| 5c396d2 | Andrew Otto | 01 December 2014, 21:46:33 UTC | Include new class misc::statistics::packages::utilities on stat1002 and stat1003 This includes packages that Ironholds has requested so that are depdendencies fr utilities he builds for R. Change-Id: I9dbf08f1b6f946a61723b0ca9214ee71572fb533 | 01 December 2014, 21:46:33 UTC |
| 3fc5fba | dzahn | 01 December 2014, 19:58:35 UTC | Merge "add bmansurov to researchers admin group" into production | 01 December 2014, 19:58:35 UTC |
| fed4a96 | Coren | 01 December 2014, 18:48:39 UTC | Merge "Add codfw labs support and labstores to DHCP" into production | 01 December 2014, 18:48:39 UTC |
| 8ff5b6b | dzahn | 01 December 2014, 18:40:21 UTC | add bmansurov to researchers admin group Change-Id: I165a2f149c07d7c1d19c2d6169e42c2fed1250e3 RT: 8887 | 01 December 2014, 18:43:33 UTC |
| 4d90c82 | Brandon Black | 01 December 2014, 17:43:58 UTC | Switch SSL loadbalancing to sh scheduler This is not optimal, but it's better than 'wrr' given the current scenario with pybal, ipvsadm, and our LVS kernels. Should reduce renegotiations considerably, other than spike events when a server is downed or upped by pybal. Change-Id: I296ca2f8a6a5e135c3b97c892241acbbfc868d96 | 01 December 2014, 17:44:05 UTC |
| e569770 | Coren | 01 December 2014, 17:23:01 UTC | Add codfw labs support and labstores to DHCP Change-Id: I40187504cd7688562543ce84e94a7c420e2377c5 | 01 December 2014, 17:36:42 UTC |
| e2b038d | Brandon Black | 24 November 2014, 17:00:33 UTC | Remove old protoproxy / ssl[13]00x config / star certs Change-Id: I5f3f7cbd6a1c8bce58bf14d04cc981baf53edc85 | 01 December 2014, 15:33:08 UTC |
| 601bb45 | Faidon Liambotis | 01 December 2014, 14:30:27 UTC | ldap: neptunium is eqiad, virt1000 is no more The monitoring change (see previous commit) revealead another error: neptunium, an eqiad host, was configured with an "ldap-codfw" certificate. Of course, this had no effect on the actual OpenDJ configuration, probably because OpenDJ puppet manifests are broken in some (many?) ways. It did affect the monitoring check, though, which noticed this discrepancy. Fix this, and while at it also remove the virt1000 stanza since the LDAP role was removed from this host with commit 0510711. Change-Id: If491512e18c4180d99957552746c3a92aec6d35d | 01 December 2014, 14:33:24 UTC |
| 70e97e5 | Faidon Liambotis | 01 December 2014, 14:07:34 UTC | ldap: fix LDAP's monitoring::service CN matching Ignore this module's series of antipatterns and fix an underlying issue: the monitoring alert is set up to check for a CN that is the host's FQDN, which isn't a valid assumption as there is no gurantee those two will match (and in fact, do not match for ldap-eqiad & ldap-codfw). Modify the alert to check instead for the certificate's subject name which we conveniently have under the $certificate variable already. Change-Id: I99b26723d554750169bddceda1508af856334bcd | 01 December 2014, 14:07:34 UTC |
| 6bea4e5 | Faidon Liambotis | 01 December 2014, 13:51:30 UTC | hhvm: remove check_procs' WARNING state Checking for an absolute == 1 number of HHVM procs isn't really a good strategy as it comes with a lot of false positives due to HHVM processes being spawned by MediaWiki. Remove the warning state of the check for now, relying only on the critical threshold of < 1 processes. Change-Id: I94f71e139828907500455ee836eea8fec1f8bf3c | 01 December 2014, 13:55:08 UTC |
| b05f928 | Faidon Liambotis | 01 December 2014, 13:33:51 UTC | Merge "Unbreak misc::statistics on <= precise systems" into production | 01 December 2014, 13:33:51 UTC |
