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
26ffa63 Brandon Black 12 April 2017, 22:26:50 UTC Revert "LVS: remove direct use of acamar recdns" This reverts commit 94132b1b53a153b7f07d98315aa667b396e42362. Bug: T155411 Change-Id: I8ad17b25ce9a1734de5f6c8251dfbcb3657849f0 12 April 2017, 22:29:34 UTC
1f0e598 Daniel Zahn 12 April 2017, 16:28:20 UTC site: remove all the "$cluster = 'misc'" There is a a "if undef then misc" statement in site.pp anyways, so all of these are not really needed. And it should be only for Ganglia, which we are deprecating. Change-Id: I785ba217353d1019c0379d0af9841761903e0764 12 April 2017, 22:20:43 UTC
6363d76 Daniel Zahn 12 April 2017, 16:22:53 UTC site/install: unify install nodes into a single regex Change-Id: I1fc19e11840ef1b5c2245f9d15846377e03ec60b 12 April 2017, 22:07:56 UTC
b12bcc3 Daniel Zahn 12 April 2017, 17:11:43 UTC site/redis_slave: remove duplicate base::firewall include This is already included in the role and the other nodes with the same role in the other DC also don't have it. Change-Id: I3a4065aa18a93562f0f7eeb6b10bd9226aed694b 12 April 2017, 21:50:03 UTC
f0c2ee3 andrewbogott 12 April 2017, 20:54:11 UTC wmfkeystonehooks: Fixup for project creation Change-Id: Ifa98d6571a1efd2237c989d9efbf1a121f6fa103 12 April 2017, 20:54:30 UTC
ba2903f Daniel Zahn 12 April 2017, 16:40:15 UTC site/ganglia: move firewall/standard/IPv6 to role Change-Id: I77a803eb7e3e4918d5c2b8ea9d306de444ef7562 12 April 2017, 20:45:10 UTC
1e6be51 Daniel Zahn 12 April 2017, 16:35:10 UTC site/thumbor: remove duplicate base::firewall include This is already included in the role thumbor::mediawiki itself. Change-Id: Ib1fe4e7b3c3c3178e9d68f0408be4566d32dd369 12 April 2017, 20:43:05 UTC
560b8a5 Brandon Black 12 April 2017, 13:16:04 UTC acamar/achernar -> jessie Also, move to the same partman layout as other recdns (simple raid1, not LVM-based) Bug: T155411 Change-Id: I68ccf1525da89ddfe88ee57b839bc49739763921 12 April 2017, 20:23:42 UTC
29a3dbf Brandon Black 12 April 2017, 20:20:04 UTC LVS: remove direct use of acamar recdns Will revert after acamar's reinstall Bug: T155411 Change-Id: Ia0e9056f80cf8f4430f6c5441a30613a76729562 12 April 2017, 20:21:42 UTC
22d26aa andrewbogott 05 April 2017, 18:33:30 UTC wmfkeystonehooks: Create project page on wikitech on project creation And delete it on project deletion! Bug: T150091 Change-Id: I3a42e3eb117da789a9f960d5b495cdd75fdc00fe 12 April 2017, 20:16:49 UTC
16dae2c Daniel Zahn 12 April 2017, 17:52:28 UTC planet/varnish-misc: switch planet to active-active The planet service can be active/active. There is an existing planet2001 in codfw already setup and running just that it was active/passive so far. So for the datacenter switch, and in general, add the second backend in codfw in misc varnish config. And in planet itself remove the "active_dc" setting which made one of them passive and run the update cron jobs in both. Bug:T88761 Bug:T156937 Change-Id: Ifbfe4b3ddfb5de259be64a9ea02067c66f24f122 12 April 2017, 19:58:52 UTC
2e11054 andrewbogott 12 April 2017, 17:15:35 UTC Repool labvirt1002. Fixed and the cause of the problem is now monitored. Bug: T162640 Change-Id: Ibdc19b54ec2eef3ca63b28f98af5bcd1702c7304 12 April 2017, 17:17:47 UTC
2cf544b Alexandros Kosiaris 12 April 2017, 14:03:54 UTC openstack: Decrease by 2 the backups held Instead of keeping 5 (actually 6 for 23/24 hours) keep 3 (actually 4) backup files. This has the drawback of making a tad more difficult to restore data from more earlier points in time as we would have to go through the regular backup system to get these files but is expected to save some disk space as well as some space on the backup system. The latter is estimated roughly at 80GB. Bug: T159524 Change-Id: I62466d356a0249fe6b65fba3a1e32b937f1a7bc7 12 April 2017, 16:51:41 UTC
