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

sort by:
Revision Author Date Message Commit Date
de987c4 ciphersuite: remove chapoly draft-mode ciphers These will go away implicitly when we switch to openssl-1.1.0, so we should make this transition first if we can to separate the effects we see in graphs. Currently ~3.8% of clients are using this. That number has been on the decline, and those clients should revert to the same strong (but slower, for them) AES-GCM options they had before we deployed the chapoly patches. The clients can fix this on their own by upgrading Chrome. Some percentage is coming from a Google Search App on iOS that has yet to see an RFC chapoly release, but I don't think we care much about it in the net. Change-Id: Id7b7bb38b2e55df17e1a1c7f942e693535ce00dd 20 September 2016, 14:37:29 UTC
ae19e1b Use 30g of heap for relforge jvms Change-Id: I8f65e7077fa6e21ba0a703fc87ee5c405623e905 20 September 2016, 13:52:53 UTC
9006aa6 package_builder: doc that gbp must be passed -sa dpkg-genchanges does add the original tarball checksum in the .changes file unless the upstream version has changed compared to the previous debian/changelog entry. I typically build packages locally in fast iterations and by the time I get a package ready for upload I already have several debian version increments which have not been pushed to apt.wikimedia.org. Thus, when publishing a package with reprepro, it complains about the tarball being missing from .changes or sometime cant find it out in the pool. The fix is to have the .changes file to always list the original tarball. Debian Policy about the .changes file states: > If a new Debian revision of a package is being shipped and no new > original source archive is being distributed [..] the .changes file vvvvvv > should leave it out. In this case the original source archive on the ^^^^^^ > distribution site must match exactly, byte-for-byte, the original > source archive which was used to generate the .dsc file and diff which > are being uploaded. So we can safely always include the original tarball to please reprepro and still not violate the Debian policy. Update package_builder README.md file example of git-buildpackage to include '-sa'. Add a hint about configuring the builder command in the git buildpackage configuration file (~/gbp.conf). Which once done solve the issue for all packages. Bug: T145797 Change-Id: I28c8e254e9e260646a8ed5f584f286f94c241f0f 20 September 2016, 13:50:43 UTC
393b141 Update varnishkafka submodule Minor README.md change Change-Id: I408819681aca9ddbbeec68339acb7c85c914c10c 20 September 2016, 13:46:24 UTC
a54bea5 Update varnishkafka submodule Minor README.md change Change-Id: I83bbd2d054207d306d717a08988ebcdda9150357 20 September 2016, 13:37:56 UTC
6459e3f Add Thumbor config values moved out of package The 0.1.21 package upgrade moves these config values out of 40-wikimedia.conf. It makes more sense to define them in puppet. Change-Id: I4318fae50e51303df3b93fbd7694bfa9a1e95129 20 September 2016, 13:33:54 UTC
9be4834 Follow-up I55032bf7: urlopen context argument was added in Python 2.7.9 too Change-Id: Id0cfcc6907c37c0960b5a614a041f7f974b7fb7b 20 September 2016, 13:33:14 UTC
1df2525 Finish adding --until param to check_graphite script Cleanup follow up of https://gerrit.wikimedia.org/r/#/c/255415/ This has been sitting for a while. The check_graphite_until_temp script that added the --until paramater has been running just fine for eventlogging and varnishkafka checks. I tested this script manually, and it does what is expected. e.g. ./check_graphite --url http://graphite.wikimedia.org check_threshold 'kafka.cluster.analytics-eqiad.kafka.kafka1012_eqiad_wmnet_9999.kafka.server.BrokerTopicMetrics-AllTopics.MessagesInPerSec.OneMinuteRate' --from 20min --until=-0min --under -C 15000 -W 18000 Varying the --until (including for the default of 0min) works as expected. Bug: T116035 Change-Id: I863367b8b5e97f501ebabc4eacbac081262462d6 20 September 2016, 13:29:49 UTC
