[06:50:32] <godog>	 greetings
[13:09:40] <hashar>	 hi, I have a fun friday question.  I got an instance for which the ephemeral disk is exposed as `sda` instead of `sdb`  and I am wondering whether that can cause troubles or if it should be fixed (and if so how :) ).
[13:09:40] <hashar>	 The instance is integration-agent-docker-1081.integration.eqiad1.wikimedia.cloud
[13:10:15] <hashar>	 (puppet failed when invoking make-instance-vg-ephem  , I reran the command manually replacing sdb by sda
[13:10:21] <hashar>	 and it seems everything works fine ™
[13:10:23] <hashar>	 ;)
[13:52:59] <andrewbogott>	 hashar: unfortunately the drive label assignment is undefined by the OS, so it's not really possible to control which is which. It might be that puppet could be smarter about detecting the right one...
[13:53:10] <andrewbogott>	 You might want to try rebooting and see if they flip back the other way :/
[13:59:36] <hashar>	 ah that is what I thought :]
[13:59:58] <hashar>	 thank you andrewbogott  :)
[14:00:13] <hashar>	 and apparnetly the make-instance-vg-ephem is simply given /dev/sdb
[14:00:19] <hashar>	 not sure whether it is hardcoded, I imagine it is
[14:04:04] <andrewbogott>	 likely is -- as you see the 2nd drive is /usually/ sdb :/
[14:10:14] <hashar>	 thanks andrew :]
[15:16:57] <hashar>	 I have filed it as https://phabricator.wikimedia.org/T422258
[15:17:06] <hashar>	 I guess I will look at it on Tuesday morning