| openstackgerrit | Elod Illes proposed openstack/ceilometer-powervm master: Install ceilometer from pypi https://review.openstack.org/609058 | 12:12 |
|---|---|---|
| openstackgerrit | Elod Illes proposed openstack/ceilometer-powervm master: Install ceilometer from pypi https://review.openstack.org/609058 | 12:21 |
| efried | edmondsw: We going to move forward with ^ ? | 13:22 |
| efried | I think we ought to be able to depend on depends-on to pick up new changes moving forward, nah? | 13:22 |
| edmondsw | I'm not sure how DependsOn matters here? | 13:23 |
| edmondsw | I think we probably proceed with this to get things working again | 13:24 |
| edmondsw | I'm still hoping tonyb can suggest something better, but let's not wait on that | 13:24 |
| efried | edmondsw: I thought the whole reason we were doing the -egit business was so that we could develop against new changes in ceilometer. The way the rest of the community deals with that is, if a new feature or bugfix is being developed in the dep, they use depends-on to get zuul to pick it up. | 13:28 |
| efried | Oh, but this is a weird one because ceilometer isn't a lib | 13:28 |
| efried | and the whole point is for us to CI against whatever they're developing so we can catch problems early. | 13:28 |
| efried | yeah | 13:28 |
| edmondsw | I don't think CI matters here. The tox.ini shouldn't be involved in CI, right? I think it's just for local testing | 13:34 |
| edmondsw | and zuul | 13:34 |
| edmondsw | for CI, I think we'll still be using ceilometer's master | 13:35 |
| edmondsw | based on how we setup the CI | 13:35 |
| edmondsw | and for zuul, you're probably right about using DependsOn | 13:36 |
| edmondsw | but that still leaves local testing where ideally we should be testing with ceilometer master, but this change leaves us testing with something older | 13:36 |
| edmondsw | efried ^ | 13:37 |
| efried | "for CI, I think we'll still be using ceilometer's master" <= howzat? | 13:37 |
| efried | we did some special hackyhack? | 13:37 |
| edmondsw | not a hack... but tox just doesn't really have anything to do with CI | 13:38 |
| efried | no, but CI uses requirements | 13:39 |
| efried | doesn't it? | 13:39 |
| edmondsw | sure, and the fact that we require _at_least_ ceilometer 11 will work when devstack is setup using ceilometer master | 13:39 |
| efried | okay, I still don't get how that's working, but I'll gladly take your word for it. | 13:40 |
| edmondsw | https://github.com/powervm/powervm-ci/blob/0e1bdae8c03228d14b9d9b0e6039ce718060a6b0/devstack/prep_devstack.sh#L127-L133 | 13:41 |
| edmondsw | if CI is testing a master commit, ZUUL_BRANCH will be master | 13:42 |
| edmondsw | if it's testing rocky, it'll be stable/rocky, etc | 13:43 |
| edmondsw | some time after the lines above, we'll do stack.sh and it will be using ZUUL_BRANCH for ceilometer as well as everything else | 13:43 |
| efried | cool | 13:50 |
| efried | And things like py* unit tests are using zuul jobs, which don't touch our tox setup? | 13:51 |
| *** openstackgerrit has quit IRC | 14:58 | |
| *** openstackgerrit has joined #openstack-powervm | 15:00 | |
| *** openstackgerrit has quit IRC | 15:22 | |
| *** openstackstatus has quit IRC | 18:28 | |
| *** openstackstatus has joined #openstack-powervm | 18:30 | |
| *** ChanServ sets mode: +v openstackstatus | 18:30 | |
| edmondsw | efried sorry, just seeing that last question... and I don't understand it | 19:36 |
| efried | it's really not important | 19:37 |
| edmondsw | efried alrighty then | 19:37 |
| edmondsw | oh, and I added my +2 on https://review.openstack.org/#/c/609058 | 19:37 |
| efried | +A | 19:38 |
| *** openstackgerrit has joined #openstack-powervm | 19:50 | |
| openstackgerrit | Merged openstack/ceilometer-powervm master: Install ceilometer from pypi https://review.openstack.org/609058 | 19:50 |
| tonyb | efried, edmondsw: I'll look into the ceilometer thing. If zuul is setup right you shoudl be getting master and Depends-On will work. | 20:29 |
| tonyb | I can hack up a 'dev' tox env for you for local development | 20:29 |
| efried | tonyb: I think the only thing we're concerned about now is the fact that there's no way to get master ceilometer when running tox locally. | 20:29 |
| efried | (well, without manually switching branches under .tox/) | 20:30 |
| tonyb | It'll probably take me a few days ... I'm sure you can understand what my inbox looks like after 3weeks ;P | 20:30 |
| efried | heh. No hurry. Development on ceilometer-powervm is *very* light. | 20:30 |
| tonyb | efried: Yeah I can setup a toxenv that'll do that for you and double check you're getting the right ceilometer in the gate | 20:30 |
| efried | oh, okay, cool. Thank you. | 20:31 |
| tonyb | Yeah you're for sure only getting released versions in the gate :/ | 20:32 |
| efried | oh. edmondsw ^ | 20:33 |
| openstackgerrit | Tony Breeds proposed openstack/ceilometer-powervm master: [DNM] Testing zuul https://review.openstack.org/609823 | 21:22 |
| * tonyb is pretty sure ^^^ will not work but it's a start for doing what we want | 21:22 | |
| openstackgerrit | Tony Breeds proposed openstack/ceilometer-powervm master: [DNM] Testing zuul https://review.openstack.org/609823 | 21:29 |
| *** openstackgerrit has quit IRC | 22:19 | |
| *** openstackgerrit has joined #openstack-powervm | 22:20 | |
| *** openstackstatus has quit IRC | 22:28 | |
| *** openstackstatus has joined #openstack-powervm | 22:29 | |
| *** ChanServ sets mode: +v openstackstatus | 22:29 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!