[11:26:20] <Krinkle>	 AaronSchulz: TimStarling I'm wondering what's easier: modify wmf-config to reduce x2 array to [0], or to introduce new modes in Rdbms for ignoring replicas, lag checks, CP etc. I assume we already don't do these things for setups that have no replicas right? Eg parsercache, LBSimple etc
[11:26:36] <Krinkle>	 Exposing the replicas seems like a risk long term
[12:18:17] <TimStarling>	 yeah, probably easier to reduce the array
[14:34:40] <Krinkle>	 > Duplicate get(): "global:Wikimedia\Rdbms\ChronologyProtector:5492c7bb49467184f4f3a09b4f4e21b4:v2" fetched 2 times
[14:34:57] <Krinkle>	 logged on mwdebug1002 when viewing a diff on enwiki
[14:42:20] <Krinkle>	 so not just one redundant fetch but two
[21:19:11] <AaronSchulz>	 Krinkle: 3 fetches? odd
[21:19:24] <AaronSchulz>	 reqId?
[21:23:29] <Krinkle>	 2
[21:23:52] <Krinkle>	 there's a few in logstash from today under mwdebug1002
[21:23:57] <Krinkle>	 Can share in meeting later
[21:24:29] <Krinkle>	 I consider both redundant since it's web request that needs zero
[21:24:47] <Krinkle>	 But one is known indeed, we have a task for that
[21:25:07] <Krinkle>	 Two of those patches are ready for re-review :)
[21:25:35] <Krinkle>	 https://phabricator.wikimedia.org/T314434
[21:25:47] <Krinkle>	 https://gerrit.wikimedia.org/r/q/bug:T314434
[21:25:48] <stashbot>	 T314434: Avoid ChronologyProtector queries on majory of pageviews that have no recent positions - https://phabricator.wikimedia.org/T314434