| fa501ac | Faidon Liambotis | 01 December 2014, 13:28:38 UTC | Unbreak misc::statistics on <= precise systems Commits f36ea9f3, 5c5cda10, c67c4aee added libyaml-cpp0.3, libyaml-cpp0.3-dev & libgoogle-glog-dev to misc::statistics::packages. However, this (incredibly generic) class is also being included in 12.04 systems, like stat1001 & helium, which do not include these packages. Revert those commits as puppet has been broken for 12 days on those systems despite this being a known problem. Change-Id: Iba7fb2c08e1523b53857c1a7ce583d16a5802fb4 | 01 December 2014, 13:32:09 UTC |
| 40b0a6e | Giuseppe Lavagetto | 01 December 2014, 12:53:48 UTC | monitoring: refine alarms on HHVM Since we're not interested in single spikes of traffic/load, but rather to sustained issues with HHVM performance on one server that may need our attention; so we need 30% of the datapoints over a 10-minute period to be affected before we send an alarm. Change-Id: I11ae78ea91a4998076f81a4aaee2e4054e6d19a1 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 01 December 2014, 12:53:48 UTC |
| 42b07f3 | Kunal Mehta | 01 December 2014, 06:57:49 UTC | extdist: Add composer location to config Bug: T70940 Change-Id: Ibb5b6c8649d747e42ee9fb4f9aeaec269500a66c | 01 December 2014, 06:58:17 UTC |
| da8ef33 | Giuseppe Lavagetto | 30 November 2014, 10:22:08 UTC | jobrunners: revert to rdb1001 Change-Id: Ia444074967fa79af48a572c94f60bf6ba0c87a5f Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 30 November 2014, 10:22:08 UTC |
| d3a2b1e | Giuseppe Lavagetto | 30 November 2014, 08:21:09 UTC | jobrunner: fix the jobrunners, not just the videoscalers Change-Id: I3e709cb60bc44d9a6c49e8227ae630fe4ed8a1c4 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 30 November 2014, 08:21:09 UTC |
| 5e67acc | Giuseppe Lavagetto | 30 November 2014, 08:10:51 UTC | jobrunner: failover the redis server in use Change-Id: I2fea105d79ae5445ca2bef274ac0f4f0508c2b9b Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 30 November 2014, 08:10:51 UTC |
| 1f02de9 | YuviPanda | 13 November 2014, 19:58:22 UTC | shinken: Setup IRC notification for shinken Also remove unused perfdata commands from notification config Change-Id: I9a9880fdf414cdf23a114651f917e97ea701b3f1 | 29 November 2014, 00:36:23 UTC |
| 59cae37 | Ori Livneh | 28 November 2014, 23:22:50 UTC | hhvm: Don't load tidy.so Segfaults. Change-Id: I335792043172e4f966bf3e2d74cd73fe898b0041 | 28 November 2014, 23:22:50 UTC |
| 6d0d732 | Ori Livneh | 28 November 2014, 23:02:55 UTC | HHVM: provision hhvm-tidy and load tidy.so Bug: T578 Change-Id: I1143954b539b0f47d284b071645c57417be8bb25 | 28 November 2014, 23:02:55 UTC |
| 7cc72cc | YuviPanda | 28 November 2014, 13:34:30 UTC | tools: Add postgis package to exec_environ Bug: T76226 Change-Id: If4ac5f88d2430f5560cd6f0954f1068ad74dd4fe | 28 November 2014, 13:34:30 UTC |
| ef83c80 | Giuseppe Lavagetto | 26 November 2014, 09:47:36 UTC | reimage: add a few configs, beautify output Change-Id: I255e94564efcd2fe8c481c3ecbf3d4a949f4cb2b Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 28 November 2014, 11:27:06 UTC |
| bfcd152 | YuviPanda | 27 November 2014, 21:03:06 UTC | tools: Update update-scripts.sh to update updated Packages properly We don't use gzipped Package lists anymore on tools, since (IIRC) it caused apt-get update to get stuck sometimes. Change-Id: If2fdf2976f8a28d7812c8c138be707b0c90f5d04 | 27 November 2014, 21:03:06 UTC |
