lucidone | How would I log from a k8s charm using the operator framework? (So that the logs show up in the pod logs e.g kubectl -n test-k8s logs -f test-charm-operator-0) | 00:53 |
---|---|---|
lucidone | Is that something debugf in ops/main.py will handle (function definition is currently just 'pass') | 00:55 |
lucidone | Ah scratch that, looks like print() does show up in logs - a separate issue meant that chunk of code wasn't getting hit :) | 01:02 |
thumper | Here is a relatively boring PR for someone: https://github.com/juju/juju/pull/11124 | 02:10 |
thumper | lucidone: oh, I don't know if we have any way to write into the pod logs | 02:11 |
thumper | wallyworld, kelvinliu: do you know? | 02:11 |
thumper | lucidone: we have a juju-log command that has things show up in the juju debug-log | 02:11 |
thumper | but I don't know if there is a specific one for the pod log | 02:11 |
wallyworld | you can also use kubectl log | 02:11 |
thumper | wallyworld: from the charm? | 02:11 |
wallyworld | not really | 02:11 |
wallyworld | kubectl is for outside the cluster | 02:12 |
thumper | wallyworld: do you think that we should have juju-log commands for k8s models go into the pod log? | 02:12 |
evhan | I think just writing to stdout does the right thing. | 02:12 |
wallyworld | they do go to pod log | 02:12 |
wallyworld | the juju logs to go both the k8s log and the juju log | 02:13 |
kelvinliu | didn't find a juju-log helper method, but you can do this https://github.com/canonical/operator/blob/dd2a0d4537ff9ef2f19354a3fd7a9598e4b4c076/ops/model.py#L605 | 02:15 |
evhan | lucidone: For debug-log you can use charmhelpers.core.hookenv.log from https://pypi.org/project/charmhelpers/, works fine with operator. | 02:30 |
evhan | In case it hasn't been reported elsewhere, https://discourse.jujucharms.com is down from here. | 02:30 |
wallyworld | seems to be back now | 02:45 |
tlm | last job!!!! | 02:46 |
wallyworld | whoot | 02:47 |
tlm | i have run them all seperate wallyworld like we discussed to avoid the wait | 02:48 |
wallyworld | tlm: i'm still seeing red dots? | 02:49 |
tlm | yeah that was the job from this morning | 02:49 |
tlm | if you click through you can see the latests ones | 02:49 |
tlm | no way to re run something in jenkins and have latest result trickle up | 02:50 |
wallyworld | ah, didn't realise that | 02:50 |
tlm | but | 02:50 |
tlm | 2.7.1 is released | 02:50 |
tlm | I think | 02:50 |
wallyworld | tlm: snap store thinks so! | 02:51 |
wallyworld | i'll check simplestreams also | 02:51 |
tlm | party time | 02:51 |
wallyworld | and we should check the ppa manually too | 02:51 |
wallyworld | using apt query blah | 02:51 |
tlm | ok i'll jump on one of my ubuntu boxes | 02:51 |
wallyworld | tlm: seem to be missing released simple streams http://streams.canonical.com/juju/tools/streams/v1/ | 02:53 |
wallyworld | metadata is in proposed but not released | 02:53 |
wallyworld | maybe there's one more job still to run | 02:54 |
tlm | it's all green when I check | 02:55 |
wallyworld | maybe it takes a few minutes to wash through | 02:56 |
wallyworld | needs to have a job run on jerff | 02:56 |
tlm | ah that guy. It passed | 02:56 |
tlm | https://jenkins.juju.canonical.com/job/release-juju-request-streams-update/120/ | 02:57 |
=== degville_ is now known as degville | ||
nammn_de | manadart: so i created the initial mocking for apiserver/spaces test https://github.com/juju/juju/pull/11088/ your comment https://github.com/juju/juju/pull/11088#discussion_r367868986 I could not 100% realize. I cannot remove all of my changes from StubNetwork, because it needs to implement Backing so I just left it there. Which can be fixed | 16:05 |
nammn_de | later when we move more and more tests from stubs to gomock | 16:05 |
manadart | nammn_de: I think we could wrap the StubNetwork with with empty stubs for those methods, and remove them. They won't be called from the old tests. | 16:15 |
manadart | Remove them from StubNetwork that is. | 16:16 |
nammn_de | manadart: how would you do that? My initial idea would be to just remove all of the code I added and just add those 2 stub methods : | 16:18 |
nammn_de | AllEndpointBindings AllMachines without any logic and just return | 16:18 |
manadart | nammn_de: Yes. | 16:18 |
nammn_de | manadart: the other "problem: i've encountered was: because we already have a test lying in spaces_test package I could either: put the new test based on macks in a newfolder or let it be in spaces for now, I put it in spaces for now | 16:22 |
manadart | nammn_de: In the spaces package is OK. As long as it is suffixed with _test.go, it will only be built for tests. We do this in other places. | 16:24 |
manadart | spaces_mock_test.go would follow the convention. | 16:24 |
manadart | That's the mock. The tests can stay in spaces_test and import the mock. | 16:26 |
nammn_de | manadart: what i found out from looking at the other packages is the following: mocks are under ../mocks/ and follow the naming of <interface/topic>_mock.go and the test itself is under <api>_test.go, but as we have both (stub and mock tests) I need to choose a new name. This would lead to spaces_mock_test.go, right? | 16:30 |
nammn_de | and later when we have moved all from spaces_test, spaces_mock_test will be spaces_test | 16:31 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!