[00:02:00] <inflatador>	 ebernhardson gotcha, thanks
[00:04:07] <inflatador>	 Wow, it's 6 PM my time all of a sudden. See y'all tomorrow!
[00:04:48] <ebernhardson>	 see ya!
[00:08:13] <inflatador>	 latersville
[00:13:59] <bd808>	 ebernhardson: y'all are going to do 7.10 rather than going right OpenSearch?
[00:14:07] <bd808>	 *right to
[00:14:57] <ebernhardson>	 bd808: indeed, expecting open search to be more painful as we make use of quite a few plugins
[00:15:12] <bd808>	 *nod*
[00:15:48] <ebernhardson>	 there is some question of what is the appropriate system to run search, because despite the name opensearch seems to have trippled down on logs compared to elastic, who were already pretty log heavy :)
[00:16:05] <ebernhardson>	 (although i haven't looked exceptionally deeply)
[00:16:08] <bd808>	 actually looking at the OpenSearch docs they recommend rolling update to 7.10.2 before moving to OpenSearch
[00:17:17] <bd808>	 yeah, I could see OpenSearch being skewed in favor of ELK use cases depending on who does the work of driving the roadmap
[00:18:59] <ebernhardson>	 i'm sure we will end up on opensearch at some point, i can't really see other options that make sense given reality.  but who knows when :)
[00:19:33] <bd808>	 right after a CVE comes out that requires going past the FOSS versions ;)
[00:20:24] <ebernhardson>	 yea thats one way :)
[00:21:02] * bd808 is mostly interested because Toolhub is a stowaway on the chi cluster
[00:22:49] <ebernhardson>	 we've talked a little, but i imagine you're safe for at least a year or so
[00:22:57] <ebernhardson>	 after the 7.10 upgrade
[00:24:45] <bd808>	 I need to watch 7.x first because of the versioned python client libs I'm using. I think they all freak out when a 6.x client tries to talk to a 7.x server
[00:25:16] * bd808 should probably start a task to POC 7.10 in the dev environment
[00:28:27] <ebernhardson>	 hh, yes that will come much sooner, es 6.8 and then 7.10 are the next project after wcqs is released this month. Previous upgrades of the same sort took maybe 3-5 months
[08:26:47] <zpapierski>	 ejoseph: let me know once you're available
[08:32:27] <ejoseph>	 zpapierski: good morning
[08:32:44] <ejoseph>	 I'll be available in 10 mins
[08:33:03] <zpapierski>	 great, let's sync then
[08:52:32] <zpapierski>	 ejoseph: I'm going to brew myself a coffee and I'm back - meet.google.com/afu-jhht-grf
[09:31:38] <gehel>	 errand, back in a bit
[10:36:13] <zpapierski>	 errand and lunch probably later
[10:39:07] <gehel>	 lunch
[11:09:29] <dcausse>	 lunch
[13:54:39] <zpapierski>	 dcausse: try as I might, I can't understand how the index mapping is constructed for the tests like TokenCountRouterBuilderESTest . Migrating to 7.10 makes them fail with "MapperParsingException[No handler for type [geo_shape] declared on field [mapped_geo_shape]]" I'm guessing the default mapping handling was changed somehow, but can't get to see how - any ideas?
[13:56:19] <dcausse>	 zpapierski: this error is fixed by adding the TestGeo plugin (from the method getPlugins)
[13:56:32] <dcausse>	 lemme find the real name
[13:56:52] <zpapierski>	 TestGeoShapeFieldMapperPlugin
[13:56:52] <zpapierski>	 ?
[13:57:06] <dcausse>	 yes that's it
[13:57:10] <zpapierski>	 just the mapper I guess?
[13:57:14] <zpapierski>	 mappers.put("geo_shape", new org.elasticsearch.index.mapper.GeoShapeFieldMapper.TypeParser()); 
[13:57:27] <dcausse>	 no just add the plugin
[13:57:36] <zpapierski>	 ok
[13:58:49] <dcausse>	 https://gerrit.wikimedia.org/r/c/search/extra/+/737664/4..5/extra/src/test/java/org/wikimedia/search/extra/router/DegradedRouterBuilderESTest.java#50
[13:58:53] <zpapierski>	 ok, much better - some test fail, but most of them passes now, thanks!
[13:59:01] <dcausse>	 yw!
[13:59:15] <zpapierski>	 btw - it's deprecated,why?
[13:59:54] <dcausse>	 no clue it's some BC testing code but I'm not sure why it got to pollute everything like that
[14:00:24] <zpapierski>	 ah, it's decompiled code, maybe source will shed some light
[14:00:49] <dcausse>	 I think they used to generate a standard mapping for all the tests and got stuck when deprecating stuff so that's why they created this Test plugin?