| 260323d | Kunal Mehta | 27 November 2014, 20:09:11 UTC | extdist: composer depends on php5-cli Change-Id: Ibe421ff94355fb7e1f18207d59086779f7fb6d89 | 27 November 2014, 20:09:11 UTC |
| 60860a9 | Kunal Mehta | 27 November 2014, 19:33:50 UTC | extdist: clone composer into /srv/composer Bug: T70940 Change-Id: I9c4c655eec649019bd40bcd397e9e8a435e448d2 | 27 November 2014, 20:00:04 UTC |
| f5ec8a3 | Giuseppe Lavagetto | 27 November 2014, 16:20:02 UTC | graphite/txstatsd: re-introduce require_package We need require_package as the two classes have overlapping package needs; however when using require_package, the resource is not always declared within the same module, so we just require to have the stub class require_package creates Change-Id: I8ce1727d4ebf180d98d507fe0d41147887e3579b Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 27 November 2014, 16:29:41 UTC |
| 1a5ebee | Giuseppe Lavagetto | 27 November 2014, 11:01:12 UTC | Revert "ocg: Temp hack to bypass hiera for redis passwords" This reverts commit cf6ff6f9e3c6aa7586d5b17ca51d8eaf7648b5a4. Change-Id: I7666c442772eb28eaf696a5d10a54a6bd19e73a7 | 27 November 2014, 11:02:29 UTC |
| b699204 | Bryan Davis | 27 November 2014, 01:11:21 UTC | labs: make mwyaml hiera backend first lookup Change the backend order for hiera lookups in labs so that settings provided by the mwyaml backend will take precedence over settings made in the on disk yaml files. This was found to be necessary by experimentation in the beta cluster. Change-Id: Ib718cc83521d0268d859a955feb3615f277a51d2 | 27 November 2014, 09:15:07 UTC |
| 8fe0944 | Sean Pringle | 27 November 2014, 03:32:30 UTC | switch remaining m2 boxen to role::mariadb::misc Change-Id: I594f735ffe30347d9509ae4e08bba8fa74aeba92 | 27 November 2014, 03:32:30 UTC |
| 48b9442 | Dzahn | 27 November 2014, 00:22:46 UTC | Merge "Further improvements to OCG apparmor profile." into production | 27 November 2014, 00:22:46 UTC |
| a5347a6 | C. Scott Ananian | 26 November 2014, 22:32:15 UTC | Further improvements to OCG apparmor profile. * Allow OCG to read default papersize. * Allow rsvg-convert to read mime type database. * Allow imagetools to read their own binaries. * Allow imagetools and xetex to write to postmortems. Change-Id: Id23b7ac9dd1cd782d6172de89052b299fb897707 | 27 November 2014, 00:18:31 UTC |
| df6b33e | Yuvipanda | 27 November 2014, 00:09:44 UTC | Revert "ocg: Temp hack to fix empty redis password, take 2" Bah, https://gerrit.wikimedia.org/r/#/c/176181/ probably did work, I just didn't check properly. This reverts commit 8da2bc7b3c370a1afae90805b25104ccf44554e5. Change-Id: Ib787fcf87c6a11ef9d8870cd16c1434fb1e20ce0 | 27 November 2014, 00:09:44 UTC |
| 8da2bc7 | YuviPanda | 27 November 2014, 00:01:05 UTC | ocg: Temp hack to fix empty redis password, take 2 https://gerrit.wikimedia.org/r/#/c/176181/ didn't actually have any effect at all. Perhaps the bug is in our heira backend, and it somehow sets the password to ''? This patch should verify that. Change-Id: I0df8f404f50617cf2acbba2925553c9369d43341 | 27 November 2014, 00:01:05 UTC |
| cf6ff6f | YuviPanda | 26 November 2014, 23:53:12 UTC | ocg: Temp hack to bypass hiera for redis passwords Just keeping OCG up until _joe_ wakes up. Bug: T76111 Change-Id: I60dd4846be5e6ab7bb728ea8860dda430b0cc4c9 | 26 November 2014, 23:53:12 UTC |
| 44c570f | YuviPanda | 26 November 2014, 23:15:29 UTC | tools: Remove experimental mongo role/class This never panned out. Tools isn't webscale enough. Change-Id: I8076da4aa033bdc400eba428bbec4f9e93155032 | 26 November 2014, 23:18:10 UTC |
