=== CyberJacob is now known as zz_CyberJacob | ||
rock | h | 05:12 |
---|---|---|
rock | Hi all. We pushed our charm to charm public store using Ubuntu SSO account. Charm will be present in store as cs:~hareeshk/kaminario-cinder-10 . Here "hareeshk" is the account username. we removed this user account. And using same mailID we created another user account. and pushed charm to public store again. cs:~kaminario2/kaminario-cinder-0. But in store previous pushed charm also there. We forgot to revoke the permissions of that c | 05:58 |
rock | How do we remove that charm from store. We tried to create same user account. But It was showing "hareeshk" username already in use. But we deleted that user account. | 05:59 |
rock | "hareeshk" will be stored some where in server Cache. | 05:59 |
rick_h___ | rock: file a bug in https://github.com/canonicalltd/jujucharms.com and will mention it to the folks that manage the charmstore | 06:22 |
rock | rick_h__: Thank you. | 06:52 |
=== babbageclunk is now known as babbageclunkafk | ||
=== macgreag1ir is now known as macgreagoir | ||
SimonKLB | how do i install the charm-terms commands? they don't seem to be in by default in charm-tools 2.1.5 | 09:12 |
pascalmazon | Hi, I've recently reinstalled charm tools (2.1.5-0ubuntu1~ubuntu14.04.1~ppa2) on a trusty. However when I try to call `charm proof`, it complains that pip isn't the correct version: "pkg_resources.DistributionNotFound: The 'pip>=7.1.2' distribution was not found and is required by charm-tools". Any ideas? | 10:35 |
SimonKLB | pascalmazon: https://github.com/juju/charm-tools/issues/274 | 10:36 |
pascalmazon | SimonKLB: oh thanks, I should have checked there… | 10:37 |
anrah | Hi all! I was about to ask what the prefer-ipv6 option should do and is it still available on juju 2.0? | 12:34 |
anrah | At least when that is set to a model nodes private-address still is ipv4 | 12:35 |
=== saibarsp_ is now known as saibarspeis | ||
BlackDex | Hello there. I want to know if it is somehow possible to have multiple hacluster's on one node? | 12:44 |
BlackDex | Currently it seems that it doesn, because haproxy isn't getting configured | 12:44 |
=== hml_ is now known as hml | ||
kiko | morning | 15:20 |
=== iatrou_ is now known as iatrou | ||
lazyPower | bdx - should i resched the elastic stack planning session? | 17:07 |
=== daniel1 is now known as Odd_Bloke | ||
bdx | lazyPower: shit | 17:18 |
bdx | I'm so sorry | 17:18 |
lazyPower | bdx no worries, we'll reschedule and reconvene | 17:18 |
lazyPower | bdx and family friendly language *shakes a finger* | 17:18 |
bdx | was stuck in a bit of traffic ... just getting in now | 17:18 |
marcoceppi | this is...the internet ;) | 17:18 |
bdx | my b | 17:18 |
lazyPower | i think the meeting time is a bit early for your TZ | 17:18 |
lazyPower | i was trying to overlap with le magicaltrout's EU TZ as well | 17:19 |
bdx | usually I'm online by 8 | 17:19 |
bdx | so 9 is ok | 17:19 |
lazyPower | ok, would you want to try to do this again later this week? | 17:19 |
bdx | definitely | 17:20 |
lazyPower | ok, i'll send something out later today :) | 17:21 |
lazyPower | marcoceppi - FYI, we're blocked atm on completing that PR for the test infra. they moved some files and i'm tracking down their result munging scripts | 17:22 |
bdx | perfect | 17:22 |
vmorris | lazyPower: https://goo.gl/images/kkpQ8j | 17:58 |
lazyPower | that ^ | 17:58 |
x58 | charm proof is complaining I don't have any "provides", so I set a provides... and now it's complaining that I don't provide hooks for that... | 18:28 |
x58 | Is there a way to have charm proof not complain, yet not really "provide" anything, and thus not need hooks for it. | 18:28 |
x58 | My charm only sets information in the OS, it doesn't provide any services or anything that anyone can relate with. | 18:29 |
x58 | What is the best way to use the juju-info interface? charm proof complains about using juju-info as the requires name, yet that is what the docs tell me to use... | 18:41 |
