| *** Shuo has joined #zuul | 00:19 | |
| *** saneax-_-|AFK is now known as saneax | 00:25 | |
| jeblair | rbergeron, mordred: well, i'm planning on writing some stuff up for the ptg tomorrow, so we can see where we are after that :) | 00:27 |
|---|---|---|
| mordred | jeblair: woot | 00:29 |
| SpamapS | question: how would one create a similar animation to the one that is used in zuul presentations? | 00:57 |
| clarkb | SpamapS: I believe its all just javascript | 00:58 |
| clarkb | SpamapS: so you edit the javascript things and run it in your browser | 00:59 |
| SpamapS | clarkb: ok cool.. I want to try and show it on a larger scale with humans approving and reviewing on a time line. | 01:05 |
| SpamapS | so you get a visual sense of why you want automated gating in the first place | 01:06 |
| *** jamielennox is now known as jamielennox|away | 01:07 | |
| clarkb | that should be possible. the animation is graph manipulation via javascript so should be generatable from logs | 01:08 |
| *** jamielennox|away is now known as jamielennox | 01:21 | |
| *** Shuo has quit IRC | 01:49 | |
| SpamapS | maybe I could abuse gource again | 02:15 |
| *** harlowja has quit IRC | 02:53 | |
| *** saneax is now known as saneax-_-|AFK | 04:07 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: Support swift logserver without Send-Temp-Url-Key https://review.openstack.org/270338 | 04:48 |
| *** bhavik1 has joined #zuul | 06:26 | |
| *** bhavik1 has quit IRC | 06:35 | |
| *** Cibo_ has joined #zuul | 07:57 | |
| *** saneax-_-|AFK is now known as saneax | 08:19 | |
| *** hashar has joined #zuul | 08:47 | |
| *** Cibo_ has quit IRC | 09:14 | |
| *** Cibo_ has joined #zuul | 09:27 | |
| *** openstackgerrit has quit IRC | 10:18 | |
| *** saneax is now known as saneax-_-|AFK | 11:49 | |
| *** saneax-_-|AFK is now known as saneax | 11:58 | |
| *** hashar is now known as hasharLunch | 12:19 | |
| *** saneax is now known as saneax-_-|AFK | 12:19 | |
| *** saneax-_-|AFK is now known as saneax | 12:29 | |
| *** hasharLunch is now known as hashar | 12:47 | |
| *** openstackgerrit has joined #zuul | 14:11 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 14:11 |
| Shrews | with any luck, min-ready nodes will now be launched ^^^ | 14:11 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 14:15 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: DNM: Re-enable devstack test https://review.openstack.org/431649 | 14:27 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Update nodepool 'list' command https://review.openstack.org/431647 | 14:27 |
| *** Shrews has quit IRC | 14:29 | |
| *** Shrews has joined #zuul | 14:32 | |
| *** saneax is now known as saneax-_-|AFK | 14:39 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Add action plugins to restrict untrusted execution https://review.openstack.org/428798 | 15:23 |
| mordred | jeblair: I think that should be about everytihng until we need to filter plugins from roles - but that should be fairly easy with the playbook plugin filtering already there | 15:24 |
| mordred | jeblair: at least, that should be everything minus adding container execution wrapping | 15:25 |
| *** yolanda has quit IRC | 15:38 | |
| *** yolanda has joined #zuul | 15:38 | |
| pabelanger | morning | 15:44 |
| rbergeron | yes it is. ughhhh | 15:44 |
| pabelanger | Shrews: are we ready to update nodedeleter for zookeeper? | 15:46 |
| pabelanger | from what I see, it is still using the database | 15:47 |
| Shrews | pabelanger: you can update the command to set node.state=DELETE, but there is nothing running to actually delete though | 15:48 |
| pabelanger | Shrews: ya, that's what I've been doing now | 15:49 |
| pabelanger | was looking to plumb waitForNodeDeletion(), but will hold off for now | 15:50 |
| jeblair | mordred: w00t, i'll resume work on roles next week | 15:55 |
| Shrews | jeblair: getting closer. min-ready support added: http://logs.openstack.org/49/431649/4/check/gate-dsvm-nodepool/e598bf9/logs/screen-nodepool.txt.gz#_2017-02-10_14_46_15_365 | 16:00 |
| Shrews | jeblair: still fails. i suspect the request is being declined b/c the images are not yet uploaded and ready. will have to add a check for that | 16:01 |
| jeblair | Shrews: cool, i will start poking after breakfast | 16:01 |