| d4822da | Gabriel Wicke | 26 November 2014, 22:13:20 UTC | Allow read access to fonts in OCG apparmor profile Change-Id: I1a90c278ebf59f303d1f3678dcd0593f8d4440e1 | 26 November 2014, 22:13:20 UTC |
| 3e50a27 | C. Scott Ananian | 26 November 2014, 21:24:16 UTC | Update apparmor profile for OCG. Co-Authored-By: Gabriel Wicke <gwicke@wikimedia.org> Change-Id: I13d76432d0c61b2af8568c076c5d7ba5a569262c | 26 November 2014, 21:48:08 UTC |
| e5a5d83 | Gabriel Wicke | 26 November 2014, 20:23:03 UTC | Fix apparmor config syntax The apparmor profile would not load due to syntax errors. This patch fixes those, so that the profile will at least be loaded. We are now testing it in labs to verify that the service actually working with this profile being enforced. Change-Id: I10b5766ab031b4ac7076a3d56e7fb0620f10f9b2 | 26 November 2014, 20:23:03 UTC |
| c62d87a | dzahn | 26 November 2014, 19:09:04 UTC | Merge "ci/Travis: ensure .ssh dir is present" into production | 26 November 2014, 19:09:04 UTC |
| 381dd85 | dzahn | 26 November 2014, 19:06:20 UTC | ci/Travis: ensure .ssh dir is present ensure the system user gets a .ssh dir in the home. fix the requires accordingly Change-Id: I705c2262c6542a4af55e6c469e3ef399c347a072 RT: 8866 | 26 November 2014, 19:06:20 UTC |
| d8d0d2e | YuviPanda | 26 November 2014, 09:43:25 UTC | shinken: Add checks for labs puppetmaster - Make all puppet failure checks a dependency on that, to prevent massive notices when there are transient puppetmaster failures - Hardcodes labs puppetmaster info, should probably use hiera for this somehow - Alerts go to labs-infra group, which is me and andrewbogott Change-Id: I208fd3a67c5c3064bfd6fa2757fee2a6d290f07f | 26 November 2014, 19:00:25 UTC |
| 92f72d3 | dzahn | 26 November 2014, 18:53:03 UTC | Merge "ci: travis user, typo 'nmptravis' vs. 'npmtravis'" into production | 26 November 2014, 18:53:03 UTC |
| 415cd47 | dzahn | 26 November 2014, 18:51:06 UTC | ci: travis user, typo 'nmptravis' vs. 'npmtravis' needs to be npmtravis, fix typo Change-Id: Ia2e8bc69466e91c9b152663939fd039185c359a8 | 26 November 2014, 18:51:06 UTC |
| 145ec26 | dzahn | 26 November 2014, 18:47:41 UTC | Merge "CI: install private ssh key for Travis integration" into production | 26 November 2014, 18:47:41 UTC |
| 08f352b | dzahn | 18 November 2014, 18:02:35 UTC | CI: install private ssh key for Travis integration setup user and install private key for Jenkins-Travis integration on hosts with role CI slave, requested in: Change-Id: I7eb3964f1be2ea419bb61f6851a3460360532772 RT: 8866 | 26 November 2014, 18:46:18 UTC |
| a1a4bff | Gabriel Wicke | 26 November 2014, 03:29:24 UTC | Move restbase config to regex.yaml This targets the restbase and cassandra config to the test hosts, and should work with the new hiera setup as per https://gerrit.wikimedia.org/r/#/c/174694/ and https://wikitech.wikimedia.org/wiki/Puppet_Hiera#Practical_example. Bug: T1228 Change-Id: I8d157b5757df898e16172ec88e657edf6c132a9c | 26 November 2014, 17:01:47 UTC |
| 34f8c2f | Giuseppe Lavagetto | 26 November 2014, 11:37:46 UTC | mediawiki: fix content-type and content-length in fcgi Change-Id: I232138e62d31ca734eb30ba2e34262a134a2d615 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 26 November 2014, 11:37:46 UTC |
