https://github.com/wikimedia/operations-puppet

sort by:
Revision Author Date Message Commit Date
253f838 Auto-merge of 05e646c89fdc620fb2757775c2a5e9a7b6031a8d 03 June 2014, 17:56:50 UTC
37828fd Merge "beta: Remove File['/usr/local/apache/common'] from ::beta::common" into production 03 June 2014, 17:47:18 UTC
7c7fdda Merge "add newline to researchdb pw file, easier to read" into production 03 June 2014, 17:36:19 UTC
c424ce8 admin yam to aluminium.wikimedia.org Change-Id: Ia164ab5ed53e47f721285163b2f209ec34e24b73 03 June 2014, 17:34:10 UTC
aa13595 add newline to researchdb pw file, easier to read Change-Id: I8a51e1513c3db9a1a5e6f9977cb4b7e931d6e46c 03 June 2014, 17:30:51 UTC
386af25 run all maintenance crons as apache user per quote from Tim "everything has to su to apache before running maintenance scripts" https://wikitech.wikimedia.org/wiki/UID Change-Id: I70bc7bb631b6d953325a2979404a097189b908ff 03 June 2014, 17:16:06 UTC
8e555fe Fix typo in 2b16da9b4137 Change-Id: I4033b405dc2a8139c2eb3e425a3475fc5dc7c317 03 June 2014, 16:59:39 UTC
1c2492c Merge "Remove searchidx1001 from scap targets" into production 03 June 2014, 16:56:42 UTC
20f408d Merge "[gdash] Add yearly graphs for frontend performance" into production 03 June 2014, 16:50:29 UTC
6bc7227 toollabs: Add MongoDB role Change-Id: I832f9017ce51d889ba6e6f20781f78e1da0200a1 03 June 2014, 16:17:58 UTC
64a49c0 [gdash] Add yearly graphs for frontend performance Change-Id: I19e03dfc52400d9f1e112dc9941ad03e322e2460 03 June 2014, 16:15:55 UTC
a7d3495 syntax error in admins.pp Change-Id: I67456147ff8f2eb06d3e9feab7cebe9816cbecad 03 June 2014, 15:52:12 UTC
ba2e6d3 rainman account absented Change-Id: I396792986bda6a2a4530f2529c893924d69675c6 03 June 2014, 15:42:54 UTC
443a56c Specify table and chain for role::dns::recursor notrack rules NOTRACK is only valid in raw table and OUTPUT and PREROUTING chains Change-Id: Iab328adecdd437bc69bda81ed7d0a16988fbda29 03 June 2014, 15:34:09 UTC
696e3d0 Merge "Fix missing semicolon in ferm in role::dns::recursor" into production 03 June 2014, 15:26:28 UTC
8717352 adding user notes to admin yaml Change-Id: I71c3e10c171c6ebf8d20d1fc1c4e293c7a738e60 03 June 2014, 15:25:54 UTC
069de6a admins.pp deprecation warning Change-Id: I23c3e4d707f9fb48596a4a0e7da2d5931594a77d 03 June 2014, 15:25:33 UTC
a2887c1 Fix missing semicolon in ferm in role::dns::recursor role::dns::recursor has rules for avoid connection tracking for DNS. Unfortunately they had a typo, a missing semicolon at the end. Fix that Change-Id: I651aba5138e715ab5b28c2dae3e167592b50d081 03 June 2014, 15:22:57 UTC
43557e4 dns recurses: add firewll Change-Id: I47de949e10161f8bf468b4bd5d4a3a3dd95b8dcd 03 June 2014, 15:10:52 UTC
76254b6 rm old admins::mortals class, replaced by yaml Change-Id: Ie9dd91ab6f7eb21686a5794d0f722ac3e1eba738 03 June 2014, 15:06:50 UTC
0255ced admin: files for /home/hashar Change-Id: I2ff976340cdb257bff161b18b30c17f5a7cd27e0 03 June 2014, 14:53:44 UTC
232e287 mediawiki: add python-imaging to required packages PIL is being used to generate captchas. RT: 6670 Change-Id: I81235b3e76f37ccd7d53e5cd46b9d455d5d85b14 03 June 2014, 14:26:48 UTC
65d82b3 use analytics-users group vs. stats group "stats is not a valid group name" use the analytics-users group instead like on other nodes Change-Id: Ia90d90ba3f89384196da6dc018d7b113e58e95bd 03 June 2014, 14:12:17 UTC
21be64b Add custom Diamond collector for RCStream This change depends on change Ia9de677c8 in mediawiki/services/rcstream. It provisions a custom Diamond metrics collector for RCStream which connects to all configured backends and reports the total count of connected clients and the maximum queue size. Change-Id: I37f3f38c9f274f17a0b6e8a430d7a75f8486dc74 03 June 2014, 14:03:18 UTC