=== zz_CyberJacob is now known as CyberJacob | ||
bdx | concerning juju resources, is the jujuresources pkg the recommended path to managing resources? | 19:00 |
bdx | I'm looking at this documentation here -> http://pythonhosted.org/jujuresources/index.html | 19:00 |
bdx | wondering if its current | 19:00 |
jrwren | bdx: that is not current for juju 2.0 | 19:01 |
bdx | jrwren: thx, do you know of current api docs for 2.0 resources? | 19:01 |
jrwren | bdx: https://jujucharms.com/docs/stable/developer-resources | 19:01 |
bdx | jrwren: no python api tho? | 19:02 |
jrwren | bdx: there is a wrapper around resource-get in charmhelpers. | 19:02 |
bdx | other than what is offered throught hookenv? | 19:02 |
jrwren | bdx: I don't know. | 19:02 |
bdx | gotcha, ok ... just wanted to make sure I wasn't missing something here | 19:03 |
bdx | jrwren: thx | 19:03 |
bdx | cory_fu: really nice job on the jujuresources pkg and docs | 19:05 |
bdx | cory_fu: would you mind updating them to indicate juju 1.x only pls | 19:05 |
bdx | concerning resources, does the 'upgrade-charm' hook run when `charm attach` is ran? | 19:07 |
jrwren | bdx: no. | 19:08 |
jrwren | bdx: you can run that yourself, or write an action which uses the resource and run that action. | 19:08 |
bdx | jrwren: I'm thinking abount a ci use case, where say jenkins might run `charm attach` following successful build of the resource | 19:10 |
bdx | jrwren: what you are saying is this might be best implemented having jenkins run `charm attach`, following that an action that gets and unpacks the resource | 19:11 |
jrwren | bdx: that is fine. it can just as easily run juju upgrade-charm or juju action run BLAH as a next step. | 19:11 |
bdx | jrwren: ah, so a better workflow might be; `charm attach` to get the new resource in the charm store, then `juju attach` to get the resource onto the controller, then `juju action run` to get the charm to grab the new resource | 19:15 |
marcoceppi | bdx jrwren yes it is | 19:15 |
marcoceppi | bdx jrwren when you attach resources, upgrade-charm is run | 19:15 |
marcoceppi | bdx: it's either charm attach, then juju upgrade-charm, or juju attach | 19:16 |
jrwren | oh no! | 19:16 |
jrwren | sorry bdx | 19:16 |
marcoceppi | then upgrade-charm runs for both. | 19:16 |
bdx | marcoceppi: oh nice, so I can just react to the upgrade-charm hook instead of a custom action following the attaching oof a new resource | 19:16 |
jrwren | bdx: for a CI scenario you describe, I'd skip the charmstore entirely. | 19:16 |
x58 | I am trying to use juju-info interface as a requires to allow my charm to be a sub-ordinate, however as soon as I add the updated charm to the relation and run juju add-relation <other charm> <my charm> it says no relations found. | 19:16 |
marcoceppi | bdx: yeah, lazyPower has done this a lot | 19:17 |
marcoceppi | oh, and he left | 19:17 |
bdx | jrwren: but when new deploys happen, I want the charm to deploy with the latest resource | 19:18 |
marcoceppi | bdx: it will | 19:18 |
marcoceppi | bdx: whatever is in the charm store, will get deployed | 19:18 |
marcoceppi | you don't need to attach post deployment | 19:18 |
bdx | jrwren: hence why I feel keeping the charm store resource up to date is important too | 19:18 |
jrwren | bdx: ok, its up to you. | 19:18 |
bdx | jrwren: if the charm store step is skipped, then deploying a new instance of the charm would require the user having the rescource | 19:19 |
marcoceppi | bdx: right, bdx I think you're on the right train of thought | 19:20 |
bdx | jrwren: what I'm thinking, is if its just fully left up to jenkins, then the resource is always current, and the dev/user deploying doesn't have to know or care about it | 19:20 |
x58 | marcoceppi: I have: https://gitlab.com/bertjwregeer/juju_staticroutes/blob/master/metadata.yaml#L16 yet when I call juju add-relation ceph-osd:juju-info staticroutes:host-system it fails to work... documentation says to use "juju-info" as the requires name, but I can't push to the charmstore when I use that... | 19:21 |
