[06:44:39] morning [07:04:46] morning [08:48:23] dcaro: when would be the best time for you to revisit the harbor upgrade together? [08:50:23] is now a good time? [08:58:54] still looking for a review of https://gerrit.wikimedia.org/r/c/operations/puppet/+/966871 [09:00:14] dcaro: yep, we can use the collab meet link [09:01:56] blancadesal: coming [13:36:01] Folks I'm doing a similar change in eqiad to the one I did in codfw yesterday [13:36:50] network reconfig - means I'll be taking down uplink from one cloudsw to reconfigure, then doing the other one [13:37:05] all should be fine, traffic will route via the other link/switch, I'll also be keeping pings running/test [13:37:10] but do ping me if any signs of problems [13:40:20] ack! [13:42:52] also, not sure if anyone has an opinion here [13:43:19] I really don't like us naming "cloudnet100[7-8]-dev" as "cloudnet" [13:43:52] on the network side cloudnet has a bunch of rules (see https://wikitech.wikimedia.org/wiki/Portal:Cloud_VPS/Admin/Network#cloudnet) [13:44:08] but I'm not sure how to define those rules going forward if we've two types of server called "cloudnet" [13:44:28] one set of which are dedicated hosts running the netron l3 agent, another being just normal k8s worker nodes [13:44:55] wait [13:45:16] this has come up in #dc-ops channel as I have to explain to them that the rules I wrote on what servers get what vlans don't apply to these new ones [13:45:22] could I get a +1 here? https://phabricator.wikimedia.org/T349378 [13:45:54] I agree that if those hosts are not going to be neutron service hosts, they should not be named cloudnet xd [13:46:34] we don't have to work it out today I think, but probably we should think of it [13:46:37] blancadesal: you probably want a project name without any dashes, but otherwise lgtm [13:47:01] ah yeah. object storage, right? [13:47:10] for the new "-dev" poc buildout it might be useful to have new server names, to easily distinguish what hosts are part of what setup (i.e. existing vs new openstack k8s) [13:47:44] yeah, it causes problems with that at leat [13:47:59] you will see the warning when running the cookbook now :) [13:48:02] also, what does qte mean in the project name here? do you see there being multiple catalyst admin projects? [13:48:22] the main project is called catalyst-qte [13:48:49] could call it just catalyst I guess [13:49:20] topranks: I'm not involved on the new setup effort, so I don't have any insight on the reasons why those names were chosen, but it looks to me that as you say it would be really confusing to named them cloudnets xd and probably something different, like cloudworker or cloud-k8s-worker or such [13:49:22] yeah and I don't like the team name in the project name, since generally wmf teams have a shorter lifespan than the software projects they create [13:51:00] dcaro: yeah I think those names are probably better [13:51:27] topranks: I guess that andrewbogot.t and Roo.k are the ones working on it? not sure if they can pitch in, but you might get better engagement pinging them in the task I guess [13:51:37] it's ok, we can rename them. we probably need to have a meeting at some stage to discuss the plans for new buildout, definitely some decisions on the network side that need to be made / planning [13:51:49] I left a comment on the task to clear it up for dc-ops yep [13:51:50] thanks :) [13:52:06] 👍 [13:52:25] taavi: renamed [13:57:15] blancadesal: do you want to have the admin word in the project name though? [14:50:40] I think just 'catalyst' is fine [15:28:15] * dcaro off [15:37:05] fyi. I left some ceph osds processes restarting for the upgrade to v15, will come back later to check up on them, don't be surprised :) [17:02:27] * dhinus off