ec19c23 Allow Elasticsearch java version to float again We used to force 7u25 because the newer version of java had issues with Lucene. These were bugs in Java, not Lucene. Anyway, its been fixed and available in 7u55 which is the newest version. Change-Id: I662a0e1fc5e2e3047ef6c00597504a9a35fd62fa 03 June 2014, 13:47:45 UTC
2b16da9 monitoring: monitor mediawiki jobs As a consequence of the last jobrunners outage, we add checks on the job queue length both for queued and running jobs. With these checks, we would have been warned of the problem way earlier. Change-Id: I054e11cea019369813eb92e2c62e7e10c5ac47cb Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> 03 June 2014, 13:29:17 UTC
10875c9 jobrunner: restart with full path. Change-Id: I299eac50e11ebcaac91227b137fba6295016e40b Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> 03 June 2014, 11:15:47 UTC
d619d67 use m2 CNAME for exim Change-Id: I514c7a400b11ef5802cf9bc8f3b2e352ebea333e 03 June 2014, 08:43:15 UTC
6f8ecec mediawiki: make jobrunners restart work. - Added the PATH needed by the MW jobrunner restart script - Corrected the cron scheduling, with randomization Change-Id: I53fa7ee981d7b5f0d6b25c64c6808ca92f7c79a8 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> 03 June 2014, 06:34:38 UTC
76212ae Remove searchidx1001 from scap targets searchidx* already has a cron that copies the required configuration directly from tin rather than the local /usr/local/apache/common/* Fun tip: we only need the messages files, InitialiseSettings and all.dblist. The rest of MW is unused. Copying to /usr like it's an Apache is pointless and doesn't acheive anything other than filling up / and spamming IRC with low disk warnings. Change-Id: I9e74476e02b113470d8c05d538feb4513bd62964 03 June 2014, 02:51:52 UTC
b120464 Use m2-master CNAME to make DB rotations neater. This allows a master switch to be a DNS change plus a simple port 3306 tcp redirect with socat until TTL. Should also help if we switch to a haproxy configuration in the future. Change-Id: I1fef849d7368679920de974ed3a9f8d1b1fab538 03 June 2014, 02:27:23 UTC
2f1b295 beta: Remove File['/usr/local/apache/common'] from ::beta::common Fixes: err: Could not retrieve catalog from remote server: Error 400 on SERVER: Duplicate definition: File[/usr/local/apache/common] is already defined in file /etc/puppet/modules/mediawiki/manifests/sync.pp at line 41; cannot redefine at /etc/puppet/modules/beta/manifests/common.pp:20 on node i-0000011b.eqiad.wmflabs Change-Id: I957dcba6a01767828871d77e1f6934ee125fae69 03 June 2014, 01:09:36 UTC
d2fb5f2 fix ethtool command for ring as well Change-Id: I25e3e66b6f22770d82aae429b0b82d18b365f6cf 03 June 2014, 00:43:09 UTC
8252098 fix s/ring/setting/ in interface::ring usage Change-Id: I71a250c73c52b0299175271b0d47b71cdc5526cb 03 June 2014, 00:39:34 UTC
35a3f4a Set rx ring params for bnx2x on 10GbE LVS Change-Id: Ibc5018145bbbdb9289525070e866c69a8efea9c4 03 June 2014, 00:35:12 UTC
0bf93a8 Periodically restart job runners to avoid pipeline shrinking issue Change-Id: I04b63144c293a7add925a919be65a97247faf443 03 June 2014, 00:29:55 UTC
6cd280c replace Service['procps'] with an Exec procps is a task job, not a service, and it is only hackery that makes it appear like a well-behaved service. So replace it with a refreshonly exec. Change-Id: I8edb51a519c68050f577aebea1df9045b2c4b933 03 June 2014, 00:27:32 UTC
324d054 mediawiki::web: make $maxclients numeric; simplify config Change-Id: I12beed5e993d750be8a1b506a6713cf3e2c1dcba 03 June 2014, 00:22:05 UTC
4000fa3 get rid of {jobrunner,videoscaler}-apache-service-stopped Execs The wikimedia-task-appserver package is no longer installed, so these are not needed. Change-Id: Icb931cf3be252285b0a0349907e9bfadd99328e0 03 June 2014, 00:21:18 UTC