x58 | (juju says "no relations found") | 19:21 |
bdx | jrwren, marcoceppi: is there any juju<->jenkins plugins work currently being done, or pre-existing that you know about? | 19:22 |
x58 | This documentation is outdated: https://jujucharms.com/docs/2.0/authors-implicit-relations (can't require juju-info...) | 19:22 |
marcoceppi | bdx: I'd check the mailing list, I'd just ask - I know a few teams have done some simple stuff, but it's mostly just scripts | 19:22 |
bdx | marcoceppi, jrwren: also, upgrade-charm shouldn't be ran on `charm attach`, because we are only interfaceing to the charm store at that point right? | 19:23 |
marcoceppi | x58: I'm about to go offline, but if you don't get an answer in an hour I'll be back to take a look | 19:23 |
marcoceppi | bdx: it's not, but the operator i stold there's an upgrade available | 19:23 |
marcoceppi | bdx: charm attach -> juju upgrade-charm -> upgrade-charm hook run, new resource is available | 19:24 |
x58 | marcoceppi: Ok... | 19:24 |
marcoceppi | x58: I think that should be provides.... | 19:24 |
x58 | Nope, all charms implicitly PROVIDE juju-info. | 19:25 |
marcoceppi | x58: is this a subordinate? | 19:25 |
x58 | Yes. | 19:25 |
marcoceppi | okay, I'll be back soon | 19:25 |
x58 | It is not explicitly listed as a sub-ordinate though... because it may also be deployed stand-alone. | 19:25 |
x58 | marcoceppi: https://api.jujucharms.com/charmstore/v5/ntp-16/archive/metadata.yaml like this. requires juju-info. | 19:28 |
thumper | x58: hey there, I think this week at the planning sprint we are discussing model level subordinates | 19:28 |
thumper | charms that should be on all machines | 19:29 |
thumper | which match the need for things like ntp and other base machine tweaks | 19:29 |
bdx | marcoceppi: just to be clear, isn't `juju upgrade-charm` overkill if all i need is the new resource? | 19:30 |
bdx | which I could get with `charm attach` | 19:30 |
bdx | bleh | 19:30 |
bdx | `juju attach` | 19:30 |
aisrael | thumper: That would be a great feature. | 19:31 |
bdx | thumper: +1 | 19:31 |
thumper | there is clearly a need for it | 19:31 |
x58 | thumper: That doesn't help me resolve the issue right now.. | 19:32 |
x58 | thumper: I can't upload a charm that uses juju-info as the relationname under requires (like the ntp charm) | 19:33 |
x58 | and when I name it something else, add-relation fails with "no relation found" | 19:33 |
thumper | x58: sure | 19:33 |
x58 | Also... the docs on the site are out of date... which as always is fantastic. | 19:33 |
x58 | What am I missing? | 19:34 |
thumper | x58: I don't believe that the system supports a charm that is both a subordinate, and not a subordinate | 19:34 |
thumper | I'm not sure why it is complaining about juju-info | 19:34 |
thumper | I'm wondering if it is considered only available for subordinates | 19:35 |
thumper | ? | 19:35 |
thumper | perhaps? | 19:35 |
x58 | charm push . | 19:35 |
x58 | ERROR charm "staticroutes" using a reserved relation name: "juju-info" | 19:35 |
thumper | if you mark the charm as a subordinate, does it work? | 19:35 |
x58 | Now, if I change it to be subordinate... I will break existing users that are not using it as a subordinate when they upgrade... | 19:35 |
thumper | ah... | 19:36 |
thumper | poo | 19:36 |
thumper | perhaps you need two charms? | 19:36 |
x58 | Yeah... no. | 19:36 |
x58 | I really don't want to maintain two charms. | 19:36 |
thumper | x58: what is the intent of the juju-info relation? | 19:37 |
x58 | To be able to add a relationship so that it is subordinate when related to an existing charm. | 19:37 |
x58 | Without breaking existing users by removing the ability to just deploy staticroutes --to <host. | 19:38 |
thumper | I get a strong feeling that this is not a supported use case | 19:38 |
thumper | I don't think you are missing anything, I think that this use case just was never thought about | 19:39 |
thumper | or it was, and explicitly not supported | 19:39 |
thumper | but I don't know which | 19:39 |
