[08:38:44] <atsukoito>	 dcausse: hiya!
[08:38:52] <dcausse>	 atsukoito: hi!
[08:45:15] <dcausse>	 atsukoito: looking at search stats on the ttmserver indices in the cirrussearch clusters, if they show 0 usage in the next couple hours, any objections me deleting (or closing) these indices there?
[09:23:51] <dcausse>	 sigh... just realized that transfer_to_es weekly failed 3 weeks ago... just unblocked it but that means 3 weekly runs in one go :/
[11:03:00] <atsukoito>	 dcausse: no objections for deleting it, I was looking around the graphs and didn't find anything worriesome in particular
[11:37:02] <atsukoito>	 it works much faster! tho i'm going to reduce the memory allocation back to our defaults since we don't do much of the requests
[12:04:15] <dcausse>	 ack, closing them for now and will delete tomorrow
[12:53:29] <ebernhardson>	 \o
[12:53:34] <pfischer>	 o/
[12:56:29] <dcausse>	 o/
[13:08:21] <inflatador>	 <o/
[13:48:24] <inflatador>	 dcausse might be slightly late for pairing, have to drop off my son at camp
[13:48:57] <dcausse>	 inflatador: np! ping me when you're around and if you have things to discuss 
[14:02:35] <inflatador>	 dcausse I'm back, but don't have much to discuss. It looks like we have to sort out some merging behavior in Puppet before we can do the OS 1->2 upgrade, ref https://gerrit.wikimedia.org/r/c/operations/puppet/+/1305321
[14:02:58] <inflatador>	 If you don't have anything else to discuss I'm fine w/skipping
[14:23:32] <dcausse>	 inflatador: I don't have much sorry, ended up being distracted calling my health care provider anyways... :)
[14:33:34] <inflatador>	 np
[14:59:12] <pfischer>	 I’ll be 15’ late for our Wednesday meeting due to a conflicting meeting.
[15:43:27] <ebernhardson>	 https://cirrus-lens.toolforge.org/
[16:32:05] <dcausse>	 heading out
[18:07:33] <pfischer>	 ebernhardson: Could you please have a look at the following jupyter/pyspark script? I tried to get the rate of (long) autocomplete requests coming from Android. Motivation: One of Tais’ designs would require interleaving semantic results with autocomplete results. So I wanted to know if this would potentially overwhelm OS. However, the numbers appear low:
[18:07:45] <pfischer>	 Mean rate: 35797.36 requests/hour ÷ 3600 = about 9.94 requests/second.
[18:07:45] <pfischer>	 For the range: min 24432/hr ≈ 6.79/sec, max 48794/hr ≈ 13.55/sec. The stddev of 6425.83/hr works out to ≈ 1.78/sec.
[18:07:52] <pfischer>	 https://phabricator.wikimedia.org/P94446
[18:24:14] <ebernhardson>	 pfischer: sure looking
[18:25:35] <ebernhardson>	 pfischer: seems reasonable. 
[18:59:07] <inflatador>	 ebernhardson are you going to OpenSearch Con again this year? Looks like it's in your neck of the woods
[18:59:39] <inflatador>	 "Gather with hundreds of your OpenSearch community peers September 22–24 in San Jose"
[19:05:57] <ebernhardson>	 inflatador: hmm, hadn't thought about it. I suppose i could
[19:08:26] <inflatador>	 No worries either way, just curious since I got an email about it
[20:36:44] <pfischer>	 ebernhardson: thanks!