| Shrews | jeblair: no need yet, unless you just want to. i have to head out for an appointment in a bit. will add the check for ready images when i return | 16:02 |
| pabelanger | Shrews: Nice | 16:02 |
| Shrews | oh, i just thought of something clever to try... | 16:06 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: DNM: Re-enable devstack test https://review.openstack.org/431649 | 16:07 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Update nodepool 'list' command https://review.openstack.org/431647 | 16:07 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 16:07 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: DNM: Re-enable devstack test https://review.openstack.org/431649 | 16:10 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Update nodepool 'list' command https://review.openstack.org/431647 | 16:10 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 16:10 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Inherit playbooks and modify job variance https://review.openstack.org/430483 | 16:12 |
| Shrews | pabelanger: we're going to need some sort of 'wait' test method that will handle the nodes launched by the min-ready path. i haven't had time to put much thought into that yet | 16:12 |
| Shrews | pabelanger: that's the reason the 'list' change is failing, atm | 16:13 |
| pabelanger | Shrews: ya | 16:13 |
| jeblair | mordred: https://review.openstack.org/430483 | 16:14 |
| jeblair | mordred: that should be ready for review now | 16:14 |
| pabelanger | we should be able to update waitfornodes | 16:14 |
| mordred | jeblair: is it going to hurt my head? | 16:14 |
| jeblair | mordred: a little, but you've already seen half of it | 16:14 |
| Shrews | pabelanger: we can change our tests to not submit node requests and just wait on the min-ready nodes, which is probably the way we should do it. If you have time to put some thought into it, would be very helpful | 16:15 |
| Shrews | pabelanger: yes, update waitfornodes sounds like the thing | 16:15 |
| pabelanger | Shrews: sure, I can trying and get it working today | 16:16 |
| Shrews | pabelanger: feel free to post a PS to my 'list' change, if you want | 16:17 |
| pabelanger | kk | 16:17 |
| Shrews | gotta step away for a bit. bbl | 16:18 |
| *** hashar has quit IRC | 16:24 | |
| jeblair | clarkb, pabelanger, SpamapS, jlk: any of you want to look at 430483 too? | 16:24 |
| mordred | jeblair: makes sense / looks good | 16:37 |
| * Shrews notes from a remote location that nodes are being created: http://logs.openstack.org/49/431649/6/check/gate-dsvm-nodepool/c796757/ | 16:51 | |
| mordred | Shrews: woot! | 16:55 |
| pabelanger | woah, I think I just got an allocation from nodepool for a zookeeper node | 16:58 |
| *** Shuo has joined #zuul | 17:00 | |
| mordred | pabelanger: zomg | 17:01 |
| pabelanger | DEBUG:nodepool.NodePool: Deficit: fake-label: 0 (start: 0 min-ready: 1 ready: 1 capacity: 95) | 17:02 |
| pabelanger | I am sure, my code is wrong | 17:03 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: [WIP] Allocation a node from zookeeper https://review.openstack.org/432386 | 17:03 |
| pabelanger | Shrews: ^ when you are back, would be interested in talking about above | 17:04 |
| pabelanger | specifically, I guess we don't have subnodes any more | 17:04 |
| *** harlowja has joined #zuul | 17:05 | |
| *** harlowja has quit IRC | 17:05 | |
| *** harlowja has joined #zuul | 17:05 | |
| pabelanger | jeblair: ^ left few comments on above too | 17:06 |
| *** hashar has joined #zuul | 17:10 | |
| jeblair | pabelanger: ah, subnodes are gone. | 17:11 |
| pabelanger | that is what I thought | 17:11 |
| jeblair | pabelanger: zuul asks for nodes as groups now, so we don't need subnodes anymore | 17:11 |
| pabelanger | ack | 17:12 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Add 'hostname-format' to provider config section https://review.openstack.org/429841 | 17:16 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Unallocate new nodes if request is pulled https://review.openstack.org/430453 | 17:17 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Add new Node ZK model attributes. https://review.openstack.org/430992 | 17:17 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Remove 'template-' from image name format https://review.openstack.org/431564 | 17:17 |
