[12:49:06] dhinus: I'm looking at those pending domains now. It's a familiar issue but usually it's an all or nothing thing, I've never seen it work for some records and not for others... [12:49:15] and the new records I created yesterday worked fine [13:26:15] ok, well, at least your new records are working now [13:33:36] andrewbogott: thanks, I was also thinking of restarting openstack based on https://wikitech.wikimedia.org/wiki/Portal:Cloud_VPS/Admin/DNS/Designate#If_things_are_misbehaving [13:33:46] but I thought I'd wait for you to have a look first [13:34:09] I should've taken my own advice and started with that [13:34:36] PENDING means that designate has accepted the change but that that pdns hasn't. [13:35:04] And in the pdns logs I could see pdns getting an update notification and then checking the serial and saying, Oh, that's the serial we already have, and skipping the xfr request [13:35:27] then even after I updated the serial in the database, pdns kept getting the same (old) serial from designate [13:35:52] at which point I decided that designate was just stuck or had a bad cache or something [18:24:14] dcaro: you are correct the interface operational status isn't making it's way into prometheus from gnmi [18:24:19] this is because the YANG interfaces model returns a string for the state ("up", "down" and a few others), and Prometheus only allows a number to be stored as a metric value [18:25:06] we hit something similar with the bgp operational state and had to add some processing of the values in gnmi to export as we need [18:25:07] https://w.wiki/D4V9 [18:25:15] we can do something similar here I'll prep a patch