| ce9ef73 | Faidon Liambotis | 24 November 2014, 22:21:24 UTC | geoip: kill geoliteupdate in favor of geoipupdate MaxMind's geoipupdate mechanism has a "hidden" feature that MaxMind themselves pointed me to: GeoLite databases have their own updates.maxmind.com product codes and there is a special UserID of 999999 with a LicenseKey of 000000000000 that has privileges to download them. Kill geoliteupdate in favor of using geoipupdate across the board. This brings us a similar update mechanism for production & Labs, plus a better program to fetch updates, as this one also does MD5 checks etc. Change-Id: I34fb5b2d5253a9161d3c86c2e92375049c241775 | 26 November 2014, 09:26:56 UTC |
| 885a1be | dzahn | 15 November 2014, 01:51:43 UTC | realm: remove pmtpa, add codfw 208.80.152.0/24 is codfw now and pmtpa's private address space is gone. Change-Id: If6c4dbb88704cd4aa336027bf2fee0daf4b55e3e | 26 November 2014, 09:17:29 UTC |
| 95e900e | Giuseppe Lavagetto | 26 November 2014, 07:07:15 UTC | mediawiki: fix passing of the Authorization header in HAT Change-Id: I56a59db10249e25ffd7fcc8904fefe6b2e91f69c Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 26 November 2014, 07:07:15 UTC |
| ef52839 | dzahn | 26 November 2014, 01:47:57 UTC | Merge "mha: replace pmtpa with codfw" into production | 26 November 2014, 01:47:57 UTC |
| c638049 | dzahn | 14 November 2014, 23:53:14 UTC | mha: replace pmtpa with codfw replace pmtpa with codfw in role/mha.pp, does this make sense or should it just be removed? Change-Id: I2b45e65d503c7866a83cb72428c1eb2bcb27a6eb | 26 November 2014, 01:16:26 UTC |
| beca8da | dzahn | 26 November 2014, 00:55:50 UTC | Merge "wikistats: add cron to enable wikia updates" into production | 26 November 2014, 00:55:50 UTC |
| 196bae8 | dzahn | 26 November 2014, 00:53:14 UTC | wikistats: add cron to enable wikia updates Bug: T61943 Change-Id: I26991a6dcec3c255b88a4eb5034621ab8900d176 | 26 November 2014, 00:55:02 UTC |
| 31656eb | Andre Klapper | 19 November 2014, 22:45:45 UTC | Phab: Change user visible strings "Execute Query" and "Real Name" Bug: T371 Bug: T798 Change-Id: I954c05c19369723b1b67c6c40aeac2fa3c546bbf | 25 November 2014, 22:46:27 UTC |
| 259f20d | dzahn | 17 November 2014, 22:28:31 UTC | delete class ldap::client::autofs per comment on I9fa24d47852ac2d3c91 can the whole class be removed then? Change-Id: I8d67f7ad7a0d8efe9cb6a97f8d168d6dd32b4e11 | 25 November 2014, 22:16:44 UTC |
| a3683b5 | dzahn | 25 November 2014, 22:13:00 UTC | Merge "delete class facilities::dc-cam-transcoder" into production | 25 November 2014, 22:13:00 UTC |
| e6d4bdf | John F. Lewis | 23 November 2014, 00:50:30 UTC | admin: grant qchris tin access (through deployers) Christian needs access to tin to do deployments of EventLogging software as per the RT ticket, manager approval has been sought and ticks the box. Grant access through deployers although adding 'eventlogging-roots' to tin is an option I guess. Change-Id: Ice7e173dd273587e34c0a74a726e3c12bf28f895 RT: 8914 | 25 November 2014, 22:01:16 UTC |
| 6336f06 | cmjohnson | 25 November 2014, 21:49:43 UTC | Merge "Adding bob west to data.yaml and to west1 to bastion only and analytics-privatedata-users RT8896" into production | 25 November 2014, 21:49:43 UTC |
| 19438b9 | dzahn | 25 November 2014, 21:15:15 UTC | Merge "ganglia: remove pmtpa varnish stanza" into production | 25 November 2014, 21:15:15 UTC |
| fc02600 | cmjohnson | 25 November 2014, 19:00:50 UTC | Adding bob west to data.yaml and to west1 to bastion only and analytics-privatedata-users RT8896 Change-Id: I860ed1cf06f15ecc097c422fa325f93d44cf8fcb | 25 November 2014, 19:25:58 UTC |