c0083fc role::mediawiki::job_runner: same config for beta & prod beta was configured to have fewer processes, presumably because the job runner VMs have fewer cores. But there are very rarely enough jobs in beta to fill up the queue anyway, and some thrashing is acceptable. Change-Id: I6fbe0e7d829fc64344474d8ac2a00a1409c3e40a 03 June 2014, 00:20:36 UTC
3259a3f Bump vm.min_free_kbytes on LVS/cache nodes Also, introduces new module "vm" and vm::min_free_kbytes() to set a somewhat-dynamic value based on memory size that's safe and reasonable, and moves old generic::higher_min_free_kbytes to vm::higher_min_free_kbytes as well (without change). Change-Id: I4682a08d967f873bd069aaf0c17ad2b02a0d4d78 02 June 2014, 23:34:36 UTC
e5745c7 File['/a/common']: symlink to common-local; replace => False Before If74ea635f, File['/a/common'] was a directory with replace => false because "# FIXME: /a/common is a symlink on some systems". The system (singular) was terbium. In fact, it ought to be a symlink everywhere except on tin, so make it a symlink and set it to replace => false. Change-Id: If923e3a6c09ac5255a25828dc56ef903460064ad 02 June 2014, 23:09:42 UTC
eea89f2 Clean up system::role * Add documentation * Make description parameter optional, to avoid things like: system::role { 'misc::fundraising::mail': description => 'fundraising mail server' } system::role { "misc::management::ipmi": description => "IPMI Management" } system::role { "misc::noc-wikimedia": description => "noc.wikimedia.org" } * Remove dead code I'm still not sure this deserves to be its own module, but at least it's a bit better now. Change-Id: Ic36f85461f6eff5ef521fef4ef4ab27e91de4a4d 02 June 2014, 22:48:59 UTC
07bfd0b hafnium: add firewall Change-Id: I959a83c72d3a8f68c9ced65d1ee4259e627d7e66 02 June 2014, 22:04:10 UTC
626e765 Merge "Remove declaration of File['/a'] from imagescaler manifest" into production 02 June 2014, 21:55:58 UTC
f07d1f6 Remove declaration of File['/a'] from imagescaler manifest Used to be provisioned by wikimedia-task-appserver; now declared in mediawiki::sync. Change-Id: I9b6013b3ab497ecb04699a6f5f970f2aaa31adb7 02 June 2014, 21:49:16 UTC
052a0a6 Merge "stat1010: stats group -> analytics user group" into production 02 June 2014, 21:48:55 UTC
4a63400 stat1010: stats group -> analytics user group puppet error on stat1010: Error 400 on SERVER: stats is not a valid group name talked to Chase, like on 1026 this should be the analytics user group instead Change-Id: Ia74f580c770cc245f6f4d082e37c85da52f179e6 02 June 2014, 21:44:55 UTC
a8d8047 rm old admins::pmacct class, replaced by yaml Change-Id: Ibbcafebe262b69ab914880586d7cfb8217bf33bb 02 June 2014, 21:29:53 UTC
9762372 Merge "rm old admins::parsoid class, replaced by yaml" into production 02 June 2014, 21:16:12 UTC
a43a568 rm old admins::parsoid class, replaced by yaml Change-Id: I490253d790bd6d5b43073a26d92ee18a2a5e1335 02 June 2014, 21:06:35 UTC
1c491dd Add zerodot support for 520-18. Change-Id: Icd00fd2930939d9714439db87d5e84f57a7e54fe 02 June 2014, 21:00:18 UTC
d989c07 add yaml group for PDF QA users, switch tantalum add a new group for these users and switch tantalum over to admin yaml Change-Id: I1d8c2a74ea7f8f623087a9f71227f14b933a365c 02 June 2014, 20:47:29 UTC
84e0352 admin yaml labstore* seems the gid's are synced. root@labstore1001:~# getent group ops ops:*:700 Change-Id: Id923ca79d1545d53ec92c019ce85a1acc3e1e73e 02 June 2014, 20:46:12 UTC
199dff1 admin yaml /virt100[8-9].eqiad.wmnet/ Change-Id: Ia61610852e22c2e65c098aa96240760ba7e64e01 02 June 2014, 20:41:14 UTC
646230f Move beta-specific configs from role::mediawiki::common to role::mediawiki::appserver::beta Change-Id: I2d19838c173441ced72d51074ea22bc0cbbcfc8f 02 June 2014, 20:17:02 UTC