| mordred | jeblair: btw - on the ansible action plugin patch - once we've landed the GH work I think we should set up a test job that cross-tests tip of ansible with tip of zuul - and add a job that checks to see if there are any new actoin plugins in tip-of-ansible | 17:20 |
| jeblair | mordred: ooooh | 17:20 |
| mordred | jeblair: that way we can at least be aware of someone submitting a PR to ansible that would add a new thing | 17:20 |
| mordred | jeblair: (honestly, this would also be a good way to protect ourselves against our necessary use of private ansible variables in places too) | 17:21 |
| jeblair | ya | 17:21 |
| pabelanger | jeblair: Shrews: any objections if I start removing the SubNodeLauncher from nodepool? | 17:22 |
| mordred | so it seems we have at least 2 cross-repo relationships that are interesting with ansible - ansible/zuul and ansible/shade (to test that shade changes don't break os_ modules) | 17:22 |
| SpamapS | jeblair: holy mother of all commit messages | 17:23 |
| * SpamapS reads | 17:23 | |
| rbergeron | dfa | 17:24 |
| * mordred waves to rbergeron and SpamapS | 17:25 | |
| jeblair | SpamapS: yeah, sorry. it's really two changes, but i wrote them together in the wrong order because i didn't realize the other was needed. i could separate them, but it'd probably take me a few hours. so i'm hoping i can get away with it. :) | 17:25 |
| SpamapS | No I think they're intertwined | 17:29 |
| SpamapS | I mean yeah, in a perfect world. But I don't mind two related changes in one commit when one is feeding the other. | 17:30 |
| jeblair | yeah, i might end up spending a lot of time trying to get tests to temporarily pass were i to try. | 17:31 |
| jeblair | pabelanger: see comment on 431756 | 17:34 |
| SpamapS | jeblair: exactly why I think the change is fine as-is | 17:35 |
| SpamapS | I found the tests to be extremely helpful in understanding the code | 17:36 |
| SpamapS | tried to look out for uncovered branches but it looks pretty solid | 17:36 |
| jeblair | heh, me too :) [that was my yesterday] | 17:36 |
| * rbergeron waves back at mordred | 17:38 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Remove subnodes from nodepool https://review.openstack.org/432403 | 17:38 |
| * mordred loves it when the tests are useful for undestanding :) | 17:39 | |
| pabelanger | jeblair: WFM, I can do a follow up | 17:39 |
| clarkb | jeblair: reviewed | 17:40 |
| *** Cibo_ has quit IRC | 17:46 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update nodepool hold to use zookeeper https://review.openstack.org/431756 | 17:48 |
| *** hashar has quit IRC | 17:52 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: [WIP] Allocation a node from zookeeper https://review.openstack.org/432386 | 17:55 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Inherit playbooks and modify job variance https://review.openstack.org/430483 | 17:59 |
| jeblair | clarkb, SpamapS, mordred: thanks. responded and updated ^ | 18:00 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update nodepool hold to use zookeeper https://review.openstack.org/431756 | 18:08 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Implement nodepool 'list' command https://review.openstack.org/431647 | 18:08 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 18:08 |
| pabelanger | UGh | 18:08 |
| pabelanger | jeblair: sorry | 18:08 |
| pabelanger | I incorrectly rebased there | 18:08 |
| *** yolanda has quit IRC | 18:14 | |
| *** yolanda has joined #zuul | 18:16 | |
| Shrews | pabelanger: umm | 18:21 |
| Shrews | can you undo what you did there? it lost a lot of stuff | 18:21 |
| Shrews | s/lost/reverted/ | 18:21 |
| pabelanger | Shrews: yes, I'll revert | 18:22 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 18:23 |
| Shrews | jeblair: yeah, sorry for the size of 431523. it's the culmination of days of work to get something functional | 18:24 |
| Shrews | jeblair: and it's not even complete. but i think it's a good base to add the "rest of the things" | 18:24 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update nodepool 'list' command https://review.openstack.org/431647 | 18:25 |
| pabelanger | Shrews: okay, reverted to your patch | 18:26 |