| 3fa175c | Gabriel Wicke | 25 November 2014, 01:43:54 UTC | Include and configure the restbase role on the test cluster Also set up the cassandra seeds in hiera, so that the cassandra cluster nodes can find each other & aren't as lonely any more. Since this is for internal testing only we leave out passwords & encryption for now. Default cassandra passwords will be used. Change-Id: I330a3b8487b647f0d7c22b28d823a5b4f858dd59 | 25 November 2014, 19:13:05 UTC |
| eb9c09c | Giuseppe Lavagetto | 25 November 2014, 17:11:45 UTC | mediawiki: move most servers from the hhvm to the standard pool We'll remove the last two later, to avoid paging on the hhvm lvs endpoints Change-Id: Idce006c44af7fdb7ac0955256790ca5407b88a8a Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 17:16:23 UTC |
| d3ef614 | YuviPanda | 25 November 2014, 17:08:22 UTC | shinken: I heard you like monitoring So let me monitor your monitoring project from your monitoring project Change-Id: Ie000d8a89d32734e1840084c3adb601ed217c57b | 25 November 2014, 17:08:22 UTC |
| b8fddf4 | YuviPanda | 25 November 2014, 17:03:25 UTC | icinga: Remove CPU alerts for contint from icinga With this all of labs alerting is gone from icinga. We'll add this back in better form as part of basic-checks for all hosts. Change-Id: I59c54f3bab8c00621041f55f877de26be7df4368 | 25 November 2014, 17:03:25 UTC |
| 150a77e | YuviPanda | 25 November 2014, 16:27:16 UTC | Merge "shinken: Cleanup to be better compatible with autoload layout" into production | 25 November 2014, 16:27:16 UTC |
| 1e4a5a6 | Tony Thomas | 25 November 2014, 16:26:59 UTC | Fix incorrect beta MX hostname Bug: T75736 Change-Id: I9081ce7eabfe1b25c6bd4e91a45615d4fab6920a | 25 November 2014, 16:26:59 UTC |
| a2c6e98 | Giuseppe Lavagetto | 24 November 2014, 13:13:28 UTC | varnish: remove redirection to the hhvm pool Change-Id: I39064328aabd1e7e5147e464e879a22e39c839fb Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 16:24:36 UTC |
| 9a9e40b | YuviPanda | 25 November 2014, 16:15:22 UTC | shinken: Cleanup to be better compatible with autoload layout Change-Id: If8d4d1af1f11b8e1cd39c1cee738f709d5bfdae4 | 25 November 2014, 16:22:15 UTC |
| fb34959 | Giuseppe Lavagetto | 25 November 2014, 16:09:04 UTC | swift_new: reorganize hieradata Change-Id: Ibddd82db898c16666feb2c58fb4c25daa32c38a1 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 16:09:56 UTC |
| 0be03ef | YuviPanda | 25 November 2014, 06:43:28 UTC | puppetmaster: Add script to track upstream changes This is currently in use by deployment-prep, and seems general enough for wider use as is. Adds a puppetvar that makes this usable. Note that before merging must munge the roles applied on deployment-salt. Bug: T1150 Change-Id: Ifa3229d6fb3424f75d14dd2569efca3b93936aeb | 25 November 2014, 15:55:18 UTC |
| 3433c68 | Giuseppe Lavagetto | 20 November 2014, 10:03:48 UTC | deployment: make scap proxies configured in one place * moved misc::deployment::scap_proxy to deployment::scap_proxy * made dsh/group/scap-proxies configurable via a class variable * use the same variable in mediawiki::common to determine if the server is a scap proxy This way we can finally change things in one place only. Change-Id: I22dd6d44a4574f679451a606fab7750e7345bb6c Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 15:32:52 UTC |
| 03c055f | Giuseppe Lavagetto | 20 November 2014, 14:44:16 UTC | hiera: a few tweaks * Make $site an expanded path as well * move most things out of equiad.yaml, and into common/ Change-Id: I86c00de8049e503d85f2a721d89feab594751cc6 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 15:07:28 UTC |
| 40d140f | YuviPanda | 25 November 2014, 12:29:22 UTC | shinken: Move alert notification to ml from icinga Change-Id: Ib41cc18c7722c663ef1f5843955d3dcda08ee0cc | 25 November 2014, 12:30:12 UTC |