131b072 andrewbogott 06 April 2017, 16:45:40 UTC nova.conf: change dhcp lease times to 12 hours. This is a followup for 222b8f856e04500b20a94e2421d87e028113e3d9. Three minutes was dangerous because it might cause outages at the drop of a hat, but 12 hours should give us a full day to mess with nova-network without causing an outage, and may well still reduce the number of leaks that we run into. Bug: T160908 Change-Id: Ie136dcecf998dca5abd4cc5db79afe8a1ce29cde 12 April 2017, 16:29:55 UTC
d2e0eee andrewbogott 06 April 2017, 16:25:26 UTC nova.conf: Reduce fixed_ip_disassociate_timeout to three minutes. This is a bit of a stab in the dark -- I'm hoping that this will encourage nova (and by extension, dnsmasq?) to clean up after itself a bit better and leak fewer IPs. Bug: T160908 Change-Id: I30e6e44c306d84790adbbe2ab54fc2d425c14895 12 April 2017, 16:28:09 UTC
7c47872 andrewbogott 08 April 2017, 20:07:09 UTC nova-compute monitoring: Check for one and only one nova-compute process Bug: T162640 Change-Id: I7b6c4b5f1ceb9bc0f3e2e88a1aac4185423e756f 12 April 2017, 16:23:04 UTC
8af8c35 Andrew Otto 12 April 2017, 15:16:03 UTC Remove unused ganglia views Change-Id: Ic8c71287af134e0789b6d93346cdf4cdf26882e2 12 April 2017, 15:17:34 UTC
94561a6 andrewbogott 12 April 2017, 15:03:06 UTC maintain-meta_p: Add a missing quote-mark Change-Id: Iff441aec8b509d3ed55cb47de21f713be5f4603f 12 April 2017, 15:11:53 UTC
16f1cb4 Riccardo Coccioli 12 April 2017, 09:12:15 UTC Switchdc: rename redis stage from t05 to t06 Bug: T160178 Change-Id: I3e43cdb2d089e7ba436d9bc0218a9ff754b4f7be 12 April 2017, 14:58:42 UTC
c1b6124 Emanuele Rocca 12 April 2017, 13:50:48 UTC Blacklist intel_uncore kernel module The intel_uncore kernel module provides performance monitoring support for NehalemEX and more modern processors. The code has been made modular between Linux 4.4 and Linux 4.9, and the module has been causing troubles bringing up network interfaces when upgrading some of the cache hosts to 4.9. Blacklist it. Bug: T162612 Change-Id: I37838eba628af41747387683dc64658917061284 12 April 2017, 14:28:13 UTC
971f3ff Brandon Black 30 March 2017, 13:03:23 UTC citoid: active/active public interface This will send live user traffic to the application simultaneously in both primary DCs! Change-Id: Ieb29299ae97fd88424dd06ac2add19fbac46ab7b 12 April 2017, 14:14:28 UTC
6f1a041 Brandon Black 30 March 2017, 13:03:10 UTC cxserver: active/active public interface This will send live user traffic to the application simultaneously in both primary DCs! Change-Id: I5a0aa2f531a8bf35a14adbae7eb66d95b5488268 12 April 2017, 14:07:37 UTC
bc3edb4 Alexandros Kosiaris 12 April 2017, 13:52:21 UTC Disable backups on labtestweb2001 There is no reason for this host to get backed up per T159524, so use the recently added profile::backup::host::enable flag to disable backups for it. That is expected to save us some 900GB Bug: T159524 Change-Id: I065e429b39b6ab8c07fd4f8ce1cc4a371858601d 12 April 2017, 13:52:21 UTC
5042098 Riccardo Coccioli 12 April 2017, 10:36:06 UTC cache: noop to test the switchdc procedures Bug: T160178 Change-Id: I31a982b76a335b88bca9181fbf78ef628dc1399c 12 April 2017, 13:39:20 UTC
d3b8067 Alexandros Kosiaris 08 March 2017, 17:08:45 UTC backup::set: Allow passing an $extras config hash Allow backup::set to accept a config Hash that is pass directly to bacula::client::job. That does expose a bit of bacula's Job Resource configuration to the caller but it allows for more flexiblity. Default to under to maintain compatibility. Amend openldapset to use the backup::set now instead Change-Id: I200662d81df7a733e7a4a63ffb3d146212dab51b 12 April 2017, 13:31:26 UTC
