=== xnox|sessions is now known as xnox [08:14] hi all [08:14] sorry to bother again for a quick one. [08:14] rax asked me to try changing the flavor (trying to bootstrap a new controller) [08:15] list of "flavors" name contains spaces character [08:16] so if I use `instance-type=4GB Standard Instance` that does not work... as juju client will consider `Standard` and `Instance` as parameters... Of course I tried to escape space character but with mo luck [08:16] does anyone has some ideas for that? [08:32] flxfoo: AFAICT from looking at the constraints parsing code, we currently split the constraint list on space so this (or any other key-value constraint combination with spaces for that matter) will not work. [08:33] flxfoo: Can you create a bug for this? [08:36] achilleasa: ok, where do I do that? [08:37] flxfoo: you can create a new one here: https://bugs.launchpad.net/juju/+filebug [08:46] ok I created a bug [08:46] is there anything I can do meanwhile? [08:51] flxfoo: are the instance types aliased or something that you can rename? Btw, what provider are you using? [08:55] achilleasa: provider is rackspace [08:55] flxfoo: if it's a blocker and you don't mind compiling juju I can push a branch with a quick patch which will allow you to use something like '\ ' to escape spaces till we fix the bug [08:56] achilleasa: it looks like juju by default uses flavorId = 5, is there a parameter to pass to juju to update that? [08:58] achilleasa: the blocking situation is that, rax deleted an image which the current ctrl would use to create containers, so the platform is pretty unmanageable. On top of that, the current ctrl has been bootstraped using a custom simplestream because the image list provided by rackspace is empty. [08:59] achilleasa: so I though to bootstrap a new ctrl with a new image id (that rax provided me) in order to migrate the model, so at least I can do something. [09:01] flxfoo: if you pass "general1-4" as the instance type does it work? [09:03] flxfoo: oh wait, according to https://developer.rackspace.com/docs/cloud-servers/v2/general-api-info/flavors/ ID 5 corresponds to the one you are trying to use (4gb standard instance)? [09:03] achilleasa: tried it nope, it returns the list of available flavors which the ones with spaces [09:03] achilleasa: yeah (my bad) I should have tried another one, but the issue remains with the spaces... [09:04] achilleasa: this is rax support answer: It looks like the issue with the command is that you're attempting to create a boot from volume Standard flavor server. Standard flavor servers don't have the ability to use a CBS volume as the system disk. You'll either want to try the command without the flags for BlockDeviceMappings or choose a different server flavor such as General Purpose: [09:07] I see. As mentioned, I can push a branch with a patch to get you unblocked until the bug is fixed [09:09] achilleasa: thanks [09:11] achilleasa: just a silly question, I suppose I can not create an instance on rackspace and tell juju to use it to setup the controller? [09:13] flxfoo: you probably can if you set it up as a manual cloud (https://juju.is/docs/manual-cloud). Don't have access to rackspace so I can't verify this though... [09:14] achilleasa: thanks for your time [09:36] flxfoo: if you checkout (https://github.com/achilleasa/juju/tree/2.7-support-escaped-spaces-in-constraints) in $GOPATH/juju/juju and run 'make install' you will have a juju cli that supports constraints with slash-escaped spaces [09:51] flxfoo: FYI, there is another related bug for zones with spaces and the fix for that (currently in review) will also take care of the issue you are having without the need to manually escape spaces [09:59] https://github.com/juju/juju/pull/11506 fix for intermittent test failure [10:00] thumper, I'll check that one [10:00] thumper, any QA steps, other than hammer the test with -race? [10:36] manadart, https://github.com/juju/juju/pull/11412#issuecomment-620525235 [10:41] stickupkid_: nah [10:41] stickupkid_: it took 10 - 20 minutes to replicate on my machine [10:41] thumper, I tried whilst waiting for aws to bootstrap and couldn't replicate [10:41] yeah, that's cause I fixed it :) [10:42] before hand [10:42] haha [10:42] ah... yeah, it took me a long time === arif-ali_ is now known as arif-ali [11:32] anybody tried rebuilding dependencies for 2.7 branch [11:59] juju ssh 0 -m controller <- why does that not work and this does -> juju ssh -m controller 0 [12:00] we totally we should have done juju ssh 0 -m controller -- bash [13:08] are storage pools model-level entities or cloud-level ones? if I run create-storage-pool in model a, will it be available from other models? [14:05] timClicks: they seem to be model-level from my experience [14:22] stickupkid: ran qa on 11412 with new code - success, though what is “ERROR no available address(es)” from at the end? [14:22] hml, yeah, that's fine it's trying to get the engine reports [14:22] hml, it's optional at best [14:23] stickupkid: rgr [14:23] udpate my results in the pr [15:54] axino, looking at your charm code, I don't see install or start being handled, just config-changed/leader-elected/upgrade-charm [15:54] jam: yeah but the "hooks" directory has a "start" symlink to charm.py [15:55] axino, gotcha. I would also create an install symlink. You can have both and that should handle 2.7.3 and 2.8 [15:55] jam: indeed, indeed. This is what the official operator doc says currently [15:55] I think I'll just resolve this bug [15:57] done [15:57] sorry for the noise o/ [16:04] axino, np. I think it is very good to understand changes between releases