| c911209 | YuviPanda | 25 November 2014, 12:02:38 UTC | contint: Move monitoring into shinken rather than icinga No CPU checks in shinken yet, so leaving those in icinga for now Change-Id: I9bcf176e39315d72a665d838faf0040e8de093f9 | 25 November 2014, 12:19:44 UTC |
| 14027a9 | Yuvipanda | 25 November 2014, 12:07:16 UTC | Merge "toollabs: lint" into production | 25 November 2014, 12:07:16 UTC |
| 76ba81a | Matanya Moses | 25 November 2014, 12:04:08 UTC | toollabs: lint Change-Id: If7d9eda0fef5ca8a961021bb1bd21f01b8aa7e93 | 25 November 2014, 12:06:03 UTC |
| f0d0a02 | Giuseppe Lavagetto | 24 November 2014, 11:23:43 UTC | mediawiki: adjust hhvm max threads to number of cpus as well It seems HHVM can't work well whenever its thread count gets very high compared to the available processors, so we also limit the max number of running threads based on that as well. This would result in: * 96 threads on smaller appservers (we had 100) * 128 threads on larger appservers (we had 536) Change-Id: I204c55c5bea378ccbf65a363cfc8ad29b29bea27 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 12:02:38 UTC |
| 6015e1a | Faidon Liambotis | 25 November 2014, 11:53:49 UTC | ocg: simplify module hierarchy There is no reason for "ocg::ganglia::module", "ocg::ganglia" should be enough. Similarly for Nagios, with the extra catch that there is no need for two separate classes, one of which is only used by the other and only contains a single resource. Change-Id: Ia5c16bfc20d7e9e10cd6860a29a06a65506335b2 | 25 November 2014, 11:54:58 UTC |
| 2a69cea | YuviPanda | 25 November 2014, 11:45:52 UTC | icinga: Remove contactgroups for toollabs & betalabs Change-Id: Iba86bded3a2ef0ad8ac94f60beb44c00790b9bcd | 25 November 2014, 11:46:44 UTC |
| 9f4c484 | YuviPanda | 25 November 2014, 11:38:38 UTC | icinga: Remove checks for betalabs and toollabs Superseeded by shinken Change-Id: I2115feb1ebbcdf66e0317a06cc72224300a7faf5 | 25 November 2014, 11:39:19 UTC |
| 8e1eac3 | Matanya Moses | 25 November 2014, 11:30:56 UTC | vm: lint Change-Id: I88768cece86ee450d1a32bc8b6b59b7a6c2e96d9 | 25 November 2014, 11:31:35 UTC |
| 9261db2 | Alexandros Kosiaris | 25 November 2014, 11:27:57 UTC | Merge "wikitech: lint" into production | 25 November 2014, 11:27:57 UTC |
| f54f465 | Matanya Moses | 25 November 2014, 11:18:36 UTC | wikitech: lint Change-Id: Ia7ff8f1792d6871b1d5aa5b62092340ea3741e44 | 25 November 2014, 11:18:36 UTC |
| 5c8d22e | Giuseppe Lavagetto | 25 November 2014, 11:01:38 UTC | dsh: add new mediawiki appservers Change-Id: I961eb9535cca34ff1eab1821319257216c249dc6 Signed-off-by: Giuseppe Lavagetto <glavagetto@wikimedia.org> | 25 November 2014, 11:01:38 UTC |
| 436f92f | Sean Pringle | 25 November 2014, 02:59:20 UTC | MariaDB config tweakes for m1,m2,m3 - Single buffer pool instance (>1 contributed to m2 lockup bug) - Enable extra port 3307 in line with prod cnf, for emergencies Change-Id: I43301d8cb5e77071629df806fb99e267b756787d | 25 November 2014, 02:59:20 UTC |
| a2fc778 | dzahn | 25 November 2014, 02:18:02 UTC | Merge "bugzilla: old-bz, keep enforcing https" into production | 25 November 2014, 02:18:02 UTC |
| e029592 | dzahn | 25 November 2014, 02:15:45 UTC | bugzilla: old-bz, keep enforcing https we want to keep enforcing http->https but we need a different rule now that we are behind misc-web, in the main vhost Change-Id: I92de40b20e362c9239c4a78ef147699ca221bf24 | 25 November 2014, 02:15:45 UTC |
