[12:07:28] <brouberol>	 hnowlan (pinged due to git blame https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/services/change-propagation/+log/refs/heads/master) I'm planning to increase the partition count for the eqiad.mediawiki.job.htmlCacheUpdate topic, that seems to be related to changeprop. Do you know if the associated consumer is using consumer groups (which
[12:07:28] <brouberol>	 will cause it to take the new partitions into account on the fly)?  
[12:08:43] <brouberol>	 cf https://gerrit.wikimedia.org/g/mediawiki/services/change-propagation/+/9f5e06a8a0c680206a52cc1121d1e0304ea4114f/config.jobqueue.wikimedia.yaml#54
[12:45:21] <hnowlan>	 brouberol: changeprop's htmlCacheUpdate only really knows about databases for partitioning - I can't quite remember whether the consumer groups will take account but it most likely will at minimum need a restart
[12:45:26] <hnowlan>	 You'd be best off asking serviceops 
[12:46:26] <brouberol>	 thanks!
[13:16:38] <atsukoito>	 Hi, I think there is an issue with puppet and IDP: https://gerrit.wikimedia.org/r/c/operations/puppet/+/1328573
[13:17:53] <atsukoito>	 When I created the first revision, pcc failed on absent secret. When I added a placeholder secret to the labs/private, it started failing on pre-compile because other fields are empty.
[13:27:23] <_joe_>	 not sure I understand. Do you have a link to a failed pcc run after you added the secrets to labs/private?
[13:28:26] <atsukoito>	 _joe_: in this diff, rev1 is done before, and rev 2 is after adding secrets to labs/private
[13:28:41] <_joe_>	 ahhh ok so it's expected
[13:28:53] <_joe_>	 rev1 will fail pcc, if it's already merged
[13:29:47] <atsukoito>	 _joe_: do you mean that is it expected that pcc will fail no matter what?
[13:29:54] <_joe_>	 uhm no wait, I'm looking at the pcc output and it does not make sense
[13:30:01] <cdanis>	 yeah
[13:30:02] <atsukoito>	 there is the run before merging to labs/private https://puppet-compiler.wmflabs.org/output/1328573/7616/idp2005.wikimedia.org/change.idp2005.wikimedia.org.err
[13:30:05] <cdanis>	 I've been scratching my head about it
[13:30:26] <atsukoito>	 there is a run after merging to labs/private https://puppet-compiler.wmflabs.org/output/1328573/7618/idp2005.wikimedia.org/prod.idp2005.wikimedia.org.err
[13:31:10] <cdanis>	 something in the CAS puppetization must be instantiating a Apereo_cas::Service[airflow_experiment_platform] because it found hiera for it, even if it was just a placeholder secret
[13:31:29] <_joe_>	 yes I was getting there
[13:31:37] <_joe_>	 that's the only explanation
[13:32:17] <cdanis>	 227   │     $services.each |String $service, Hash $config| {
[13:32:19] <cdanis>	  228   │         apereo_cas::service { $service:
[13:32:21] <cdanis>	  229   │             * => $config,
[13:32:23] <cdanis>	  230   │         }
[13:32:25] <cdanis>	  231   │     }
[13:32:27] <cdanis>	 in modules/apereo_cas/manifests/init.pp
[13:32:33] <_joe_>	 yeah
[13:32:50] <_joe_>	 ok so yes, that's an unfortunate choice :)
[13:32:52] <cdanis>	 atsukoito: if you make that conditional on the id key being present in $config, I think both sides will work
[13:33:09] <_joe_>	 I would rather invert the logic
[13:33:27] <_joe_>	 and use $config as the source of the loop, but it's not on atsukoito to fix it now I guess
[13:37:28] <atsukoito>	 cdanis: my diff is instantiating it i guess, because I'm adding a service
[13:37:42] <cdanis>	 atsukoito: yeah exactly, and the labs/private change is used for both sides of the PCC
[13:38:04] <atsukoito>	 i'll make a diff now
[13:38:11] <cdanis>	 i'm happy to review
[13:38:21] <cdanis>	 thanks :)
[13:48:58] <atsukoito>	 fixing in https://phabricator.wikimedia.org/T435815
[13:51:39] <brouberol>	 atsukoito: can I merge you puppet-merge patch?
[13:52:09] <atsukoito>	 yup!
[13:52:30] <brouberol>	 {{done}}
[14:09:58] <atsukoito>	 cdanis: _joe_ : https://gerrit.wikimedia.org/r/c/operations/puppet/+/1328610
[14:25:46] <inflatador>	 Does anyone have an example of selecting 10Gbps hosts with cumin? The only Puppet fact for connection speed I found is `net_driver.${device}.speed`, which varies based on NIC hardware
[14:26:03] <atsukoito>	 thanks cdanis, _joe_, PCC on IDP is back
[14:26:18] <inflatador>	 of course it is very likely I missed something ;)
[14:28:42] <cdanis>	 inflatador: worst case, you can do something like this https://phabricator.wikimedia.org/P8744
[14:33:48] <atsukoito>	 cdanis: oh-h.. now IDP fails in prod with `Error while evaluating a Resource Statement, Evaluation Error: Error while evaluating a Function Call, airflow_experiment_platform: $client_secret required when using OidcRegisteredService `
[14:34:21] <atsukoito>	 I was following our old doc on spinning up OIDC, and it seem that we need to actually generate a secret in advance
[14:34:23] <inflatador>	 cdanis ACK, will take a look. Thanks!
[14:34:36] <atsukoito>	 https://wikitech.wikimedia.org/wiki/Data_Platform/Systems/Airflow/Kubernetes/Operations
[14:57:28] <cdanis>	 atsukoito: oh, yeah, you will need to have a secret in the actual private repo to have puppet work on idp in prod
[14:58:04] <atsukoito>	 from the documentation i felt like it will be generated by provider, but I was mistaken XD
[14:58:12] <atsukoito>	 I'll update our docs here