[01:31:06] Could somebody take a look at https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#WMF_Board_of_Trustees_Election_Bug? [06:35:11] <_joe_> uhm no one on clinic duty? [06:35:59] _joe_: been that since 1am Monday [06:36:04] UK time [06:42:56] <_joe_> RhinosF1: thanks, I can read ;) The problem is why, not when. I suspect it has to do with yesterday being a bank holiday in US-canada [07:56:19] so the source calendar has d.enisse overriding for X.ioNoX but it looks like the override is just wrong, so I'll fix it. [07:57:17] (and they noted yesterday that they thought they were on clinic duty this week) [08:31:04] stupid question. Why is socket.getaddrinfo('moss-be1001.eqiad.wmnet',None,flags=socket.AI_CANONNAME, type=socket.SOCK_STREAM) [in python] only returning the v4 address for that host? [08:31:19] (leaving flags and type unset doesn't change this) [08:32:30] ah, but _only running on that host_ [08:32:45] because /etc/hosts only has v4 :( [08:34:12] Is this a bug in python's socket.getaddrinfo, or a bug in how we deploy /etc/hosts ? [08:43:33] Emperor: IIUC getaddrinfo is behaving as it should, prioritizing what's in /etc/hosts.. Is there a specific issue getting v4 as default? You can always query the AAAA record to get ipv6, if needed (trying to understand the issue) [08:46:39] I'm trying to track down a misbehaviour in ceph (where it isn't starting up a service), and the cluster is generally using v6 everywhere. [08:48:29] ...so I was wondering if the lack of a v6 address might be confusing ceph [08:48:41] ah okok, no idea then [09:34:31] I guess also wondering if it's a bug in our setup too [09:37:56] if I had to guess I'd say that we defaulted to ipv4 to avoid issues like random services picking up v6 by default (maybe when v6 was introduced, we had mixed setups and v6 sometimes lead to connectivity bugS)