[00:05] wallyworld: part 1 - https://github.com/juju/juju/pull/8828 (the simplest...) [00:05] anastasiamac_: will look as soon as i finish talking to vino [00:05] nws, thnx! [00:34] anastasiamac_: lgtm with a question [01:45] wallyworld: part 2 - api consolidation for cli and worker access - https://github.com/juju/juju/pull/8829 (largely mechanical....) [01:46] ok [02:02] wallyworld: query, the resource-get command currently returns, well prints, the file path of the named resource. For docker stuff that doesn't make sense, should we update expectations so it either: 1) prints some json/yaml with the needed details (image, secret) or write a json/yaml file and return the path to that? [02:03] veebers: it prints the file path brecause it has streamed the content to that location. i think for other hook commands like config-get we print to stdout [02:03] i'll check [02:03] wallyworld: right sorry it does print to stdout [02:03] it's the content of that I'm querying [02:04] currently resource-get only prints the filepath of the resource (after it's been downloaded/cached/whatever) but for the docker bits we don't want a filepath, we want more [02:04] the content for a docker inmage resource will be the result of the charmrepo api call [02:05] we print a filepath for file resources because that fits the semantic [02:07] wallyworld: ok, so the resource-get hook command will change from just ever printing out a file path, to printing out whatever is right for the resource type. filepath for filetype, json for docker-image type [02:26] veebers: correct [02:27] although i think the output format should be either json or yaml [02:27] depending on what the charm asks for [02:27] anastasiamac_: review done, 1 thing to fix i think [02:27] * anastasiamac_ looking [02:28] wallyworld: ack, so optional arg for format (for when it's not just a straight string of 'filepath') [02:28] yup [02:30] wallyworld: i disagree about auth. it is the saem auth is for model migration (and upgrader, I think) [02:31] hmmm, that's wrong then [02:31] auth need sto check for correct callers [02:31] at the very least IsAuthClient() || IsAuthAgent() etc [02:32] which model migration facade? [02:32] yeah, i can do that [02:33] * anastasiamac_ digs up the magnifying glass.. [02:33] you'd want AuthController() || AuthClient() [02:34] those other cases just checking for != nil should be fixed [02:36] not authcontroller... it will be a model surely?... and we do not have an equivalent.. [02:37] that's why all model-scoped api's auth do that [02:42] i guess, i could check if it's any agent (machine|unit|application) at the very least?... i'll update a PR [02:47] wallyworld: apparently, i was talking nonsense - there is no precedence for just a nil check :D [02:47] whew, thought i was going crazy [02:47] \o/ not yet... and hopefully, not from me :D [02:48] it's AuthClient() || AuthXXXX() [02:48] whatever is the agent that makes the call [02:49] well, it's authclient() for one of the api (the one that cli will use, client one) [02:49] and a 3 agent check for the other... [02:49] i've pushed the changes [02:58] wallyworld: I have pushed a commit addressing ur review comments. [02:58] looking [03:05] vino: just need one more test for truncation as there's a coding error [03:05] test would have picked it up [03:06] ok. [04:44] babbageclunk: did you have some emacs bits so you could like all definitions of a function declared in an interface? [04:45] veebers: I use go-guru for that [04:45] babbageclunk: would you know what the command is? [04:45] veebers: go-guru-implements [04:45] babbageclunk: ack, cheers! [04:46] babbageclunk: ah, that's for a type to list what interfaces it implements right? Can you do the other way around? [04:47] I want to find places where a func signature is implemented [04:47] If you point to a method in an interface definition, it will list all of the method defs that implement that interface method. [04:47] oh wait, no I think that's done it ^_^ [04:47] B) [04:48] babbageclunk: ack, yeah I misread something. Thanks! that's really helpful [04:48] yeah, it's great - I use it pretty frequently [04:49] wallyworld: pushed a commit [05:00] wallyworld: plz ping me once u have reviewed. I can merge and start the next phase [05:11] vino: looking [05:13] just a small correction i am doing now [05:13] vino: omit empty bit is not quite right [05:13] `json:"charm-vesion,omitempty" yaml:"charm-version,omitempty"` [05:14] yes [05:14] thats wat [05:14] ok :-) [05:14] i am correcting that now [05:14] gimme few mins. I found it in CI [05:20] :q [05:39] KingJ: what does `ip route get ` tell you? the correct interface/subnet? or a wrong one? [06:03] wallywaorld: fixed the status charm-version issues [06:03] pushed the commit [06:09] wallyworld: [06:09] looking [06:12] vino: looks good to land [06:13] sure. [06:13] thank u [06:24] wallyworld: Can we discuss abt the next phase ? whenever u have time. please ping me. [06:24] vino: ok, now? [06:24] sure. [06:24] standup HO ? [06:24] yep [06:46] wallyworld: i didnt git add the verison file created in testcharms [06:46] thats the issue [06:47] ah ok [06:47] i will commit that and squash [06:47] thank u [06:47] :) [06:53] kelvin: have to run, had a quick look at charmm pr. mostly ok a few small fixes. will check again later === frankban|afk is now known as frankban [08:03] wallyworld, thx, just got the keys and back home. looking on it now [09:04] Speaking of charm versions, anyone recall when $CHARM_DIR/.juju_charm became a thing? [12:15] suddenly every hook gives an error on "subprocess.CalledProcessError: Command '['config-get', '--all', '--format=json']' returned non-zero exit status 1" Anyone have occurred a similar thing? Gonna restart machines and charms but might as well ask [12:16] + juju status gives no error it's when I run a hook in debug-hooks [12:41] TheAbsentOne: hmm, are all the agents up and in the green? [12:52] at that moment it was like it lost connection to the agents but status showed everything green yeah rick_h_ [12:53] rick_h_: I'm having the same problem now when spinning up a new charm/machine :s I'll show error gimme a sec [12:53] TheAbsentOne: gotcha, yea I know it pings every so often and it needs to miss X pings before it shows down [12:53] TheAbsentOne: kk [12:54] "connection is shut down" [12:54] https://pastebin.com/4zDsica3 [12:55] I'm having a feeling it's not my charm but rather something else :/ [12:57] rick_h_: the error is also "insta" so I don't think he is running anything [12:57] TheAbsentOne: yea, so the agent being down means that it can't ping the API server and get the details about the config-get information [12:58] TheAbsentOne: might have to restart the agent, check the list of services running for juju related ones [12:58] and what agent is this? The unit agent on the charm? [12:58] How do I check that? [13:00] rick_h_: I got a juju status where I saw agents being lost (both to new/recent charms as to charms that are already running a longer time), but they are green again [13:00] TheAbsentOne: sec, otp. [13:01] TheAbsentOne: oh hmm, maybe the agents came back then [13:01] TheAbsentOne: and you just need to juju resolved the units [13:02] but inside debug-hooks I still get the same error though rick_h [13:04] is there a log I can consult for more information? show-status-log doesn't give me anything [13:09] I think the problem lies with the vmware-cluster, sometimes juju status takes a while and I catched 2 machines being down for a few secs. Weird stuff [13:10] rick_h_: is it correct that if the agent is lost while being in a debug-hooks he won't fix the connection and it is lost? [13:11] TheAbsentOne: doubtful? I'm not 100% sure on the debug-hooks if it can help with that or not [13:42] got it working now rick_h_ think the problem is not caused by me but lies deeper int he infra where I don't have access to, thanks for the help though! [13:54] TheAbsentOne: ah ok cool. Glad you got around it === frankban is now known as frankban|afk [20:38] Morning all o/ [20:42] veebers: morning! [20:42] veebers: is quiet today [20:43] hey hml o/ how's things? Hopefully quiet is a good thing ^_^ [20:44] veebers: not bad… you? [20:51] hml: can't complain :-) [20:57] babbageclunk: a small +18/-3 PR to fix a regression from yesterday https://github.com/juju/juju/pull/8831 [21:17] what does a blocked status mean: "backend relation required" o.O All I did was juju deploy pgbouncer :o [21:17] ahn I guess it was still waiting for postgresql nvm my question! [21:25] if it's following this guideline: https://docs.jujucharms.com/2.3/en/reference-status then you probably should take a look to unblock it. [21:29] thx catbus it's just that it feels like my model/controller is a bit unhealthy. Juju gui keeps loading too :/ [21:30] can I manually restart juju gui? [21:40] kelvin: i've had time to do a more thorough review on the charm PR; let me know if you have questions [21:41] wallyworld: oops, missed that - looking now