x58 | That'd be nice to have documented on https://jujucharms.com/docs/2.0/authors-implicit-relations | 19:39 |
kiko | x58, out of curiosity, what are you charming? | 19:39 |
x58 | as well as the fact that you can't name the relation "juju-info" like it shows in that example. | 19:39 |
x58 | kiko: adding staticroutes becaue MaaS doesn't provide a feature for us to add routes on a specific interface. | 19:40 |
x58 | We have 3 racks of Ceph nodes. | 19:40 |
x58 | THe backend cluster network is on a /64, I need to add a static route to a /52 on the interface for that backend cluster network | 19:40 |
x58 | so that backend cluster traffic goes over the right interface, instead of over the default route (which is the frontend cluster interface where clients communicate) | 19:41 |
kiko | x58, we are doing that work this cycle, fwiw -- why don't we work together to do that in MAAS instead? | 19:41 |
x58 | kiko: https://gitlab.com/bertjwregeer/juju_staticroutes/ | 19:41 |
bdx | x58: not sure if this is the best way, but what I use to do was just edit the maas dhcp conf | 19:41 |
x58 | kiko: Because I am already using it in production. | 19:41 |
kiko | x58, I see, it's complete already | 19:41 |
kiko | well | 19:41 |
bdx | x58: or dhcp template that maas uses | 19:41 |
x58 | Oh yeah... I am just trying to make it a sub-ordinate. | 19:41 |
kiko | if it makes you feel any better, we're adding the functionality now | 19:41 |
kiko | because we're aware of that limitation | 19:42 |
x58 | bdx: No DHCP used for the IPv6 network... | 19:42 |
bdx | oooh IPv6 .... nm | 19:42 |
x58 | bdx: As in the MaaS server doesn't even have an interface on that network. | 19:42 |
kiko | bdx, how would dhcp help with static routes, though? | 19:42 |
x58 | You can push more routes with DHCP | 19:42 |
bdx | yeah | 19:42 |
kiko | interesting | 19:42 |
bdx | I still have an openstack with that hack | 19:42 |
bdx | lol | 19:43 |
kiko | the more general case of not-DHCP is the killer there of course | 19:43 |
kiko | anyway | 19:43 |
x58 | I really wish two things: 1. Juniper would support RFC4191 and 2. Linux would accept /52 routes from router advertisements. | 19:43 |
bdx | totally | 19:43 |
thumper | x58: how are you declaring the relation? | 19:43 |
x58 | thumper: juju add-relation ceph:juju-info staticroutes:system-host | 19:43 |
thumper | x58: no, in the charm's metadata.yaml | 19:44 |
x58 | kiko: Basically on a subnet defined on a Fabric we would need the ability to specify extra routes that are added to the networking config when the host is brought up. | 19:44 |
kiko | x58, yeah, ivoks has been asking for that since forever, and it's now on the plan for 2.3 | 19:45 |
x58 | kiko: Also, please please please test that everything works with IPv6... | 19:46 |
tychicus | I'm trying to get juju gui up and running, but it hang at "Connecting to the juju model" | 19:47 |
tychicus | looks like a websocket issue Uncaught DOMException: Failed to construct 'WebSocket': The URL '<no value>' is invalid. | 19:48 |
tychicus | are there any docs that I can consult to help me resolve this issue? | 19:48 |
tychicus | following instructions from here https://jujucharms.com/docs/2.0/tut-google#use-juju%27s-gui | 19:50 |
tvansteenburgh | tychicus: you could try `juju upgrade-gui` in case it's a bug that's been fixed | 20:04 |
tychicus | Juju GUI at version 2.2.2 | 20:06 |
tychicus | if it helps it's the version that was automatically loaded when I ran juju bootstrap | 20:08 |
tvansteenburgh | tychicus: yeah looks like that's the latest | 20:09 |
tychicus | any other things I can do to troubleshoot? | 20:10 |
tvansteenburgh | tychicus: hatch may be able to help you | 20:10 |
* hatch waves | 20:10 | |
tychicus | hi hatch | 20:10 |
tychicus | hatch: what information do you need from me? | 20:11 |
hatch | hey tychicus I hear you have an issue in the GUI? | 20:11 |
tychicus | correct | 20:11 |
hatch | I just joined, so I have no idea what's going on :) | 20:11 |
tychicus | Uncaught DOMException: Failed to construct 'WebSocket': The URL '<no value>' is invalid. | 20:12 |
