[06:08:17] <elukey>	 hello folks
[06:08:26] <elukey>	 afaics we have two transport links down
[06:08:45] <elukey>	 - esams <-> eqiad (Lumen), that should be tracked by https://phabricator.wikimedia.org/T287469 (just added a comment)
[06:09:00] <elukey>	 - eqiad <-> codfw (Telia), that should be scheduled maintenance
[06:09:27] <elukey>	 nothing really on fire but this Lumen situation is sad :(
[06:17:03] <volans>	 indeed
[06:37:33] <elukey>	 I am running the dns netbox cookbook to clear an alarm, there was an asset record to remove for ms-be1020 (offline - https://netbox.wikimedia.org/dcim/devices/1653/)
[06:39:08] <volans>	 thanks! I was digging a script and didn't yet check it 
[06:58:18] <elukey>	 TIL https://github.com/mediawiki-utilities/python-mwapi (ores uses it)
[07:07:26] <topranks>	 Luca, thanks for the updates on cct status.  Can’t really do much unfortunately. 
[07:08:04] <topranks>	 Lumen outage for almost 10 days is far from ideal! 
[07:10:22] <elukey>	 topranks: yes yes I know :(
[15:01:44] <Krinkle>	 godog: RE T257093 - so you can shard within a swift container even on something that isn't the left-most portion of the path? e.g. split score/ up as "score/v*/a/*" and "score/v*/b/*" etc?
[15:01:45] <stashbot>	 T257093: Add cache key information to metadata json - https://phabricator.wikimedia.org/T257093
[16:04:28] <godog>	 Krinkle: in practice yes, within a container the object name is a collection of bytes to swift
[16:05:19] <godog>	 Krinkle: i.e. / doesn't mean anything, prefixes do obviously depending on how you are planning to iterate on the container
[16:16:17] <godog>	 Krinkle: gotta go now, we can followup here and on task
[20:49:17] <Krinkle>	 godog: ack, I understand that at rest the name is just metadata, but I guess my question is more on where we materialize/map to containers and shards. To some extent (up until the wiki ID $site/$lang mostly) this is wired from MediaWiki's side with the swift rewrite rules. But deeper within that, e.g. if you were to shard score into 16 segments from 0-f or some such, afaik that is not something MediaWiki (and thus me) is aware of where 
[20:49:17] <Krinkle>	 and how that happens. I'm imagineing there's a config file of sorts where in perhaps a prefix pattern or regexp of sorts is used to dynamically create these buckets perhaps. If that part of the system requires a clean prefix, then that would be a reason to go for score/0/01/ instead of score/cacheversion/0/01 as otherwise it'd be pretty difficult to map that /0/.