7d20938 elasticsearch: Remove id_hash_mod.groovy We are ensuring absent /etc/elasticsearch/scripts as well as the mwgrep.groovy script already. While investigating this on our elastic search clusters, id_hash_mod.groovy as an elastic search script was noticed. Assuming we don't use it, set it as asbsent as well Change-Id: I8bc5cef55a5f8ab87b1d098afa765b39335f0222 20 September 2016, 13:26:05 UTC
15bbdee puppet-merge: submodule diff should honor QUIET We were not honoring the QUIET parameter on puppet-merge for submodule diffs. Honor it. Also force temporary directory removal in order to avoid prompts Change-Id: I317e3c7399bdb7172d37986b2b20855c12995206 20 September 2016, 13:06:24 UTC
9b1b708 letsencrypt: acme_tiny: Skip my ignore-HTTPS-validation code for old versions of python We don't need it for those older versions (default behaviour does what we want) and it broke on precise (e.g. carbon) Change-Id: I55032bf7573ab006d4b9e441421bc0ca35fc9450 20 September 2016, 12:46:09 UTC
fb256cc Always refresh Package/Release/Translation files Taken from squid-deb-proxy. Without this, we've occasionally been running into installation errors after changes to the Packages files (e.g. after a jessie point release), caused by stale cache returns for Packages files. Change-Id: I264b5a8f01afdf719f21fc2764ef4ce475d7c792 20 September 2016, 12:32:37 UTC
816acb5 mariadb: implement custom mysqld_safe script to all servers Enforce it to those that are using the wmf-mariadb or wmf-mariadb10 packages. Bug: T145378 Change-Id: I204dd59b0569572062c62894546b567297cc9a6f 20 September 2016, 11:28:05 UTC
5b40682 Add jobrunner02 to deployment-prep Bug: T144006 Change-Id: If60190ca89d8cb7fd53f0d52026807da36077cb8 20 September 2016, 11:12:09 UTC
13a5343 puppetmaster: Ping before sending requests to backend ping argument will actually send a 100-Continue request to the backend server. This should fix a set of errors where the backend will close the connection while the frontend will have sent the request already, resulting to the frontend issuing a 502 to the agent Change-Id: I251c259b1336e650061abd5959eec292cea57e67 20 September 2016, 09:53:12 UTC
1a92a1f puppetmaster: add proxy-initial-not-pooled to frontends We've been seeing a few errors that might be related to this; let's see if it helps. Change-Id: Ica4085cf0d578709cbb374eba244e07464a7d6ac 20 September 2016, 07:04:44 UTC
320459f labs: Make puppet-enc read from the config file Change-Id: I2e2bf176993f22c6c805f8118147f6e961f70655 20 September 2016, 01:36:52 UTC
199f62d labs: Add config for the puppet enc Contains info on the host on which the rest backend lives Change-Id: I56fe1945e10e1771292b2a2305408a105eb92d6d 20 September 2016, 01:31:06 UTC
3caa2df varnish-backend-restart: increase sleep times We still see very small 503 spikes on most of the cron-automated restarts, and insufficient sleep is a likely reason why. Change-Id: Ifebc99a932440dc03650012be528f04b5c03b083 19 September 2016, 23:06:05 UTC
476e61e Puppet Panel: Remove the ?-in-circle This is nice when it works, but most browsers cannot render that icon. After this patch, docs will appear as hover text over the puppet role name. Change-Id: Ic7c7bee8aab7832aaa42d9a6b083a43e28cd90a7 19 September 2016, 22:46:47 UTC
bfb2bd0 upload frontend: limit at 256KB more ongoing experimentation. 1MB seems to work pretty decently. I've tried 512KB before, but reverted after seeing very little effect. I may not have waited long enough to see it, though. Also, recent data analysis for the backends indicates 256K might be a much better value here anyways. Change-Id: I378525339347361286ef2a5187fb7912244dcd54 19 September 2016, 22:38:26 UTC
