[03:57:11] !log tools.stewardbots Deploy 435a16d [03:57:13] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools.stewardbots/SAL [05:38:33] !log admin replaced cloudservices200[2,3] with cloudservices200[4,5] [05:38:35] Logged the message at https://wikitech.wikimedia.org/wiki/Nova_Resource:Admin/SAL [13:18:19] Hi. Could I get an update on https://meta.wikimedia.org/wiki/Special:OAuthListConsumers/view/68c7f240238dcf87f8cd607d05caba9d [13:22:28] roy649: done [13:35:17] cool, thanks. [13:36:15] At some point, I'll want to get a production key. I'd be more than happy to have a security review of my code prior to that. [14:15:47] I'm having some problems figuring out the right puppet hiera values in cloud, is there a way to see how the puppet master is resolving these vars? Ref https://phabricator.wikimedia.org/P25970 [16:34:57] inflatador: I've never found a great way to understand that short of trial and error. You can see the lookup order on the puppetmaster in /etc/puppet/hiera.yaml if that helps [16:35:07] (Rook ^ also related to your recent question) [16:35:45] yeah, this is a blind spot to me, coming from ansible where I can just echo the var out with a "debug". I'm sure there's a way to do it in puppet, but coming up empty so far [16:35:58] There's an attempt to explain how this happens here https://wikitech.wikimedia.org/wiki/Puppet_Hiera [16:36:13] Yeah, been digging through that, so far I'm at https://wikitech.wikimedia.org/wiki/Portal:Cloud_VPS/Admin/Hiera#cli [16:36:17] I feel like there's some way to do a value lookup without running puppet but I don't see it in that page... [16:36:54] inflatador: on cloud-vps typically you would be setting those values using the horizon puppet UI which has a pretty simple lookup order. [16:37:12] If you need the hiera values to come straight from operations/puppet though... I'm frequently surprised by how that works [16:37:34] I'm not sure I'm helping :( [16:37:36] s/surprised/confused for me ;) [16:37:57] actually, yes...I feel like I'm not the only one now ;P [16:38:07] See, with years of experience I have the advantage of having expectations that are never correct. Rather than mere confusion :p [16:39:03] inflatador: you can do something like `taavi@deployment-puppetmaster04:~$ sudo puppet lookup --explain --compile --node deployment-mwmaint02.deployment-prep.eqiad1.wikimedia.cloud 'profile::mediawiki::php::icu63'` [16:39:22] replace the instance hostname and hiera key of course [16:40:19] ah, there it is, 'puppet lookup' [16:40:20] Thanks taavi , will give it a shot. I tried that cmd already but not with those args [16:40:37] And that only works on the puppetmaster, right, not the client? [16:40:41] not sure :/ [16:40:55] I think that's right re: puppet master only [16:41:13] inflatador: `--compile` is the important part that I usually forget, otherwise some data that's needed is not available [16:42:47] taavi: I need advice from a netbox expert, but https://gerrit.wikimedia.org/r/c/operations/dns/+/785182 should help get codfw1dev less broken [16:42:51] yeah, I think that's exactly what I missed [16:43:40] andrewbogott: I am unfortunately not a netbox expert! but if you have also somehow reserved those IPs in netbox then it looks ok to me [16:44:33] Yeah, it's the 'somehow reserved' thing that's giving me pause. I guess we have two sources of truth for this kind of thing now? [16:52:45] I think my problem here is that I'm using a SAN cert, and pretty much everywhere assumes the fqdn of the cloud server and the filename of the certificate are one and the same [16:53:03] Should be easy enough to work around [17:05:10] Hello, i have noticed my server crashes with the erro notification: "MySQL server has gone away (BrokenPipeError(32, 'Broken pipe'))"). if this is an expected behaviour, how can i work around it? i tried session.rollback handling, but it doesn't work. [17:05:11] A little information about the technical implementation: i am using Toolforge with python and Flask SQLAlchemy [17:07:37] Guest21: this sounds like it could be related to https://wikitech.wikimedia.org/wiki/Help:Toolforge/Database#Connection_handling_policy and the database server terminating an idle connection. [17:08:02] You may need to configure something like a NullPool for your sqlalchemy connections -- https://docs.sqlalchemy.org/en/14/core/pooling.html#sqlalchemy.pool.NullPool [17:08:46] nicer link: https://docs.sqlalchemy.org/en/14/core/pooling.html#switching-pool-implementations [17:22:22] @bd808 thank you for the informative links, that seems to be my problem. Do you have an example of a nullpool implementation from maybe another toolforge project ? [17:27:33] Guest21: I do not. I had to web search for the sqlalchemy links I gave you. [17:29:24] ok, thanks again for your help :)