[10:49:58] Any puppet/rake/ci experts around? I'm having some weird issues when running the ci tests: https://integration.wikimedia.org/ci/job/operations-puppet-tests-buster-docker/22057/console, it complains that there's no `parallel_spec` rake task :/ (I just added a new module test) [10:52:46] dcaro: looking [10:53:58] running rake global:parallel_spec works though (using `rbenv exec bundle exec` ...) [10:58:53] jbond42: thanks! [10:58:56] dcaro: tl;dr you need to add a Rakfile just copt one from a different module (which i have allready done for you). The way the main rake task works in CI is that it cd's into each module directory and and runs rake parallel_spec (use to be rake spec). so you need a Rakefile in the module dir so that you can run rake tasks from the module level. this is also usefull as it means you can run the [10:59:02] rake tasks just in the individual module and ... [10:59:04] ... not have to run all tasks [10:59:26] it's a very confusing error though xd [11:02:03] the global:parallel_spec task is a taks i created to try and change the forking `cd modules/$module ; rake spec` flow and have just one `rake parallel_spec` task run all required spec test. however there are other issues there around the mocker used in various modules see here https://gerrit.wikimedia.org/r/c/operations/puppet/+/636923 [11:02:18] ack to the error message i think we should be able to fix that fairly easily [11:02:45] i.e. global:parallel_spec is a bit experimental [11:05:35] dcaro: global:spec will force a spec test of all modules in the same way that CI runs. bundle exec rake spec will run in the same way as CI i.e. only acting on changed modules and there dependencies [11:28:55] ack [11:52:01] dcaro: https://gerrit.wikimedia.org/r/c/operations/puppet/+/672698/2 [11:52:05] failed e.g. https://integration.wikimedia.org/ci/job/operations-puppet-tests-buster-docker/22070/console [11:54:39] and this time with the correct thing failing https://integration.wikimedia.org/ci/job/operations-puppet-tests-buster-docker/22071/console :) [12:35:04] moritzm: jbond42: thanks for the java package duplicate resource work ;) [12:35:23] np [12:35:38] the other change about puppet compiler I dont quite know https://gerrit.wikimedia.org/r/c/operations/puppet/+/672660 ;) [12:35:51] I though about adding profile::java but gave up finding how to apply that [12:36:00] seems puppetdb package already depends on java so that is good enough [12:39:00] sgtm [12:39:49] hnowlan: FYI there's an alert on icinga for "PROBLEM - Uncommitted DNS changes in Netbox on netbox1001 is CRITICAL" that is related to the change from reserved to active of the "similar-users" SVC IPs [12:40:18] although no-op in prod, it actually requires to run the sre.dns.netbox cookbook. I can run it for you (have already run it in dry-run to know what was the diff [12:40:21] ) [12:42:22] hashar: indeed, puppetdb was converted to profile::java a while ago, so that seems good enough [12:42:56] which would let me switch to profile::java in the profile used for the Jenkins agents (some being puppet compilers) [12:42:59] that is a long tail [12:43:01] ;D [12:45:01] volans: uhhh that's news to me [12:45:05] what changed their state? [12:45:09] Go ahead anyway :) [12:45:17] hnowlan: https://netbox.wikimedia.org/extras/changelog/51541/ [12:45:51] was part of changes related to kubernetes2009 [12:45:54] is that IP allocated there? [12:46:05] if so the puppetdb import after the reimage would have picked it up [12:46:12] https://netbox.wikimedia.org/extras/changelog/?request_id=16319bb6-2e81-4a34-a03b-b7fb3be0f411 [12:48:37] aha [12:48:59] if you have the cookbook waiting to run go for it, or I can [12:49:07] ok I'm running it [12:49:13] thanks! [12:49:20] is there a related task? [12:49:25] I can link to the commit [12:51:29] T268837 is probably best [12:51:29] T268837: Create Helm chart for Similarusers - https://phabricator.wikimedia.org/T268837 [12:51:34] ack thx [12:52:01] ofc only the codfw one was changed, the eqiad will happen when an eqiad host will be done [12:52:14] sorry for the trouble, cookbook running [12:55:47] hnowlan: all done :) [13:01:56] * volans lunch [13:05:08] thanks volans! [13:05:54] everyone is grateful when volans goes to lunch [13:21:27] lol [16:13:52] xd [16:59:57] Hello. Anyone around willing to +2 https://gerrit.wikimedia.org/r/c/operations/puppet/+/672483 ? [17:00:08] (updates to logspam-watch) [17:12:49] dancy not entierly sure with the funtional change however i se brennen the original author hasgiven it a +1 so will merge [17:16:12] dancy: merged and deployed to mwlog1001 [17:20:13] thanks jbond42 [17:22:23] np [17:23:19] Thanks! [18:24:25] I could use help from a puppet guru on https://gerrit.wikimedia.org/r/c/operations/puppet/+/672537 - not sure what the right syntax for query_facts() is [18:30:28] blind guess: does it want quotes around the path, inside File[]? [18:50:33] * legoktm tries [18:52:30] rzl: well it compiles now :) though it doesn't look like it's matching the hosts I want it to [18:54:11] but since this relies on a private hiera value, not sure if it's a PCC limitation? I have it set in labs/private [18:54:47] ah yeah, PCC can't help you there [18:56:41] I remember there's some weirdness in how exactly PCC handles that -- obviously it doesn't have access to the actual secrets, but I think it's more complicated than that for implementation reasons [18:57:44] j.oe or j.bond could tell you the details but TLDR it doesn't necessarily mean your code is wrong [18:59:24] gotcha [18:59:49] it worked fine for when it was just Class[k8s::kubelet] returned one of each type of host, which is what I remember reading about in the PCC docs [19:05:55] legoktm: there is also https://wikitech.wikimedia.org/wiki/Nova_Resource:Puppet-diffs#How_to_update_the_compiler's_facts?_(e.g._INFO:_Unable_to_find_facts_for_host_conf2001.codfw.wmnet,_skipping) [19:06:00] possibly related; I can't say for sur [19:07:24] legoktm: I don't recall how that works behind the scenes but the puppetdb API for resources doesn't allow to query for both a fact and a resounce in the same call [19:07:40] (it's the reason why cumin has the global grammar) [19:09:16] from the code it's doing: [19:09:16] puppetdb.query(:facts, query, :extract => [:certname, :name, :value]), facts [19:10:48] sorry, can't dig into it right now though [19:30:37] hmm [19:31:04] I guess then I could make two queries and do the intersection in code?