e8b482c labspuppetbackend: Set charset explicitly Also make sure empty checks check for both str and byte Change-Id: Icad5a92bbd5282e0726ffd3071c98d627cf35346 19 September 2016, 22:32:58 UTC
2f077c3 upload storage: no restart cron w/ experiment Bug: T145661 Change-Id: I9551d542bea3c1e21cdeac8d01a5c807291eaf30 19 September 2016, 22:23:47 UTC
dca9740 upload storage: fix CL comparisons Failed to use std.integer(), cleaned up related logic a bit after checking that std.integer() properly handles up to signed 64-bit values on our hosts. Bug: T145661 Change-Id: If79a9ad86d159000e597115ab73712a0e6aed0ad 19 September 2016, 22:13:26 UTC
4ab94f9 upload storage size class experiment This will probably need tweaking. Uses cp1099 as first test host. Bug: T145661 Change-Id: I024b54ed3ae1446743f1a3b15dd3842987a0fda9 19 September 2016, 22:02:19 UTC
3c696e1 wmflib: Return default dir when role::puppet::self isn't used Bug: T120159 Change-Id: Ia59e587ff56fce628556fe556f9fa9f759b82945 19 September 2016, 21:56:48 UTC
ec6a1d4 Reimage: improve output in case of errors Bug: T143536 Change-Id: Idc8fe2152d1684c3a0e32100f975305f61585755 19 September 2016, 21:48:18 UTC
e7ed079 Puppet Panel: Use the right template for the remove-role dialog Change-Id: I8447488b90a0cf494d57967afeb669f142c52289 16 September 2016, 10:11:11 UTC
1e62ebd labspuppetbackend: fix a bug with detecting project-wide prefixes The db returns b'' instead of '' Change-Id: Ia93b974d9f812e910bd70210fa2144e1bdbb7959 16 September 2016, 09:20:03 UTC
d649e7b labs: Add a standalone puppetmaster role - Uses the puppetmaster module rather than the puppet module. - Based off the labs puppetmaster code, but much simplified - Uses apache + mod_passenger like prod rather than the default WEBrick server - Probably does not support precise! Should work on trusty & jessie. - Is standalone puppetmaster, does not support replication / HA like the prod puppetmasters DEATH TO THE PUPPET MODULE Bug: T120159 Change-Id: I5f11761bdd2a1f292d3d061363fa53346d1eb768 19 September 2016, 19:45:54 UTC
bf9f5c3 Puppet Panel: Don't choke on a (user-provided) "" hiera string. Bug: T91990 Change-Id: If842893fe08099bd98493bb46f3bf46afe04d529 16 September 2016, 08:18:07 UTC
e2accbc Puppet panel: Use "{}" to represent an empty hiera config We were using "" in some cases before, which wound up unparsable down the line. Bug: T91990 Change-Id: I1e3886270ef5a14d1073aac7e2fa80b6271fa2ff 16 September 2016, 08:12:01 UTC
dfe4fb5 mariadb: Add additional node (db1061) for mysqld_safe testing Last test before cluster-wide deploy. Bug: T145378 Change-Id: I2e5cd1d78b891c7dd357230019c69a19581b7a27 19 September 2016, 17:20:10 UTC
97f3e31 Puppet Panel: check for empty hiera before we start enumerating keys. Change-Id: I3d8f6709bfce442cf76903ab9e1b0fee4517933b 16 September 2016, 07:22:42 UTC
f1c4fbd Puppet Horizon Panel: Deploy on Labs Bug: T91990 Change-Id: I89f9ef6a896b711acfaef305dc767a50ed50c672 16 September 2016, 06:24:20 UTC
40d02ab mirrors: fix autoindex for tails Indent the location /tails block under the location / that has some defaults. Change-Id: I4681abd6cb7bd4de4355c2a016a7bde085d0a832 19 September 2016, 15:48:13 UTC
a9e812b robh back from vacation Revert "robh on vacation, remove from paging" This reverts commit 3d38196b79dc478da1c08be96daa1447bd72be2f. Change-Id: Ie01b3ec258e13ef5d933d4cd2c67025e8d5ecd6a 19 September 2016, 15:40:26 UTC
2002233 puppetdbquery: fix argument ordering Change-Id: I68105ab79c4a67ef9f4bfff9c6c61cab9ca713e1 19 September 2016, 15:00:26 UTC
fcf9c75 ssh::client: order by host name with puppetdb as well Change-Id: I47d1f6492f02b38765cbd040acd0a5710bef69e6 19 September 2016, 14:21:42 UTC
03db64c puppetdbquery: add ability to order resources Change-Id: I28bf175231ab025f367bb51bf83c41b29c83a2c6 19 September 2016, 14:20:30 UTC