| d039a00 | C. Scott Ananian | 12 November 2014, 18:40:52 UTC | Give parsoid-admins access to ruthenium ruthenium is now used for parsoid round-trip testing, so grant access to parsoid-roots & admins. RT: 6980 Change-Id: I974e3a80155caa27f34ef1c5d711a3817367e7c1 | 25 November 2014, 01:27:43 UTC |
| 448c61c | Faidon Liambotis | 25 November 2014, 01:17:41 UTC | Switch Cassandra test hosts to the new role class - Move system::role under the role class. - Get rid of the vm.max_map_count sysctl, included in Cassandra's Debian package now. - Split ruthenium off the host list, to be repurposed. Change-Id: I3e5a4b95a906fd9797a731cafc80d8d94ca45db9 | 25 November 2014, 01:25:38 UTC |
| 880e1ae | Brandon Black | 24 November 2014, 16:58:00 UTC | Switch LVS to use localssl at all sites Change-Id: Ib791d78a28a26a49d729d0082ba51cfe8674a98a | 25 November 2014, 01:19:49 UTC |
| 880d3e7 | dzahn | 25 November 2014, 00:52:15 UTC | Merge "bugzilla: remove SSL Apache config for old-bz" into production | 25 November 2014, 00:52:15 UTC |
| 65dcc69 | dzahn | 25 November 2014, 00:29:45 UTC | bugzilla: remove SSL Apache config for old-bz to put behind misc-web varnish we need to adjust the Apache config on the backend. remove SSL part and change RewriteCond to be based on X-Forwarded-Proto keep bug-attachment unchanged Change-Id: I3ce54f7d59efade97f0449f5f13380bab9d3cb96 | 25 November 2014, 00:29:45 UTC |
| 957b8d8 | Brandon Black | 24 November 2014, 16:37:05 UTC | Turn on r::c::ssl::sni locally for varnishes This switches the ulsfo caches from r::c::ssl::unified to ::sni for actual prod traffic flow. For eqiad/esams, it configures the ::sni -style local nginx service on the cache hosts themselves, but LVS will still be sending the traffic to the ssl[13]00x machines at these datacenters instead of the new local ssl services until further changes are merged. Change-Id: I24013da78641970733649749b6dd2c5eaf507d8e | 25 November 2014, 00:20:25 UTC |
| 4e0d43e | Brandon Black | 24 November 2014, 15:17:37 UTC | r::c::ssl::misc: switch to r::c::localssl like prod SNI Change-Id: Ia261825e50d6473e6ee78bd2af38ad8774863e9a | 25 November 2014, 00:08:01 UTC |
| 83c740b | dzahn | 24 November 2014, 23:20:16 UTC | move old-bugzilla behind misc varnish put old-bugzilla.wikimedia.org behind misc-web varnish. to fix the current SSL cert warning. disabled caching, just send straight to zirconium as the backend which is the original Bugzilla box Change-Id: I17a9a34f361fe71fa46322063395d520fe66f6c7 | 24 November 2014, 23:30:28 UTC |
| c8a6046 | dzahn | 24 November 2014, 23:02:55 UTC | Merge "misc varnish: do not handle bz-attachment URLs" into production | 24 November 2014, 23:02:55 UTC |
| 8bbe109 | Andrew Otto | 24 November 2014, 22:57:57 UTC | Update varnishkafka module with logrotate changes Change-Id: Idc59675fe84a64df48f0ec65a93ac0b8191ef786 | 24 November 2014, 22:57:57 UTC |
| 1d70884 | Faidon Liambotis | 24 November 2014, 22:54:35 UTC | Merge "bits varnish: serve 204s for /statsv" into production | 24 November 2014, 22:54:35 UTC |
| 75f1da2 | Ori Livneh | 24 November 2014, 22:37:18 UTC | bits varnish: serve 204s for /statsv statsv is a logging beacon endpoint for processing client-side statsd-like performance metrics, generated by MediaWiki's JavaScript code. * Make //bits.wikimedia.org/statsv/* reqs HTTP 204s, like EventLogging's /event.gif endpoint. * Fix the RxURL arg for the Varnishkafka instance to not require an additional slash. * Add a comment explaining both endpoints. Change-Id: I6774c0b6508879226dfb41fb123c9ac4dfe132d3 | 24 November 2014, 22:46:26 UTC |
