[18:56:50] Amir1: What's the recommended way to debug mainstash now that x2 is gone? [18:57:07] I'm looking at updating https://wikitech.wikimedia.org/wiki/Debugging_in_production#Debugging_databases [18:57:27] it seems while ms1 is in dbctl, it is not in $wgLBFactory so not reachable as --cluster ms1 [19:02:46] I wonder if anything bad would happen if we let those proceed to wgLBFactory for debugging only. [19:03:15] the etcd.php loop `continue`'s them which seems to only serve to make debugging break. [19:03:28] (I havnet' tested that removing it makes everythign works, that's an assumption) [20:18:30] Krinkle: in the new architecture. Ms sections are exactly like parsercache hosts. We probably need a general way to allow connections from mysql.php or something like that that would work for both pc and ms [20:21:23] They stopped looking like a external cluster and more like a parsercache one [20:22:32] yeah, I've updated the page to rephrase it in that sense. Its not great for UX but at least it's possible to figure out [20:23:11] What do you think about allowing them as ~unused entries in $wgLBFactory? Would that pose an isuse/risk? [20:23:28] The idea woudl be to let it work as --cluster ms1 --cluster pc1 [20:23:53] I need to think about it. Right now [20:24:03] Nothing crazy comes to mind [20:24:11] But can't be sure [20:30:36] * Krinkle adds docs for connecting to x3 [22:09:32] Amir1: test coverage https://gerrit.wikimedia.org/r/1182228 and proposal in https://gerrit.wikimedia.org/r/1182229