2d4863d naggen2: order resources by title in puppetDB as well Also, sort all the parameters for reducing differences further. Change-Id: Iebc821b9adf465fb48115b23543c56f6ce8fe7b0 19 September 2016, 13:48:28 UTC
eb273ba deployment_server: Daemonise redis when running on systemd Otherwise the redis startup fails since the systemd unit uses Type=forking. Bug: T144578 Change-Id: I7d2b015d6c1aab9af65d999e0c884378266a3845 19 September 2016, 12:31:20 UTC
bf9bac7 remove nfs mount of labstore host from datasets unused for some time. Change-Id: If6b8f18c9f1ac40034cbaa74a615f7ec26f01174 19 September 2016, 12:11:57 UTC
661d19e set up static nfs lock manager ports for dataset hosts Change-Id: I316d8d387c743492326a8e3bc8df41be16b5a0cf 19 September 2016, 10:34:10 UTC
b81da8d mirrors: disable ETag for Tails, at their request Change-Id: I54b06909372cd67b427e7e47204757c7b06cb1b3 19 September 2016, 09:06:58 UTC
323b97d fix another possible netmapper-1.3+v4 FE crash Change-Id: Iddb35c4438e4a0efe4b394f2aae839fa0766eaf7 18 September 2016, 17:56:41 UTC
606aea9 cache_upload: two-hit-wonder experiment, hit/2+ Bug: T144187 Change-Id: Ibe2fcfba9005f72295c179b63a3fe67f7048395d 18 September 2016, 15:29:26 UTC
cbb2cf7 cache_upload: FE size limit 1MB 512KB didn't offer notable improvement over 1MB, and 2MB seems to have slightly dropped FE hitrate, so 1MB seems reasonably-optimal for now. Change-Id: Id748ebd4a513867b1306c41b8712753e14da02eb 18 September 2016, 15:21:44 UTC
d6c822a htcppurger: quoting bugfix for host_regex Even with a simple string, apparently it's still possible to get quoting wrong through the various layers :) Change-Id: I537ebc0e25224c37a683fdc11ab7aa52511665e2 18 September 2016, 14:07:28 UTC
e9e9308 cache_upload: vhtcpd host regex filter Only about 0.1% (or less) of the total multicast purge flow actually applies to the upload caches. The rest is the crazy mess of redundant/template/backref-driven text purges. Reducing the rate of actual PURGE into varnishd may help a lot with various problems in cache_upload for now, and vhtcpd is capable of regex-filtering on the purged URL. The full flow can still be seen in vhtcpd's inpkts_received and inpkts_sane, but the reduce number from this fitering will be seen in inpkts_enqueued and other stats further into the pipeline. We never got around to this optimization before because the perfect was the enemy of the good and it was hard to think through and test the quoting for a proper, perfectly-matching regex with several layers of interpolation and quoting between puppet and the actual vhtcpd commandline. However, using the literal simple string 'upload' as a regex filter does the job fine in this case, and is all we really need. Change-Id: I605638d8e60c9c52ecc8368afc4a1705112f57d1 18 September 2016, 13:54:59 UTC
3941129 cache_upload: increase FE size limit to 2MB Part of ongoing experiments to find an approximately-optimal value for maximizing FE hitrate. Change-Id: I8893a839c9fa23bb048a5469994b3e8353bc55ed 18 September 2016, 13:05:18 UTC
748a3dc varnish-backend-restart: fix service invocation 'service' is under /usr/sbin, which is not in cron's path Change-Id: If413fcef0167cb49469acec13c6ec44ea80b3a06 18 September 2016, 10:30:38 UTC
36d82b8 cron_splay() with first use in cache_upload Change-Id: I995c8e55018bbd6544a55cae744658e972c72726 17 September 2016, 19:03:17 UTC
0148fc4 Attempt to fix salt key monitoring sudo rule The script doesn't attempt to re-run itself with sudo, it runs `/usr/bin/salt-key -l un` Bug: T144801 Change-Id: Ib594b0284029172f30929461ac45f12e5f7756ee 17 September 2016, 17:12:04 UTC
3571509 Revert "upload: splay restart cron ever betterer" This reverts commit 2eb4b4c4fe189824a110a7ddca2f0783dbccd6a0. Change-Id: Ied2bdc8d8df9cdd0b729bde6f52063a208d754c3 17 September 2016, 16:12:54 UTC
