[15:57:42] <James_F>	 Are there any examples of sidecars in production services (other than the service mesh)? I'm thinking about an MW object cache service, maybe subscribed to the change-prop feed, to avoid network fetches to MW instances (WF.org and WD.org) for the function-orchestrator.
[16:35:03] <mszabo>	 James_F: there was T244340 which sounds similar to what you are proposing but it got decommissioned AFAIK
[16:35:04] <stashbot>	 T244340: Reduce read pressure on mc* servers by adding a machine-local Memcached instance (on-host memcached) - https://phabricator.wikimedia.org/T244340
[16:38:11] <inflatador>	 James_F we are running git-sync sidecars in the dse-k8s cluster, see https://gitlab.wikimedia.org/repos/data-engineering/blunderbuss
[16:41:54] <James_F>	 mszabo: Yeah, good find; that was a pre-k8s thing for the MW appservers, whereas we're thinking of something on the orchestrator side to avoid load on the appservers, but something worth thinking about, thank you.
[16:43:07] <James_F>	 inflatador: Aha, interesting. Thank you!
[18:12:56] <Krinkle>	 Is there documentation about how to do schema changes for removed tables? It seems that's not covered by https://www.mediawiki.org/wiki/Manual:Schema_changes
[18:13:38] <Krinkle>	 That page assumes a json patch (which isn't needed for drop table), or a manual sql file (idem).
[18:14:15] <Krinkle>	 But to get it right requires quite a few changes.
[18:14:38] <Krinkle>	 Eg https://gerrit.wikimedia.org/r/c/mediawiki/core/+/1117849/2
[18:26:08] <denisse>	 James_F: We have the Thanos sidecar for metrics, but I'm not sure if it's relevant for what you're looking for.
[20:23:45] <James_F>	 denisse: Ack, will look at that anyway, thanks.