hatch | oh kay, that's a new one | 20:12 |
tychicus | ran juju bootstrap test-maas-controller test-maas —to=test-juju.maas | 20:12 |
hatch | ok and then to load the gui, `juju gui` } | 20:13 |
hatch | ? | 20:13 |
tychicus | yep | 20:13 |
tychicus | ran juju gui --show-credentials | 20:13 |
hatch | hmm | 20:14 |
tychicus | browser renders "Connecting to the Juju model" | 20:14 |
hatch | can you try `juju upgrade-gui` ? | 20:14 |
hatch | and then try again | 20:14 |
tychicus | Juju GUI at version 2.2.2 | 20:14 |
hatch | ok give me a few minutes to dig into this | 20:15 |
tychicus | sure | 20:15 |
tychicus | thanks | 20:15 |
hatch | tychicus: oh, what browser? | 20:15 |
tychicus | hatch: tried FF 49.0.2, chrome 54.0.2840.87 (64-bit), and safari 10.0.1 (10602.2.14.0.7) | 20:16 |
hatch | hah ok | 20:17 |
tychicus | safari gives a divvernt error | 20:18 |
tychicus | Wrong url scheme for WebSocket and SyntaxError (DOM Exception 12): The string did not match the expected pattern. | 20:19 |
hatch | yeah I think I know what might be causing the issue | 20:19 |
tychicus | ok | 20:19 |
hatch | could you open the browser console and copy the value in `juju_config` and pm me the values? | 20:19 |
tychicus | sure | 20:20 |
hatch | thanks | 20:20 |
x58 | thumper: Looks like setting it as subordinate did the trick... | 20:22 |
x58 | thumper: Think the only users are internal to $WORK, I'll just make this breaking change. | 20:23 |
thumper | hmm... the docs need to be more clear if juju-info is only available to subordinates | 20:24 |
thumper | seems a bit weird to me | 20:24 |
x58 | Not the first time I've had issues with docs... | 20:26 |
bdx | concerning resources, is there a way to attach a specific revision of a resource? | 20:29 |
bdx | say if you wanted to roll back or something | 20:30 |
thumper | bdx: I would say you should be able to, but I don't know they syntax | 20:33 |
lazyPower | bdx - doesn't appear to be the case when using resources from the charm store. juju deploy cs:~containers/kubernetes-e2e --resource e2e_amd64=e2e_amd64-0 -- but the --resource flag seems to assume its coming from local disk | 20:42 |
=== babbageclunkafk is now known as babbageclunk | ||
bdx | lazyPower: do you feel this is a functionality that should exist? | 20:58 |
bdx | should I create a bug/feature request against charmstore, or charmstore-client? | 20:58 |
lazyPower | bdx - i haven't tested this, but i think the way it would work is if you rolled back to a previous version of the charm... you would get the resource thats associated with that revision of the charm. | 21:02 |
lazyPower | s/would work/does work. | 21:04 |
bdx | I see | 21:04 |
rick_h___ | bdx: resources are revisioned just like charms and follow the released channels/etc. It should be possible. | 21:04 |
* rick_h___ goes to look for syntax | 21:04 | |
bdx | lazyPower: I dont see any mapping though .. take mattermost for example `charm show cs:~cmars/mattermost` and `charm show cs:~cmars/mattermost-16` both show rev 3 for the resource | 21:07 |
lazyPower | bdx - its entirely possible to use the same resource between two revisions of the charm | 21:07 |
bdx | lazyPower: but the mapping that charmstore keeps | 21:08 |
lazyPower | bdx: it appears that its been the same resource since revision 8 of the charm | 21:09 |
lazyPower | charm show cs:~cmars/mattermost-8 resources | 21:09 |
lazyPower | anything prior to that, no resource that i can see, has been associated with that charm release. | 21:09 |
bdx | lazyPower: I see, thx | 21:15 |
=== rick_h___ is now known as rick_h | ||
tychicus | hatch: sudo add-apt-repository ppa:juju/stable did the trick | 22:59 |
tychicus | thanks again for the help | 22:59 |
hatch | tychicus: glad to hear it, np | 22:59 |
tychicus | now to see if I can get ceph up and running :) | 23:00 |
hatch | good luck! | 23:05 |
tychicus | will juju show you all of the machines that you have commissioned in maas? | 23:14 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!