2eb4b4c upload: splay restart cron ever betterer Change-Id: Ide899e2fbaacb0787dea9dcff066e8429735f443 17 September 2016, 16:00:02 UTC
cdb1ea8 upload: splay restart cron a lot better Change-Id: I4e0bebc0fc97dd48f4b1edfceae042fa76b67316 17 September 2016, 15:25:29 UTC
d34d187 upload: drop FE size limit 1MB->512KB Change-Id: I459fea3da58f856498d71c1afc7efaea79036e4d 17 September 2016, 14:53:28 UTC
1b77869 upload: splay restart cron a little better Change-Id: Ib97ff2652fb523373af5d62f0de58cb612bde926 17 September 2016, 14:47:40 UTC
da097c1 varnish: restart upload backends once a day Automate varnish-be restarts in cache_upload in order to avoid prod issues while we're debugging T145661. Change-Id: If07089ac19059250ab5c3429d686fed33db17092 17 September 2016, 14:21:17 UTC
188daca salt/icinga: fix permissions to run plugin script Change-Id: I4718170cf687ed9f1410e6cde0417eca3802148f 17 September 2016, 01:29:51 UTC
860a843 salt: add Icinga plugin to check for unaccepted keys Add an Icinga plugin script to check for unaccepted salt keys and install it on the master via the salt module. Needs sudo to run salt-key. Fails if it can't get the key list, otherwise counts them and alerts if >= 1. Bug: T144801 Change-Id: If15055ca58673b12dfa0f64f214fece7335d1085 17 September 2016, 00:57:05 UTC
0b6e939 Puppet Panel: Cache the project panel tab Without this we load it twice sometimes Change-Id: I1c3b556ba03bee681222081d77d85748d7d9a4cc 16 September 2016, 20:45:38 UTC
0114c57 Update Alphos' blog URL for fr.planet Feed URL had an extraneous /wordpress. Initially requested on Freenode #wikidata-fr. The site editor has been notified of the opportunity to switch to HTTPS and will consider to use Let's encrypt in a near term. Change-Id: I74d2a6f204ec5266172a0f98136af5daca11ce2c 16 September 2016, 18:38:48 UTC
2073363 Update list of mailman site languages Remove ast, gl and sk from mailman site list. The debconf script parses the currently used mailing list languages and presents the admin a list which to enable (the debconf dialogue is pretty ridicilous these days, since all it does it potentially save 150k of disk space per language...), but we currently don't seem to have any more lists in Asturian, Galician or Slovak, so the debconf::set tries to enable a setting which isn't presented to begin with. This leads to the mailman services getting stopped as fallout of an inconsistent installation status. The cleanest fix would be to have the puppet module parse the list of used mailing list languages and pass that to debconf::set, but I rather made the quick fix to allow to reenable puppet on fermium. The list of mailing list languages is rather static anyway and mailman3 will have a complete makeover anyway, so that will very likely be obsolete anyway. Bug: T144933 Change-Id: I166ddc1e1182697d5c6ae4d16f195f6c5e1b391e 16 September 2016, 17:59:41 UTC
3610285 contint: drop now unused sudo rule The gerrit-sync-ve-push.sh is no more. It was to workaround a bug in Gerrit which is now solved. Remove the sudo rule from gallium, the script has already been deleted everywhere. Bug: T51846 Change-Id: I9b10f588cc97af0c7af8a75eaec48cd1fd7768e0 16 September 2016, 17:49:53 UTC
2ecba9e Update ebernhardson ssh key Change-Id: I07b268750623f002d7798e30514ea036c0dbc45c 16 September 2016, 17:45:51 UTC
21fa7ae Puppet Panel: Remove use of breadcrumb_nav This turns out to be unavailable on Liberty, and doesn't seem to do much in any case. Bug: T91990 Change-Id: I41764250b950e211b5e15a85f579c4a11bc6c4ae 16 September 2016, 17:41:48 UTC
5d472cd mediawiki_singlenode: Use require_package instead of package latest Bug: T115348 Change-Id: Ia6b14aa2ea4d38e813d1257e455a318ac0859cf3 16 September 2016, 17:28:54 UTC
4324103 Puppet Panel: Add captions to instance tabs Since a given instance can be getting config from lots of different places, this should clarify things a bit. Bug: T91990 Change-Id: Idf71e3293d82ccaabc81fd52203a23ec02acb496 16 September 2016, 17:19:08 UTC