1334d10 Alexandros Kosiaris 08 March 2017, 17:22:14 UTC Move backup::openldapset into the respective roles It's place is clearly not in site.pp, so move the calls into the respective roles. Had to fix a bit the memory check for openldap so as to adhere to the 140 characters long line rule Change-Id: Ib3bf7d425fa617c1de8edb30b799ade42c55f08e 12 April 2017, 13:28:59 UTC
f6cf327 Brandon Black 11 April 2017, 18:11:53 UTC ulsfo recdns: prefer codfw to eqiad This is similar in nature to 392aa4fc for the LVS nameserver overrides. For now (ulsfo lacking local recdns), ulsfo should be prefering to use the codfw resolvers to the eqiad ones, as they're closer. We probably forgot to make this change long ago when codfw stabilized. Change-Id: I02726ae1f0af240fc13d1167e22862b882116473 12 April 2017, 13:05:27 UTC
5a54a52 Alexandros Kosiaris 08 March 2017, 16:36:16 UTC Change bacula retention policies and volume number The retention policies are a threshold below which it is guaranteed no file/job/volume will be deleted from the database or the storage daemon. We 've started with this being 6 months but had to lower it as we would end up with our backup schedule being paused for long periods of time as the guaranteed space would equal occasionally the available space. While we will upgrade the infrastructure to add more space, lower the guarantee low enough to make sure we don't end up with the same problem in the near future. The reimage to jessie of helium allowed restructuring the filesystem structure freeing up some free space. That allows us to increase the number of volumes per pool somewhat. Bump it to 60 After we get more space, we can proceed with more tuning Change-Id: I219ba9aa34df625465e5791f5782eb55d5a4b89d 12 April 2017, 12:13:07 UTC
ae92e95 Giuseppe Lavagetto 12 April 2017, 11:58:42 UTC profile::conftool::client: accept protocol as a parameter Change-Id: Id00e7a8f9197a7734bbb4b4f5e7fa731bf77c80a 12 April 2017, 11:58:42 UTC
830a038 Giuseppe Lavagetto 12 April 2017, 11:25:25 UTC role::puppet_compiler: fix etcd protocol Change-Id: Ie2d1b0a09b85537e98778c619b5e091ee1eef08c 12 April 2017, 11:25:25 UTC
de72c36 Alexandros Kosiaris 12 April 2017, 10:21:09 UTC puppetmaster: Remove the jessie-backports pinning jessie-backports currently has 4.8 puppet which we do not want to install. Thankfully the install could not succeed due to a Requirement of facter >= 2.4 which due to the way apt pref downprefed jessie-backports for the facter package made the puppet packages from jessie-backports non-installable. Remove the pin to jessie-backports since that version would break our installation Bug: T162462 Change-Id: I755d2200937664d30d99f114eda4248c7958b252 12 April 2017, 10:46:13 UTC
9e53c96 Giuseppe Lavagetto 12 April 2017, 10:36:25 UTC wmflib: add some information to failures of conftool Change-Id: I7665f7458f7d982fabaef46600c484da163d5145 12 April 2017, 10:36:25 UTC
a91d360 joal 12 April 2017, 09:41:00 UTC Update pivot config template file Mediawiki history dataset is now updated regularly, this patch reflects the change in pivot documentation. Change-Id: Id8866eeffede7bced09630280f64f6dfa38bdb11 12 April 2017, 09:47:26 UTC
76aca96 elukey 12 April 2017, 08:17:13 UTC Set Xms settings for Hadoop Workers daemons This will allow to have Xms and Xmx settings for Hadoop Hdfs datanode and Yarn nodemanager daemons. Also upgrading the last cdh submodule sha that carries a change that is needed. Change-Id: I38b048c957efbd3ae6c331e3b06ee91f7e1c3b1d Bug: T159219 12 April 2017, 08:49:25 UTC
acf2c67 Marko Obrovac 11 April 2017, 19:57:35 UTC RESTBase: Clean-up unused variables With the switch to Scap3 code and config deploys, there are now certain variables in the restbase module that are not needed any more. This commit removes them and cleans up the module a bit. Change-Id: Ia10ee1cacbe69181ca6b2307a4cd6fbfcfbe5673 12 April 2017, 08:32:35 UTC
