Conversation

Notices

1. Embed this notice
   Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 05:04:48 JST Evan Prodromou

   Wow, running ElasticSearch and Kibana in Kubernetes is poorly documented, requires a lot of manual steps that don't lend themselves well to automated deployment, and works badly.

   There's a lot of "when your ES pod starts, it will blurp out a bunch of garbage that you have to copy into your Kibana UI".

   I don't want that to happen, ever. I want 12-factor apps. I want to configure whatever crap ES needs in environment variables, and whatever crap Kibana needs in another set of env variables.

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 05:05:53 JST Evan Prodromou

     in reply to

     Oh, and a bunch of self-signed certificate nonsense. I don't give a fuck! This is all running inside my imaginary network in Kubernetes.

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 05:29:54 JST Evan Prodromou

     in reply to

     • Lambic

     @lambic I just want simple little black boxes that work by default and that I can tune if there are problems. Is that so much to ask?

   • Embed this notice
     Lambic (lambic@social.linux.pizza)'s status on Monday, 13-Feb-2023 05:29:55 JST Lambic

     in reply to

     @evan yep it's horrible infra to install and maintain, gets worse when you add logstash into the mix.

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 05:31:49 JST Evan Prodromou

     in reply to

     • Lambic

     @lambic Oh, and I gave up on logstash entirely and wrote our own indexer in Python. We spent days trying to get it to do anything, and it just never worked.

   • Embed this notice
     JR (csosorchid@universeodon.com)'s status on Monday, 13-Feb-2023 05:32:54 JST JR

     in reply to

     @evan That sounds sort of technical and specific, can that AI program be sweet talked into writing a program?

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 05:32:54 JST Evan Prodromou

     in reply to

     • JR

     @Csosorchid hmm.

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 09:20:57 JST Evan Prodromou

     in reply to

     • Mike Garuccio

     @mgaruccio nope! Thanks for the tip

   • Embed this notice
     Mike Garuccio (mgaruccio@hachyderm.io)'s status on Monday, 13-Feb-2023 09:20:58 JST Mike Garuccio

     in reply to

     @evan

     Are you using the eck operator? I don’t remember it being super involved last time I worked with it.

   • Embed this notice
     Evan Prodromou (evan@prodromou.pub)'s status on Monday, 13-Feb-2023 12:33:43 JST Evan Prodromou

     in reply to

     • Mike Garuccio

     @mgaruccio so, these are the custom classes in Kubernetes? How do they work? I strongly distrust them, and don't like having to install and maintain them on every cluster we use. The lifecycle, compared with just using Docker images, is really opaque. Also, where's our data going?

   • Embed this notice
     Mike Garuccio (mgaruccio@hachyderm.io)'s status on Monday, 13-Feb-2023 12:33:44 JST Mike Garuccio

     in reply to

     @evan

     Happy to help! I run a fairly absurd number of prod clusters on their older custom orchestrator so feel free to post back if you run into other roadblocks.

   • Embed this notice
     Dr. Matt Lee ?:verified: (mattl@social.mat.tl)'s status on Tuesday, 14-Feb-2023 10:19:47 JST Dr. Matt Lee ?:verified:

     in reply to

     @evan computers are shit