fdc2044 role::mediawiki::common: remove if $::realm == production guard We run twemproxy in labs now, so we don't have to guard against setting up the checks there. Change-Id: Ie2d150eb82dd1c1b995a88181966418e57c8d68a 02 June 2014, 20:16:26 UTC
d9f61a2 Merge "Comment out diamond::collector::nginx on rcs100x" into production 02 June 2014, 20:12:47 UTC
b7f2271 admin yaml analytics1026 refs 'analytics-users' Change-Id: I7861f6a594051e36e5fcda2293f92b1486374f0f 02 June 2014, 20:08:23 UTC
27d3830 Comment out diamond::collector::nginx on rcs100x Diamond hasn't been packaged for Trusty yet. Change-Id: Iac330d8385c2d4902263c8104f4ac562c6b3872a 02 June 2014, 20:07:29 UTC
2565fd3 cleanup old ircd and ekrem REVERT THIS TO FAIL BACK TO EKREM IRCD Change-Id: Icf6240a4655d83ceaf0e5de0803046f07858392e 02 June 2014, 19:54:43 UTC
1b6a5e6 Merge "include ::diamond in ::diamond::collector" into production 02 June 2014, 19:53:31 UTC
5ba61a7 Use proper analytics/refinery git url Change-Id: Ia60f69d263868d14b80aeedaf4beb6038dda50b5 02 June 2014, 19:51:50 UTC
3819adc include ::diamond in ::diamond::collector Otherwise the require => File['/etc/diamond/collectors'] is liable to fail. Change-Id: Ibc4b98f99e7c79c3132edb09db1eab5da7d43875 02 June 2014, 19:50:39 UTC
decbec4 Add deployment config for analytics/refinery (this is the repository that will replace analytics/kraken). Change-Id: I5715cf6a1c897c15e9bf933ea6c2cea179769c08 02 June 2014, 19:50:09 UTC
df3b002 Merge "rm admins::ldap, replaced by ldap-admins yaml" into production 02 June 2014, 19:48:08 UTC
b6ba928 Merge "rcstream: add 'stream' subcommand to rcstreamctl" into production 02 June 2014, 19:44:48 UTC
39b53bb Merge "diamond: add diamond::collector::nginx resource" into production 02 June 2014, 19:44:31 UTC
869ec7e admin yaml sanger Change-Id: I8110589d7ce1a3f462da9f1bfc85acf81138d556 02 June 2014, 19:42:38 UTC
c087175 diamond: add diamond::collector::nginx resource * Update nginx submodule for I9450b8cf0. * See <https://github.com/BrightcoveOS/Diamond/wiki/collectors-NginxCollector> for details. * Provision a collector for the rcstream Nginxes. Change-Id: I79611947724ba6dc424c1267128687abd30c10bb 02 June 2014, 19:42:05 UTC
1474525 rcstream: add 'stream' subcommand to rcstreamctl Running 'rcstreamctl stream' connects to redis and PSUBSCRIBEs to 'rc.*', as a quick way of debugging whether edits are streaming in. Change-Id: Ib585234564939275af75af8627c56e5e9f0dd5df 02 June 2014, 19:41:34 UTC
433f00a Fix /a/common duplicate def'n Introduced in I02c9c85d9. The canonical definition ought to be in the MW module, so remove the other. Change-Id: If74ea635ff48ce49534ff52372a8824605afd6e0 02 June 2014, 19:38:24 UTC
f3cd785 rm admins::ldap, replaced by ldap-admins yaml Change-Id: Ifd7bbb1d6b4f7bdf965cdb8150a87f12801b2f83 02 June 2014, 19:35:32 UTC
e2c221f rcstream: enroll in ganglia; add system role * Add a Ganglia cluster entry for rcs1001 & rcs1002 * Specify rcs1001 as aggregator * Add system::role entry Change-Id: I08835e7f346f2a3144672fb27bb846be0f94e506 02 June 2014, 19:16:16 UTC
d4a6125 Add service check for rcstream backends Adapt the rcstreamctl shell script to include a 'check' subcommand that emits Icinga-compatible output / exit status, and configure it as a service check. Change-Id: I7319d5fd67747147bd6628c7f109cecdcb0d14c6 02 June 2014, 19:14:53 UTC
06f74bc absent csalvia admin yaml Change-Id: Ia54aaab0d75873b14959afa08427ff1d2d4eb60e 02 June 2014, 19:10:19 UTC
35b6f43 Get rid of role::mediawiki::appserver::test This dates back to when mw1017 was used to test the app server stack on Precise. The only difference between role::mediawiki::appserver::test and include role::mediawiki::appserver is the maxclients setting (60 vs. 100), which isn't usefully testing anything. So chuck it. Change-Id: I15dd145a13bb4f767a13e101d0e95fef80b3ccfa 02 June 2014, 19:06:16 UTC