37271d4 elukey 12 April 2017, 08:00:23 UTC netboot.cfg: set correct priority to the videoscaler's recipe Videoscalers need to run without the /tmp partition to avoid ffmpeg tmp files to fill it up. The name match for videoscalers needs to happen before the rest of the mw* to be effective. Change-Id: I26a18d2751c2bd08e835a68b72d5c218a7d7eb96 12 April 2017, 08:02:20 UTC
f050a28 elukey 12 April 2017, 07:12:16 UTC Set Trusty as PXE image for mw2152 (videoscaler) Change-Id: I0f966a3ff2020ea9048d42ca60703c56b6b2a66b 12 April 2017, 07:12:16 UTC
0ebc4d7 elukey 12 April 2017, 06:26:05 UTC Set the correct partman recipe for mw2246 and mw2152 The videoscalers need to avoid a separate and small tmp partition since ffmpeg temporary files can easily fill it up. Change-Id: I311f60d4b0b2c0051bce8c1baf32b28718bfa0a3 12 April 2017, 06:27:49 UTC
7aa41fe Daniel Zahn 07 April 2017, 00:35:42 UTC mw_rc_irc: convert to profile/role structure Change-Id: I28e481efaa6b50ad2c57ea32dc03f6a1579c33f6 11 April 2017, 23:33:26 UTC
b3e82c9 Daniel Zahn 11 April 2017, 22:22:15 UTC ocg: enable ocg1003, disable ocg1001 ocg1001 needs to be disabled for hardware repair (T161158). The docs say to do this to temp. disable one of the hosts for cases like this. ocg1003 on the other hand was once reinstalled and disabled but never re-enabled again. The service is running there though. So re-enable that while 1001 is down (and use it anyways, separate from this). Bug: T84723 Bug: T161158 Change-Id: Iaf6a9c956fec0050ef1409b7b4dd433bf75c5d64 11 April 2017, 22:27:29 UTC
baf055b Daniel Zahn 07 April 2017, 00:19:05 UTC lists: convert to role/profile structure Change-Id: Ifdffd277d091b31f500084dfb3037e159b2b5ddd 11 April 2017, 21:07:00 UTC
8d4626e Andrew Bogott 11 April 2017, 20:07:06 UTC Revert "slapd conf: Allow for unlimited paged searches" /etc/ldap/slapd.conf : line 54: unable to add limit in "limits <pattern> <limits>" line. This reverts commit 8d39289f8bed18daacbe05539d9b82c191b9a4fd. Change-Id: I6148375bcb7187188aa5426be24db32befbab08d 11 April 2017, 20:07:06 UTC
8d39289 andrewbogott 06 April 2017, 03:02:15 UTC slapd conf: Allow for unlimited paged searches Change-Id: Icbcc60c46df7ac308f4fedc7fbf5fc6789bde300 11 April 2017, 20:01:36 UTC
fd62091 Alexandros Kosiaris 31 March 2017, 11:29:14 UTC changeprop: Add an ores_uris parameter Per T159615, one approach to solving the issue of having ORES's cache warmed up in both datacenters, is to duplicate changeprop's requests to both ORES clusters. In order to facilitate that, add an ores_uris parameter that is meant to be a list of the URLs changeprop should contact ORES clusters. For now we don't remove the ores_uri parameter on purpose to make the migration easier, since the config is tightly coupled with the jinja template generating the final configuration on the host by scap. Should be removed in a followup commit Bug: T159615 Change-Id: I3282c00754da7cff4e8c62359806a110f67e9921 11 April 2017, 17:16:40 UTC
9187994 Andrew Otto 11 April 2017, 17:01:33 UTC Include ores::base on hadoop clients and workers This lets us use ORES code in Hadoop. Bug: T162706 Change-Id: I1071e989ea8ba530d984570551d881e710ea2b33 11 April 2017, 17:03:38 UTC
4d244a3 Marko Obrovac 11 April 2017, 16:53:56 UTC RESTBase: Use the provided logging name and statsd alternative prefixes Bug: T116335 Change-Id: Icc981765aa18e322705886c097573cb81fb5e1f6 11 April 2017, 16:58:50 UTC
13d7507 Gabriel Wicke 08 March 2017, 19:17:38 UTC PDFRender: Delay service shut-down to work around xpra race Bug: T159922 Change-Id: I0a73a44dbd48c615efd63f1db0ac5da1b5f56e3c 11 April 2017, 16:49:34 UTC