| Shrews | pabelanger: many thanks | 18:26 |
| pabelanger | Ya, sorry for the noise | 18:26 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Implement node launching https://review.openstack.org/431523 | 18:28 |
| jeblair | Shrews: no apology necessary. i'm in the same boat. :) | 18:29 |
| jeblair | clarkb: did you want take another pass over 430483, or should i go ahead and approve it? (i only made the minor changes you and SpamapS suggested) | 18:32 |
| clarkb | jeblair: let me take a quick look at interpatch diff | 18:32 |
| Shrews | If anyone else wants to review https://review.openstack.org/431523, I'm happy to do some hand holding and walk you through the important pieces. Would probably make it easier. | 18:34 |
| jeblair | i've restored my +2 on it | 18:35 |
| clarkb | jeblair: re self.playbook where is that being used? It isn't used internally to that class at least | 18:35 |
| jeblair | clarkb: in runPlaybooks | 18:35 |
| jeblair | job_status, job_code = self.runAnsiblePlaybook(self.jobdir.playbook) | 18:35 |
| *** Shuo has quit IRC | 18:36 | |
| jeblair | that happens after preparePlaybookRepos which chooses which of the candidate playbooks to run | 18:37 |
| jeblair | (via preparePlaybookRepo and findPlaybook) | 18:37 |
| jeblair | [we check out all of the repos which might have a playbook, and examine each in the order supplied by zuul. the first one we find we set jobdir.playbook to, and that is the one we run] | 18:38 |
| clarkb | looks like line 593 ish? | 18:39 |
| jeblair | clarkb: yep | 18:39 |
| pabelanger | Shrews: sure, I can take a look. If you mind looking at 432386 to see how I updated waitForNodes() | 18:44 |
| Shrews | pabelanger: ooh, yeah. looking | 18:44 |
| pabelanger | 432403 also removed subnodes | 18:44 |
| Shrews | pabelanger: oh, that won't work | 18:45 |
| pabelanger | Ya, figured there would be some issues | 18:46 |
| Shrews | pabelanger: you're attempting to use the old Nodepool stuff. There is no allocation history, and getNeededNodes() doesn't exist in the new one | 18:46 |
| jesusaur | Shrews: i'd like to take a look at 431523, it seems pretty meaty | 18:49 |
| Shrews | jesusaur: cool. yeah, it's got some meat to it | 18:50 |
| Shrews | jesusaur: let me know if you need a quick walkthrough | 18:51 |
| jesusaur | Shrews: will do | 18:52 |
| pabelanger | Shrews: right, I replaced the database calls with zookeeper. But, I am not yet familiar with the post v3 status of how allocation looks | 18:55 |
| Shrews | pabelanger: for the subnode removal stuff, I had planned to do a bulk code removal to get rid of all the unused code (everything you're changing there is not active code now). That would avoid the modifications you have to make to dead code that you had to do there. | 18:55 |
| pabelanger | Shrews: okay, that works for me | 18:55 |
| Shrews | pabelanger: but that removal is ok. probably don't want to do any more just yet | 18:56 |
| pabelanger | maybe I'll hold off until I see how you implement the allocation of nodes | 18:56 |
| Shrews | pabelanger: allocation? could you give me more words as to what you need there? | 18:57 |
| jesusaur | Shrews: im a bit confused by removeCompletedRequests, when is that called? | 19:00 |
| Shrews | pabelanger: like, how nodes are assigned to providers, or ... ? | 19:00 |
| Shrews | jesusaur: ah, that's part of the min-ready code path. The main NodePool thread code sits in a loop and makes repeated calls to createMinReady() (which calls removeCompletedRequests) | 19:02 |
| pabelanger | Shrews: right, the logic where nodepool decided to start building more nodes. if I understand, 431647, you manually allocated the node request with the _createNodeRequest() function, over nodepool launching it | 19:03 |
| pabelanger | I _think_ that is the last step to bring on devstack gate job back online | 19:03 |
| Shrews | pabelanger: ah, we needed the manual request BEFORE i added support for min-ready (which happened this morning) | 19:04 |
| Shrews | pabelanger: now we don't need the manual request | 19:04 |
| Shrews | so i should update that.... but i was hung up and getting the automatically created node | 19:05 |
| pabelanger | okay, so does this mean nodepool/allocation.py will eventually be deleted? | 19:05 |