6b3bd11 ldap::client::utils: don't ensure => latest packages Bug: T115348 Change-Id: I98e3b7bd4c06d62a1d3e24bf7673c63a3c381277 16 September 2016, 17:05:09 UTC
d7605cf ldap: Use require_package instead of package latest Bug: T115348 Change-Id: Ic263fd61a6023486c02735e0a9163dc29b131243 16 September 2016, 16:54:19 UTC
a0eb4eb partman: delete some more unused recipes To start reducing the number of different partman recipes we have (as Brandon pointed out it's like 57(!sic), detecting the recipes that are not used in netboot.cfg and deleting them. Used this: ~/wmf/puppet/modules/install_server/files/autoinstall/partman$ for recipe in *.cfg; do if ! grep -q $recipe ../netboot.cfg; then echo "$recipe not used"; git rm $recipe; fi; done I already deleted some for cassandra hosts in I352b73878036dbc4, rebased on that, these are left now: analytics-cisco.cfg not used rm 'modules/install_server/files/autoinstall/partman/analytics-cisco.cfg' lvm-etcd.cfg not used rm 'modules/install_server/files/autoinstall/partman/lvm-etcd.cfg' raid0-lvm.cfg not used rm 'modules/install_server/files/autoinstall/partman/raid0-lvm.cfg' raid1-squid.cfg not used rm 'modules/install_server/files/autoinstall/partman/raid1-squid.cfg' raid5-gpt.cfg not used rm 'modules/install_server/files/autoinstall/partman/raid5-gpt.cfg' Change-Id: Ic05660855aab95e91909425f8548f9e9215ef642 16 September 2016, 16:15:08 UTC
53397ef puppetmaster: re-enable puppetmaster2002 Change-Id: I1c77662a6ff824d0f1fad6592bb794e33ca11f8f 16 September 2016, 14:48:55 UTC
96d90f4 puppetmaster: Send reports to puppetdb We 'd like to evaluate the reports handling of puppet agent run by PuppetDB. Enable it on puppetDB enabled puppetmasters Change-Id: I8015e6ee6e397027579841b8113497d6130da401 16 September 2016, 14:47:21 UTC
c2bc5f1 puppet.conf: Remove reports = statsd from agent puppet agent does not process reports in any way. It's only puppet apply and puppet master the process reports. So that setting is irrelevant. Remove it Change-Id: Ia2b59281356387706e54f9b9ef4060335c6342c3 16 September 2016, 14:47:04 UTC
eb88835 puppet_compiler: clean older output dirs Bug: T143671 Change-Id: I3c66a0debff7788e3ef97792d35c49681eaacb34 16 September 2016, 13:55:02 UTC
4436a25 varnish: add varnish-be restart script varnish-backend-restart is a script to depool, restart and repool a varnish backend. Change-Id: I9c29d57c46a982af165417125acb10a9768accef 16 September 2016, 13:49:22 UTC
2640b72 Revert "puppetmaster: throw away reports" This reverts commit f75f274e13defd1d4fdf9f8a7165d54d1eeaf111. Change-Id: I9f87e3e94c4956aba16c3ed0be205022934d2288 16 September 2016, 13:35:50 UTC
4b13f79 prometheus: ruby has Array.push, not append Change-Id: I83162c8b945c899a0c56ff73c76c83760450b036 16 September 2016, 12:08:08 UTC
5376596 prometheus: use scope.function in template Change-Id: I3f009738af039bb597610df5da871452cc8d7dc5 16 September 2016, 11:51:19 UTC
af20b64 Add the new aqs nodes to conftool-data This patch should add the new aqs100[456] nodes to conftool-data to finally put them in production. Some considerations: 1) adding nodes to conftool-data does not mean that they will get traffic from LVS since the 'pooled' tag will need to be manually set to True 2) there are no extra networking settings to allow the new AQS cluster nodes to get traffic from LVS The plan is to enable one node at the time (maybe with weight:5 or similar) and observe how it goes. Eventually aqs100[123] will be completely removed and decommissioned. In case of fire during the transition period, we'll just need to use conftool to remove the new aqs nodes. Both clusters have the same data, we have done consistency checks. Bug: T144497 Change-Id: I011f8262d1581e8c879a5778915240451a65991a 16 September 2016, 09:18:32 UTC