724ca67 Paladox 10 April 2017, 22:05:34 UTC ircecho: Convert to base::service class to maintain the script Also migrate to systemd at the same time. Change-Id: I5a1e899bec6e0cea34256917d9818ca35017a9f2 11 April 2017, 16:49:20 UTC
d33240a Alexandros Kosiaris 11 April 2017, 13:44:15 UTC Shortcircuit profile::backup::host to be usable always Provide a profile::backup::enable boolean flag. Rely on it being set to true to actually do anything. Since we can not provide meaningful default values for any of the other hiera calls, but still want to use the profile class in environments where backup is not set up, we are force to do that simple hack Change-Id: I52936e5bed637098bfc20c8cd096587b0623b76e 11 April 2017, 16:47:05 UTC
b7d9be7 Marko Obrovac 10 April 2017, 20:20:07 UTC RESTBase: Switch to Scap3 config deploys Bug: T116335 Depends-On: Ie3ea20b2e1fd7b627288484d4a8ff91201872e1a Change-Id: I88a7bb4a0bb686cf78979dfca7a7da0a50c82c73 11 April 2017, 16:28:56 UTC
4f1b3f0 Filippo Giunchedi 11 April 2017, 16:16:59 UTC Decommission prometheus[12]00[12] Ganeti VMs, unused now. Bug: T162712 Change-Id: I5a6b6b43598dc716ee7276767da7bd041cea690a 11 April 2017, 16:16:59 UTC
15e4d31 Filippo Giunchedi 11 April 2017, 15:41:38 UTC hieradata: allow codfw prometheus to talk to netmon eqiad Bug: T148541 Change-Id: I4521b3ddf7e3b47e2f22f779e82081201dafef2e 11 April 2017, 15:46:53 UTC
7a5a6ca Tyler Cipriani 05 April 2017, 17:50:27 UTC Scap: update version to 3.5.5-1 Bug: T127762 Change-Id: Idb765b8492348957562d3030f97cc0da3e79e83c 11 April 2017, 15:19:36 UTC
c67a10c Giuseppe Lavagetto 11 April 2017, 14:05:02 UTC role::puppet_compiler: fix conftool config Change-Id: I2cb14504d155a9d4b90a058c3dfe035633abd0be 11 April 2017, 14:41:12 UTC
59a74d9 Jaime Crespo 05 April 2017, 17:53:30 UTC Indicate install recipes for newest db1* and db2* DB servers Servers to be installed for the first time must use the db.cfg recipe, instead of the one for already installed servers. Bug: T162159 Bug: T162233 Change-Id: Ied766f9b2ecf9bd79231c51c27ac6b7c8140c407 11 April 2017, 14:09:31 UTC
dddf28f Andrew Otto 11 April 2017, 13:33:58 UTC Install jessie by default on stat1004 Change-Id: I851dbc72f3758a53d6a8064294e05ed3b81bb8a7 11 April 2017, 13:49:54 UTC
f83d1ae Alexandros Kosiaris 10 April 2017, 14:33:12 UTC Make backup::set effectively a virtual resource Check if profile::backup::host has been defined and only then proceed with the rest of backup::set. This effectively makes backup::set a virtual resource that needs to be realized by including profile::backup::host on the node. This is driven by the need to have $jobdefaults present in backup::set and derived from the profile. The approach of making this a parameter was misguided since it turns out no callers ever declare it, nor want to Change-Id: I3aefc0800b74b8023017f3fbc3fef6a7c81bbff4 11 April 2017, 13:40:27 UTC
c93b322 Ottomata 11 April 2017, 13:27:03 UTC Revert "Move hadooop python package declaration into client" stat100[24] are trusty, and don't have these packages. This reverts commit f72ae307f496e7742cd6aef433be65a14e6e5c3d. Change-Id: I7697b7bc6b027f1b7aa793082457cf0dfde8036b 11 April 2017, 13:27:33 UTC
f72ae30 Andrew Otto 11 April 2017, 13:19:59 UTC Move hadooop python package declaration into client This will install packages on workers as well as clients (like stat100[24]). Change-Id: Ib810b3247828f8400ac9bfb06a68c2a01a21c6f8 11 April 2017, 13:20:47 UTC
f1ac395 elukey 11 April 2017, 12:44:18 UTC Set Trusty for mw2246 default PXE OS This host is currently a Mediawiki videoscaler running Debian, but it should run Trusty instead. Change-Id: I3c7b9240b3455ef6893b131a09c5554d24c9c9db 11 April 2017, 12:44:18 UTC
