoDB instance. Change-Id: I6c34c479f00b7b82aad12463998df22f2cb2c7f1 19 February 2016, 22:27:29 UTC
8a0ea43 Andrew Otto 19 February 2016, 18:48:08 UTC Adding iTerm2 shell integration for otto's bash shell Change-Id: Ieddcd31ee758075f1a62271870a732362859ab46 19 February 2016, 18:48:08 UTC
b1666b4 Jaime Crespo 19 February 2016, 15:00:33 UTC Add nagios@localhost the permissions to read the heartbeat table This will allow the nagios-npre user to read the replication lag table. Bug: T114752 Change-Id: I089d159a758fea0a25087bfc1b1fb8dc712160b1 19 February 2016, 18:09:51 UTC
1f67dac Filippo Giunchedi 18 February 2016, 17:30:38 UTC restbase: adjust alarms with new metric names following the internal/external split Change-Id: I237e8a27ab3eaa5f700a84b4609442cbcf2b761e 19 February 2016, 17:51:28 UTC
ea777a9 Greg Grossmeier 19 February 2016, 16:57:45 UTC Fix sprint rewrite rule Bug: T127348 Change-Id: I96e3007c0a92dc01b6219999ebac94d3710bd0d1 19 February 2016, 17:20:00 UTC
9e498d0 cpettet 19 February 2016, 17:11:36 UTC sge: ensure grid master is running Previously this was misconfigured and we removed it during triage to prevent multiple masters from starting suspected of corruption. Change-Id: I2bba4d3701d159456589c3541eb9f793a2652845 19 February 2016, 17:11:36 UTC
b6ee132 Brandon Black 19 February 2016, 11:00:21 UTC cache_text: mangle RB requests just before fetch This switches RB request mangling to happen just before fetch in miss/pass, like the mobile hostname mangling. The effect is that the original URL is consistently the one that matters for all VCL purposes (including purges). There's a one-shot side effect that this will effectively invalidate the backend caches for RB content on deploy (but the frontends should cover that reasonably in terms of perf spike). Bug: T127370 Change-Id: I9905c9c4c265cacc0a3566e6fac4f1cc1f9c3f63 19 February 2016, 16:44:26 UTC
3f268a4 dzahn 19 February 2016, 04:56:36 UTC otrs: fix role name, move to module/role - fix role class name to match role description - move to module/role into autoload layout - fix lint warnings - add minimal docs and WP link Change-Id: I32f5c8627812604295febf61219bfd1d8886103c 19 February 2016, 11:15:44 UTC
57e263d Ori Livneh 19 February 2016, 00:47:22 UTC Fix-up for I00db1abc3: disable LightProcess for fcgi as well Change-Id: Ida79fcc20fe0d4765d6ca3ccb7fe68a8c216e8e2 19 February 2016, 00:47:22 UTC
back to top