[06:32:06] XioNoX, paravoid_ o/ I merged the pending turnilo changes, I tested briefly and it looks ok, but please check as well! [06:34:35] elukey: lgtm! [09:36:08] [heads up] if you need to merge any patches against the operations/dns repository today please get in touch with me first [09:38:22] elukey: looks good, thanks! [09:41:36] XioNoX: note that this patch merges eqdfw/codfw + esams/knams and calls it a "region" -- which is not what you wanted in the other patch [09:41:50] I know, but it's temporary, right? :) [09:41:53] it preceded your comments, in my defense :) [09:41:57] yeah [09:42:08] but... I think it's a useful way to view the data [09:42:29] like peer AS dst: 1299, split by: region [09:43:23] paravoid: we could keep the "sites" in Druid, and then have a "region" in turnilo, but that might be overkill [09:43:30] heh [09:44:09] my main concern is to keep a manual list of exporter IP -> site/region [09:44:30] what do we gain by having knams/eqdfw separately? internal netflow? [09:50:03] paravoid: what I'm hoping is to be able to re-use the definition from https://github.com/wikimedia/puppet/blob/production/modules/network/data/data.yaml#L32 [09:50:33] for which part? [09:51:22] for https://gerrit.wikimedia.org/r/c/analytics/refinery/source/+/634328 [09:51:47] for which dimension? [09:52:06] site/region [09:52:47] so knams/eqdfw would _not_ be separate then? [09:52:51] correct [09:53:00] ah, I had understood the opposite [09:53:10] the issue would be eqord part of eqiad [09:53:19] right [09:53:42] what's the best tradeoff? I don't know [09:53:43] yeah that really needs to be separate in turnilo [09:54:37] I also wonder how we're going to use hiera data in (presumably compiled) java code :P [09:55:04] dunno [09:55:39] if we have to define yet another mapping, then whatever works for me :) [09:56:26] but for our turnilo usage region make more sens than site indeed [10:18:19] XioNoX, elukey, cdanis: the retainMissingValue misconfig fix revealed a bunch of unmapped (unnamed) TCP flags that were hidden before [10:18:36] nice [10:18:50] 7 15 23 27 30 31 35 37 42 43 44 45 50 51 52 54 56 57 58 59 [10:18:51] as far as I can [10:18:54] tell :) [10:20:15] I wrote a script to try to generate them all but I guess I missed some [10:20:29] it's a binary match [10:20:49] I wonder if we can write a piece of javascript to parse the bitfield [10:21:01] in theory these functions can be javascript [10:22:12] Watch Faidon nerdsniping himself :) [10:22:34] :P [10:22:39] if you have a script ready, then maybe not :) [10:23:19] no I don't :( [11:06:36] XioNoX: https://phabricator.wikimedia.org/P13050 [11:07:57] :) [11:08:01] nice! [11:09:31] pushed a couple of changes your way, lunch now [13:08:43] paravoid: all turnilo configs deployet [13:08:46] *deployed [13:08:57] :D thx [13:15:58] thanks! [14:02:24] cdanis: I can't ask Faidon any fee but for you and Arzhel it is 5 euros for each config change [14:02:28] :D [14:03:02] (we can also ask 10 to Arzhel directly, ok for me as well) [14:03:42] :) [14:03:53] * elukey sends wikilove to XioNoX [14:04:18] wow, I would have paid you the cost of one macchiato, but the price or two seems exorbitant [14:04:23] s/or/of/ [14:06:41] ahahahha [14:07:55] I'm wondering how much I'll start billing elukey for ACL changes [14:08:13] jokes aside, really glad that turnilo is improving, hopefully we can also get short urls soon-ish (dan already upstreamed a parch) [14:08:16] *patch [14:08:39] XioNoX: I think you should start bill me for any error that I made in the configs, to become truly rich [14:08:46] *billing [14:09:05] yeah I saw that upstream PR [14:09:15] hopefully they will release the new version soon :) [14:09:42] small changes that make our life so much better [14:21:43] <_joe_> cdanis: a macchiato in italy is ~ 0.8 EUR [14:21:59] _joe_: in the touristy parts of Rome it's 1.25 ;) [14:23:56] <_joe_> cdanis: that's the tourist surcharge [14:24:15] I ordered in horrible Italian, I was trying 🤷 [14:24:36] so that's teh horrible italian surcharge :D [14:24:44] sounds about right [14:51:10] <_joe_> volans: nah in Rome's center? they surcharge me and you too :P [14:55:41] bblack: ema, would you have some time for a short meeting next week regarding my work on architecturing media backups? [14:56:05] XioNoX: I would also want some with you regarding networking side of things, like I did for database backups [14:56:52] sure [14:57:25] I know you may think you are the wrong people re:traffic, but I think you will have better understanding of rates and overloads of file uploads and downloads [14:57:42] (I have, of course, already talked to filippo and others) [14:57:57] sure yeah [14:58:09] if you have doc I can read ahead of time the better [14:58:12] for XioNoX is more of capacity planning [14:58:42] Yeah, I will share with both what we have so far, but the ongoing design is in early planning [15:06:49] XioNoX: I am sorry, I don't know which is a good hour band for you? [15:07:45] jynus: anything not booked in my calendar, 8am to 9pm France time [15:07:55] thanks [15:08:49] can I upload a standalone file on phab or wt? [15:09:15] ah [15:09:18] Permitted file types: tiff, tif, png, gif, jpg, jpeg, webp, xcf, pdf, mid, ogg, ogv, svg, djvu. [15:09:25] for wikitech [15:09:58] XioNoX: https://phabricator.wikimedia.org/file/ [15:10:03] XioNoX: you can ACL it as well if you need [15:10:09] I use phab files all the time [15:11:15] beautiful https://phabricator.wikimedia.org/F32411089 [16:36:56] ema: please merge my -> Buster patch when you merge your change, thanks [17:26:59] [HEADS-UP] We're about to migrate eqiad DNS records to the Netbox-generated ones, please do not merge DNS patches for the next ~30 minutes. No impact is expected. [18:49:44] the above is all done now and you can resume normal operations in the operations/dns repository, thanks for you patience :)