c68c43e BBlack 11 April 2017, 02:55:00 UTC Revert "dnsrecursor: update to backports for transition" This reverts commit c4881821c4eaf92a064c5132f090183f4d3d7937. Change-Id: I85dfb0655d2ccd555516538d09763f6dd8583e8a 11 April 2017, 02:55:04 UTC
4b9475d Brandon Black 11 April 2017, 02:52:37 UTC Revert "esams lvs: do not directly use maerlant, temporarily" This reverts commit 1bd02f40759f9c3754a14bfbc2077259be9625f1. Change-Id: If786479d29c812a24218fe067a74162b562ffa8d 11 April 2017, 02:52:54 UTC
1bd02f4 Brandon Black 11 April 2017, 02:49:15 UTC esams lvs: do not directly use maerlant, temporarily Change-Id: I29d2060dd5008837466d0b23baa308a091f21707 11 April 2017, 02:49:15 UTC
4b955ba Brandon Black 11 April 2017, 02:39:03 UTC Revert "eqiad lvs: do not directly use chromium, temporarily" This reverts commit 54438d40c2a42b018b1d94e15b753b70d7c5f1d7. Change-Id: I5218be7b6f43f4b8a061c80807e80590f427994a 11 April 2017, 02:39:11 UTC
54438d4 Brandon Black 11 April 2017, 02:28:46 UTC eqiad lvs: do not directly use chromium, temporarily Change-Id: I0375a6367d2fd39619fe5c4458576810d417e761 11 April 2017, 02:28:46 UTC
bcfc9e8 BBlack 11 April 2017, 02:15:06 UTC Revert "esams lvs: do not directly use nescio, temporarily" This reverts commit 3c028b4a955b35b2f778c0f0a0d68f56e4620669. Change-Id: I84c6f4f1bae23bc4c337828cad48069312a9495c 11 April 2017, 02:15:06 UTC
3c028b4 Brandon Black 11 April 2017, 02:10:10 UTC esams lvs: do not directly use nescio, temporarily To revert after upgrade complete Change-Id: I1007043917bbcda0d2ec3fb9994709eff33ecdfe 11 April 2017, 02:10:20 UTC
22e2235 Brandon Black 10 April 2017, 23:35:51 UTC Revert "eqiad lvs: do not directly use hydrogen temporarily" This reverts commit f80b9b7d97f044e3378151c2f51c26ff68557dbc. Change-Id: I46ebeb4827c38fdaa6d709d3a2b09fdab4d256c5 10 April 2017, 23:36:25 UTC
f80b9b7 Brandon Black 10 April 2017, 23:09:11 UTC eqiad lvs: do not directly use hydrogen temporarily hydrogen's recdns is being upgraded, will revert this afterwards. Change-Id: I1b60363e79370ab90685410c2d87030bcf5934a4 10 April 2017, 23:09:15 UTC
392aa4f Brandon Black 10 April 2017, 23:08:15 UTC ulsfo lvs: prefer codfw dns This should have been this way to begin with, as they're lower latency to ulsfo. We probably forgot to switch this after codfw initial stabilization. Change-Id: I94d235b945781d9f53ec4d460ba1143e3c98c8ce 10 April 2017, 23:08:18 UTC
c488182 Brandon Black 07 April 2017, 13:09:23 UTC dnsrecursor: update to backports for transition This is temporary to ensure smooth rollout on existing jessie recdns machines. Should revert after all are upgraded via puppet. Change-Id: Id1876231c33e644f6f3f645c085574bf490907b2 10 April 2017, 22:28:51 UTC
b2eb1b9 Brandon Black 07 April 2017, 03:05:28 UTC dnsrecursor: 4.x backport and edns-client-subnet I've manually tested the backports package with these settings, and it functions correctly as expected (including correct discovery geoip effects for private-network clients via edns-client-subnet). Change-Id: I73cfea9e56800624f1353a381540c0f410dd826d 10 April 2017, 22:28:51 UTC
bf45a72 andrewbogott 08 April 2017, 14:49:16 UTC keystone.conf: Whitespace cleanups Change-Id: I1c6bdbfdae3453c71e1777830a2db70262db39cb 08 April 2017, 14:49:16 UTC
c7dc72c andrewbogott 06 April 2017, 07:23:41 UTC Keystonehooks: Add two more ldap ous for sudo handling. These are pointless stubs, but apparently we need them. Change-Id: I62b2300b6fbf0b806755bcd6f0e5ee931f1a0aa7 08 April 2017, 14:25:12 UTC
9d69ad1 andrewbogott 06 April 2017, 10:46:40 UTC fullstack: Switch to the normal jessie image During testing I've been building a bunch of different images and naming them 'debian-jessie-fullstack-test' to keep them out of the way of users. Now that that has settled down, I've built a new public image with all of the new tweaks. Fullstack should use this image too, so it's actually testing the same thing as what users get. Change-Id: Ib172a754c7a0aa6258458ba84728e701aa39500e 08 April 2017, 14:15:30 UTC