[14:01:38] <dcausse>	 it's possible that using their standard gradle build it's getting imported automatically
[14:01:51] <zpapierski>	 "Some tests depend on the GeoShapeFieldMapper. This mapper is registered in the spatial-extras module, but used in many integration tests in server code. The goal is to migrate all of the spatial/geo pieces to the spatial-extras module such that no tests in server depend on this test plugin"
[14:01:51] <dcausse>	 s/imported/discovered&loaded/
[14:01:54] <zpapierski>	 you might be correct
[14:03:18] <dcausse>	 other classic problem I saw is that now elastic is more strict when dealing with term queries on undeclared fields
[14:03:33] <dcausse>	 they get rewritten to MatchNone
[14:04:03] <dcausse>	 so that's why some tests now requires defining the field they work on
[14:04:21] <zpapierski>	 this testgeoshape plugin generates a huge wall of deprecation warnings now :/
[14:04:37] <dcausse>	 :/
[14:04:57] <zpapierski>	 yeah, not much we can do, unless we want to disable those
[14:05:11] <zpapierski>	 maybe I can disable this specific warning somehow
[14:05:28] <dcausse>	 deprecation from elastic?
[14:09:17] <inflatador>	 greetings!
[14:09:34] <dcausse>	 hey!
[15:00:53] <zpapierski>	 inflatador: let's get to our 1/1 meet actually - if anyone wants to join for a "get to know" https://meet.google.com/cex-ouqe-ybj
[15:24:11] <gehel>	 inflatador: looks like we have a disk issue on elastic2051, having a look at https://icinga.wikimedia.org/icinga/
[15:25:28] <gehel>	 and that's one of our master eligible, we need to switch the master
[15:33:55] <gehel>	 inflatador: we can have a look at that server during our pairing session later today
[15:36:21] <zpapierski>	 ejoseph: I pushed the changes for that geo stuff, we can move on further
[15:37:22] <inflatador>	 gehel sounds good, just had a nice discussion with Z
[15:38:15] <gehel>	 inflatador: or if you are around now, we can have a look
[15:38:51] <inflatador>	 gehel sure, whenever you want
[15:39:33] <gehel>	 inflatador: meet.google.com/oam-gbmu-eef
[16:03:16] <ebernhardson>	 \o
[16:08:34] <zpapierski>	 o/
[16:32:21] <zpapierski>	 dcausse: can you help a bit with that failing rewrites with match_none?
[16:32:35] <zpapierski>	 I'm not sure which way we need to go with fixing that
[16:54:23] <gehel>	 inflatador: now that we've moved the master eligible node, we still need to get a new disk for that server. Let's connect during our pairing session this evening
[16:57:39] <gehel>	 Dinner, back later
[16:59:34] <inflatador>	 gehel ACK, taking a workout break, back in ~30
[17:51:47] <inflatador>	 back, but cable guy is on his way, so may be right back out
[18:10:50] <inflatador>	 aaand back for real
[18:18:46] <inflatador>	 The cable guy was very confused by my MoCA (ethernet to coax) adapters
[18:19:14] <ebernhardson>	 heh, i don't think i would recognize those either :)
[18:24:12] <inflatador>	 It's always confusing, because I have multiple cable lines but no "cable" service (just Internet)
[18:25:10] <inflatador>	 I have an old house with coax taps in every room, it's much faster than Wifi for my wife's office (she is WFH also)
[18:27:41] <ebernhardson>	 makes sense, had to google what they are but reusing any wire thats already run everywhere is a huge win :) 
[19:03:46] <inflatador>	 lunch, back in ~30
[19:26:28] <gehel>	 inflatador: I might be 5' late for our meeting, need to get Oscar to bed first
[19:26:57] <inflatador>	 gehel perfect, gives me a little more time to stuff my face ;P
[20:35:02] <gehel>	 good night!
[20:44:57] <ebernhardson>	 lunch
[21:29:17] <ebernhardson>	 back
[22:06:18] <inflatador>	 ebernhardson  or anyone else, LMK if you have time/feel comfortable giving a +1 here: https://gerrit.wikimedia.org/r/c/operations/puppet/+/752005
[22:15:34] <ebernhardson>	 inflatador: lgtm
[22:17:02] <inflatador>	 Thanks, let the +2 begin!
[23:07:34] <inflatador>	 Out slightly early to make pizza, see y'all tomorrow!
[23:10:44] <hare>	 🍕