5b9fd5a Merge "mediawiki::web: Remove $::lsbdistrelease guard" into production 02 June 2014, 19:04:35 UTC
9572365 Merge "mediawiki::sync: mkdir /a/common" into production 02 June 2014, 19:03:59 UTC
62b63f1 admin yaml stat* and analytics fixups There is a use case for the stats group not easily handled by a centralized grouping system. The workaround here is to allow group names in yaml to not be the on box final name. including: include accounts::dr0ptp4kt # RT 7569 include accounts::brion # RT 7568 Change-Id: I6a3ef6d1cb5d835f3cdb9bbf2ca8bad099af2ecf 02 June 2014, 17:56:19 UTC
14ddc98 ircd stats enabled vs enable Change-Id: If3555db52710507081a543ceabc509b6dbbdb623 02 June 2014, 17:26:16 UTC
74e4240 Only run ifup for tagged if something changed Change-Id: I5db57cadfb039b5b5ed8ce332dca319f9d9a4c8b 02 June 2014, 16:37:28 UTC
534efca only re-apply RPS on change Change-Id: I974c59109abb52b36ba0918a8fa47af4355a4349 02 June 2014, 16:26:58 UTC
842c39f Not saving Kafka jmx metrics via jmxtrans to an outfile on disk We don't use this file ever, and it just takes up space! Change-Id: If6af6bb648ced0c134a4eeefe30ad8290ec676d8 02 June 2014, 15:51:31 UTC
260ee56 Add path param for txqueuelen check (I hate you, puppet) Change-Id: Ic24c8f22565a0aa63b89b4d377e8f1c8c5e9fc37 02 June 2014, 15:16:14 UTC
17d3c89 txqueuelen 10K for 10Gbps LVS interfaces Change-Id: I840fab97dc7e57c4bf37dc43f4b3b9fc40e508d7 02 June 2014, 14:56:51 UTC
7a41d85 ircd_stats path fix Change-Id: I7a43901ea91dcfb5c08dd1b1b0d3728732bc9cc1 02 June 2014, 14:48:19 UTC
62935c6 send user and channel count to statsd for ircd looks like: servers.argon.ircd.users 5 1400868275 servers.argon.ircd.channels 271 1400868275 Change-Id: I8352e4804d225a8bcb03a6ab28b82c935a3c4d67 02 June 2014, 14:44:43 UTC
77853ce Merge "disabling/removing users kwang and csalvia from admins, site.pp, data.yaml, icinga contact groups" into production 02 June 2014, 14:32:00 UTC
29586c7 keep rc-pmtpa name for now I didn't see any mention of changing the broadcast bot name in the notice that went out. I would rather keep the changes to a minimum for migration, especially without lots of notification. Change-Id: I5951d2159a026d19ee929a5ac995f08a8569c215 02 June 2014, 14:27:30 UTC
762473c disabling/removing users kwang and csalvia from admins, site.pp, data.yaml, icinga contact groups Change-Id: Ie4f99b27888b151863dfd7d8cd5cda62f2559faa 02 June 2014, 14:21:45 UTC
2e14c11 Merge "Tools: Install user-requested packages" into production 02 June 2014, 13:55:18 UTC
4d3f9ba Add bgwriter ganglia stats to postgresql Background writer buffers inserted into ganglia stats Change-Id: If12667592094d5a0038d1db3ac695bb4eeca4744 02 June 2014, 13:30:46 UTC
a99ee29 granting brion and dr0ptp4kt stat1003 access RT768 and RT7569 Change-Id: I03c56d9492a80c283c2afbeb6c1c3f424024f172 02 June 2014, 13:12:53 UTC
d37109d set datasets user shell back to /bin/bash Change-Id: I600769b907be8e6aa7a0d3ab4f30f39d2b04f39e 02 June 2014, 09:27:38 UTC
ec0eda5 admin yaml for tin Change-Id: I35780b2c9e023ddff5fee26ac7ad872ab4803ccb 30 May 2014, 22:08:48 UTC
69908a8 admin yaml for wtp* (parsoid) Change-Id: I2f147976e1c28a580772d3fa6c4ae10df3420489 30 May 2014, 22:06:35 UTC
3e50e3c admin yaml terbium.eqiad.wmnet typo Change-Id: Ifd09c262089d076b66f75feccce78dec55cf721b 30 May 2014, 21:31:46 UTC
8477ccb admin yaml terbium.eqiad.wmnet Change-Id: I715869bbef12f12fe4995266c620bc7356bc34ec 30 May 2014, 21:30:08 UTC
ed896cc admin yaml for labnet1001 Change-Id: I0bc558733823d28c2afce1f8187294497ea9e9d0 30 May 2014, 21:26:11 UTC
back to top