0810a02 andrewbogott 08 April 2017, 14:15:08 UTC Remove the striker-users admin group Standardize on wmcs-roots which also includes Bryan. Change-Id: Id8010806b2f211ce5f8412b5da2e7470920860c6 08 April 2017, 14:15:30 UTC
522c079 Antoine Musso 10 April 2017, 19:47:50 UTC jenkins: also fix permissions for access.log Previously failed with a dependency cycle: File[/var/log/jenkins/access.log] => Systemd::Syslog[jenkins] => File[/var/log/jenkins] => File[/var/log/jenkins/access.log] 350837fd75 used: before => Systemd::Syslog['jenkins'] Change it to before => Service['jenkins'] Change-Id: I24daeb5b91c3522d580a0695b68563722a0fc7db 10 April 2017, 20:25:31 UTC
467d70a Andrew Otto 10 April 2017, 20:15:52 UTC Add new hadoop workers to hadoop/net-topology.py.erb Bug: T152713 Change-Id: Id65cb747b5775ad5d94f0f880ff674daea5bddcf 10 April 2017, 20:20:16 UTC
f554af2 Erik Bernhardson 16 March 2017, 00:15:31 UTC Upgrade logstash to 5.x Logstash 5.x is compatible with elasticsearch 2.x through 5.x, whereas the currently deployed version of logstash is not compatible with elasticsearch 5.x. This should roll out prior to upgrading elasticsearch in the logging cluster to 5.x. * JVM config is now set via /etc/logstash/jvm.options, rather than /etc/default/logstash. Provided config is the same as upstream with one update to specify our custom heap sizes. * Depends on updated operations/software/logstash/plugins repo to provide a 'plugin pack'. * The logstash/plugins trebuchet repository is now git-fat enabled to allow deployment of the new plugin pack. * renamed logstash::heap_memory_mb to logstash::heap_memory to be consistent with other java modules. * renamed logstash::filter_works to logstash::pipeline_works to match upstream rename * the kafka input plugin now takes a list of kafka brokers, rather than a zookeeper host, to bootstrap itself with. * Made all the logstash config files 0444 instead of 0440. There is nothing secret in them and it makes debugging more of a pain by requiring sudo. Bug: T154473 Change-Id: Ica36805c6fb09a33e9cae6788711d4ed3d7beb9b 10 April 2017, 20:07:39 UTC
a193ed8 Antoine Musso 10 March 2017, 14:09:06 UTC jenkins: tweak log permissions The Jenkins logs are not readable by others. Make them to belong to the 'wikidev' group so at least mortals can get access to it. That redo 350837fd75 which has been reverted by 2bc1ab8666. The reason is I defined a File[/var/log/jenkins/access.log] that is already defined by systemd::syslog resulting in a cycle: File[/var/log/jenkins/access.log] => Systemd::Syslog[jenkins] => File[/var/log/jenkins] => File[/var/log/jenkins/access.log] Change-Id: I2949c4a5d502c8e434d790fcbd66cb5e8282cb7d 10 April 2017, 19:35:22 UTC
0e291c5 Daniel Zahn 07 April 2017, 20:26:18 UTC typos: make "include profile::backup" a typo Make "include profile::backup" a typo after Change-Id: Iaa7a5c8b661c0f6 has standardized on "include ::profile::backup" so that jenkins-bot will downvote the variant without leading ::. Change-Id: I220aac18603ee1fa98e6f485394776b28ccc0a39 10 April 2017, 19:27:46 UTC
28ef2eb Daniel Zahn 07 April 2017, 17:29:48 UTC standardize "include ::base:*" ~/puppet$ grep -r "include base" * | wc -l 6 ~/puppet$ grep -r "include ::base" * | wc -l 203 find . -name "*.pp" -print | xargs sed -i 's/include base/include ::base/g' Change-Id: I2981864ebab5a4a5756e2efcf5dc8beb9a712f35 10 April 2017, 19:03:02 UTC
a4fdcf5 cpettet 10 April 2017, 15:37:47 UTC admin: add bd808 to wmcs-roots Bug: T162404 Change-Id: If5be328395ab070a94d3097384d134cbb944ba0f 10 April 2017, 18:45:06 UTC
