[09:32:31] pfischer: JVM Language at https://meet.google.com/zyr-jfio-wie if you're around [12:19:02] Hi all! I'd love to try uploading another wdqs JNL file from a test server to somewhere public folks can download it (gonna try archive.org this time) [12:19:18] Still okay with me doing this? and which is the right server to be doing this on today? :) [12:22:08] addshore: our test servers (wdqs1009 and wdqs1010) would be the right place. Could you coordinate with ryankemper and inflatador? They are in the process of migrating to Debian Bullseye and are doing some journal transfers already. [12:22:17] They will be around later today [12:22:26] gehel: sounds great! I'll coordinate with them :) [12:23:05] The main discovery we will make from this test is what the upload bandwidth is like going to archive.org :D and if curling a 1-2TB file directly to them will work in the first place, or fail [12:23:29] however, at least it'll be free for folks to download, rather than costing over 100 USD like when i put it on google cloud [12:24:57] addshore: we should probably expose this journal as part of the dumps, but doing it properly would require some work on our side (I think we already talked about that). [12:25:17] There is some ongoing work in modernizing the dumps infrastructure, which might help move this forward. [12:26:00] yeah, I think in the long run that'd be nice :) [12:26:05] it might be the right time to define the need a bit better and see if there is bandwidth to do it. [12:26:20] If you have time to create a phab task, at least we can get that conversation restarted [12:26:46] I'll look at writing something down once i carried out this test, then i'll have some data to include in the task too from the 2 experiements [13:16:56] addshore happy to help out! wdqs1010 is out of commission. As far as where to put the file, I recommend Cloudflare R2 as it has no egress charges. In fact, I was following your WDQS blog posts last week and already uploaded a dumps file there [13:22:28] oh nice, whats the pricing for storage? *looks it up* [13:22:56] 38usd for a TB, not bad [13:23:31] inflatador: so you put a JNL file there? [13:42:27] test test. I'm still having IRC (probably bouncer) related problems [13:51:44] pong [13:51:58] (very delayed) [13:53:31] OK, I think I fixed it...fingers crossed [13:58:01] gehel do you think we need both DSE k8s and Search Update Pipeline mtgs? I was thinking maybe drop the DSE k8s mtg [14:17:26] ebernhardson I know you mentioned getting a test service up k8s...I think we can add the ZK config to the current flink-app chart to test. Re: https://phabricator.wikimedia.org/T328675 [14:34:34] pfischer: Flink meeting on https://meet.google.com/pup-xwxi-oqw [14:51:18] gehel: sorry, I missed it, have to start my commute earlier to adapt for that meeting. Going through the notes... [14:52:28] The main question was around decision on Kafka. I'm trying to schedule s meeting on Wednesday to see if we can reach an agreement [15:02:10] ebernhardson: weekly triage: https://meet.google.com/eki-rafx-cxi [17:21:29] reindex is still going, not nearly as far along as i might have expected. It's up to cs/cy/de in each of the three clusters [17:26:02] lunch, back in ~1h [18:31:00] back [20:29:45] errand, back in ~30 [21:06:03] hmm, the README.md in deployment-charts says: The `values.yaml` files provided by any chart are defaults and their purpose is to help deployment under development clusters. Actual values used in production are not kept here.⏎ [21:06:17] Is that just unclear? My understanding is that the files in helmfiles/* are the actual prod values?