6131188 prometheus: fix calls to functions from template Change-Id: I60ec3b5a48b1461423c1e02bcc66294e2e9a60e8 16 September 2016, 09:14:19 UTC
ca8aad2 templates: move apache templates to the role module Change-Id: Id267b2d1f692bbe5659ccbf5e1ef200fb57fc369 16 September 2016, 08:55:11 UTC
cf1e7ea monitoring: validate check_prometheus args Change-Id: I9338a627d07c7d5d80914b3c06b2f081c6ee9598 16 September 2016, 08:34:11 UTC
bd56cf6 Remove access credentials for ironholds Initially aimed for continuing access via MOU, but the process stalled. Change-Id: Ib8d266b8d5e6f314648c39312b3e030baadf7792 16 September 2016, 07:41:59 UTC
26fa711 puppet-merge: Run conftool-merge in only 1 frontend No need to run conftool-merge on multiple frontends. It won't hurt but we don't really need to and gitpuppet user does not really have the right to run it resulting at an error Change-Id: I8b386fa468c8f787d4d20141d7a0a5f86b6713c1 16 September 2016, 07:26:38 UTC
0541c54 exim: move templates to the role module Change-Id: I60dd84e00a582846c471874442571c00f6d09a6e 16 September 2016, 07:16:29 UTC
7fb14ee Add a directive to mod_proxy_html's yarn configuration Sadly a lot of a hrefs to replace in the yarn proxied content are embedded in inline JS. This will slow down a bit the overall rendering time but it will improve the user experience. Bug: T116192 Change-Id: I90a66f1fbbce76b5f539d149398a4efdd36fdc65 16 September 2016, 07:07:05 UTC
48f615b beta: import scap_masters list from wikitech Duplicate information between puppet.git and Wikitech is crazy. Import the scap::dsh::scap_masters hiera key in puppet. Source: https://wikitech.wikimedia.org/wiki/Hiera:Deployment-prep Change-Id: Ib6c5149888149576d2204c30b683a5d042f19cc6 16 September 2016, 06:58:40 UTC
fbc945c Puppet Panel: Speed up role formatting These format functions are really the responsiblity of the view, but moving them into the model (puppet_roles.py) allows us to cache the formatted role properties. This should save us a tiny bit of render time. Bug: T91990 Change-Id: I1af176860e6c0ac42012f2a0bd8c349f19c74e11 15 September 2016, 22:53:46 UTC
602ba50 labs: Fix httpyaml backend for precise puppetmasters Apparently ruby version on precise doesn't like two params with the same name Change-Id: Ia3ed8f776955149ed0aab316f692b9a11679e8bd 15 September 2016, 21:51:17 UTC
9df3208 Puppet panel: Remove some accidental copypasta from the puppet tab Change-Id: Ic4b8111152c3f33ca8cc3845a6bb42521ba8a8d8 15 September 2016, 21:11:13 UTC
dc272a0 puppet panel: Add a button to remove prefixes. Bug: T91990 Change-Id: I6364cd347dcbb1e319c9acee65ac6affafb5a9f2 15 September 2016, 20:42:58 UTC
5bbefc9 labspuppetbackend: fixed mysql typo Bug: T133412 Change-Id: I681a078f1b132b3f869b3c011d5bdd69d40f29b1 15 September 2016, 06:12:19 UTC
b3e47d5 labspuppetbackend: Add prefix delete Bug: T133412 Change-Id: Ieda0b9a74b25ef94b699daebdd41a1a33d5128f4 15 September 2016, 20:24:28 UTC
bed63a2 labs: Use puppetmaster service URL for hiera backend This would allow it to work for self hosted puppetmasters Bug: T91990 Change-Id: I9493a978f4892c4accf09794c7027cb50f6beb79 15 September 2016, 20:06:40 UTC
5250817 labs: Add httpyaml hiera backend to labs Bug: T91990 Change-Id: Ideb7cdb3261ec9469a977e3124ba469535e900de 15 September 2016, 19:58:32 UTC
d9055e2 Revert "labtest hiera: use labtestwikitech, not wikitech" This broke puppet runs on labtest. Bug: T145808 This reverts commit 52bccb23fc36a1192d153589ca8448d5f066f5ec. Change-Id: Ie307273b8c2a719348f6aa20c07d6a90147e96a1 15 September 2016, 18:22:54 UTC
1a674ab labs: Remove debugging from httpyaml backend I also found how to test changes - the files actually live in /var/lib/puppet/lib, so just changing the files in /var/lib/git has no effect. Change-Id: Ibaf90d639c75aaa58945c9a17f76bea1bd311f4b 15 September 2016, 18:19:07 UTC
back to top