76546db cpettet 06 April 2017, 20:44:00 UTC admin: add a group for cloud services roots (wmcs-roots) * With initial role applications Change-Id: I28d028ac7e6203b7f00a11043ff0d2907c232b85 Bug: T162404 10 April 2017, 18:44:51 UTC
992c116 Andrew Otto 10 April 2017, 18:27:53 UTC Add new hadoop worker nodes to site.pp Bug: T155065 Change-Id: If69a3957eb04df03c8ffe493fdb1804dec272cf6 10 April 2017, 18:27:53 UTC
6fbf9cf Andrew Otto 10 April 2017, 18:14:31 UTC Reorder more python sklearn deps to work around require_package issue Change-Id: I137980904af7e6439a4fc22e406946ba3011d4c0 10 April 2017, 18:16:45 UTC
7a1f6d2 Andrew Otto 10 April 2017, 18:07:24 UTC Install python3-sklearn-lib and make sure python3-numpy is installed first Change-Id: I10342127217e3271ca94c1f7d3cc46e27bc90328 10 April 2017, 18:07:24 UTC
5613bae Andrew Otto 10 April 2017, 18:01:54 UTC Use package resource instead of require_package for ensuring specific python-numpy versions Change-Id: Ie19ba586d81cc38a41dd20a542a6dfde03958155 10 April 2017, 18:01:54 UTC
7d38533 Andrew Otto 10 April 2017, 17:59:32 UTC Fix missing comma in package list Change-Id: I907e542a1b43a91fe92e634067914f5cd0f4cd46 10 April 2017, 17:59:32 UTC
14c12f6 Andrew Otto 10 April 2017, 17:57:33 UTC Ensuring python and python3 versions of numpy are the same on hadoop workers Change-Id: Ia921a2666dce87d19e93f3fb675c66259dc60317 10 April 2017, 17:57:33 UTC
8aea9c4 Andrew Otto 10 April 2017, 17:54:07 UTC Remove python3-numpy=version from list of installed packages to test fix require_package error Change-Id: I9da9106edd212999286248cc438fe7efb58ff859 10 April 2017, 17:54:07 UTC
04b8a49 Andrew Otto 10 April 2017, 17:52:40 UTC Using array argument for require_package Change-Id: I03cb876c58989758405070bab9004db88d681093 10 April 2017, 17:52:40 UTC
2ad8bd1 Andrew Otto 10 April 2017, 17:50:10 UTC Separating out require_package calls in attempt to figure out weird error message undefined method `function_create_resources' for nil:NilClass at /etc/puppet/modules/role/manifests/analytics_cluster/hadoop/worker.pp:120 Change-Id: I95d2d77909551c911e4566bb8a0126a4538e552b 10 April 2017, 17:50:46 UTC
33b0aff Ottomata 10 April 2017, 17:44:08 UTC Revert "Revert "Add python3 packages to hadoop workers for ORES in hadoop"" Let's try again... This reverts commit c38ef051dd508ee767cab39789f74988738bcff6. Change-Id: I63814ad125f2bc8385039386ad8822e07792f60d 10 April 2017, 17:44:52 UTC
030319a Guillaume Lederrey 05 April 2017, 13:24:39 UTC wdqs: active/active public interface This will send live user traffic to the application simultaneously in both primary DCs! Bug: T162111 Change-Id: I4dbb7a4a30e2c87f35801d9a6c42cad669cec933 10 April 2017, 17:06:55 UTC
a276afb andrewbogott 08 April 2017, 09:08:24 UTC nova-scheduler: Temporarily depool labvirt1002 It's causing us issues, might as well depool while I debug so that CI doesn't worry so much. Bug: T162529 Change-Id: Ie80057f42e1cfde81c2e802711ce1562bf6e8342 08 April 2017, 09:09:05 UTC
c38ef05 Elukey 10 April 2017, 16:31:07 UTC Revert "Add python3 packages to hadoop workers for ORES in hadoop" This reverts commit d7c5662895ba760f153bcdbffb50bdf208b568dc. Change-Id: Ie4ee42cff1264a80b239ab78c45d514bb2c994c2 10 April 2017, 16:31:21 UTC
00d5ba8 cmjohnson 10 April 2017, 15:44:22 UTC Adding mac address for analytis1068 Change-Id: Ibd689522660980aa212d5f1b03c9c548061eae4c 10 April 2017, 15:44:22 UTC
d7c5662 Andrew Otto 06 April 2017, 18:57:37 UTC Add python3 packages to hadoop workers for ORES in hadoop Change-Id: I6940bc40b0842d5a7ca8fad7413d89452b288c4b 10 April 2017, 15:30:03 UTC
2f7e990 Guillaume Lederrey 07 April 2017, 10:08:45 UTC postgresql - cleanup dead code after migration to check-postgres package Bug: T162345 Change-Id: I0d8a3ba02071ff161e8b419180666734f80b886e 10 April 2017, 15:22:17 UTC
back to top