| Shrews | yep | 19:05 |
| pabelanger | Okay | 19:05 |
| pabelanger | now I understand | 19:05 |
| pabelanger | cool | 19:05 |
| pabelanger | so ya, my patch is wrong :) | 19:06 |
| Shrews | pabelanger: capacity is calculated by querying ZK now (see _wouldExceedQuota method) | 19:06 |
| jesusaur | Shrews: ah, ok. the function name made search to see if it would deallocate nodes created in the main body of NodeRequestHandler | 19:09 |
| Shrews | jesusaur: so, fwiw, the main pieces of that change are NodeLauncher._launchNode() which is an almost copy-paste of the original node launching code. And the createMinReady() method (which starts the min-ready support code path). Everything else is just fodder in support of those | 19:14 |
| Shrews | jesusaur: and keep in mind that it isn't quite complete. it's just the base to get a minimal working nodepoold. i'll build on it in future reviews | 19:17 |
| Shrews | jesusaur: this log shows it actually working, if that helps: http://logs.openstack.org/49/431649/6/check/gate-dsvm-nodepool/c796757/console.html#_2017-02-10_16_38_09_806118 | 19:18 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNodes() to support zookeeper https://review.openstack.org/432429 | 19:20 |
| pabelanger | Shrews: okay, have a review of ^ | 19:20 |
| pabelanger | that should pass your test_list_nodes now | 19:20 |
| pabelanger | however, don't know how to properly get getNeededNodes() now | 19:21 |
| Shrews | pabelanger: that might be a bit racey with the node state there. maybe we should have waitForNodes take a label to wait for, and return when state==READY? | 19:25 |
| Shrews | if node.type == label and node.state == zk.READY: return | 19:25 |
| pabelanger | Shrews: Seems sane, we can add a secondary check | 19:25 |
| Shrews | to be more complete | 19:25 |
| jesusaur | Shrews: my concern is around what happens when NodeRequestHandler sees that no nodes are available and launches a nodeset; can't that fulfilled request then get picked up in removeCompletedRequests and then that nodeset is no longer associated with the request? | 19:33 |
| jesusaur | I guess I'm confused about the path from state READY to state USED | 19:34 |
| Shrews | jesusaur: That's EXACTLY what we want to happen with min-ready. We don't want the nodes associated with a request. We're just preparing them for some other request down the line. | 19:35 |
| Shrews | jesusaur: What I'm doing there is usurping the node request pipeline to prime the ready nodes | 19:36 |
| Shrews | jesusaur: once I unset allocated_to on the nodes, they'll be available for the requests from zuul | 19:36 |
| jesusaur | Shrews: right, but aren't there two different paths? there are nodes booted for min-ready, and there are nodes booted for a specific request? | 19:37 |
| Shrews | jesusaur: no. they both go through NodeRequestHandler | 19:37 |
| jesusaur | or do all nodes go into the available pool before getting associated with a specific job? | 19:37 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNode() to support zookeeper https://review.openstack.org/432429 | 19:38 |
| pabelanger | Shrews: ^took inspiration from waitForBuild() | 19:38 |
| Shrews | jesusaur: for the min-ready nodes, they are _temporarily_ associated with the request from nodepool. but once I delete that request and reset the node assignment, any zuul request will see them as available for us | 19:39 |
| Shrews | use | 19:39 |
| jesusaur | aha, ok | 19:39 |
| Shrews | pabelanger: except we won't have the node_id to pass in. specifying the label would help to get around that. we just have to iterate through all nodes looking for that label | 19:43 |
| pabelanger | Shrews: ya, there is a little bit of guess work there | 19:45 |
| pabelanger | need some coffee | 19:46 |
| Shrews | pabelanger: i think if you just add a getNodes() call, loop through those, use the type/state check you already have, then that might work | 19:46 |
| *** hashar has joined #zuul | 19:59 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNodes() to support zookeeper https://review.openstack.org/432429 | 20:07 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNodes() to support zookeeper https://review.openstack.org/432429 | 20:26 |
| Shrews | pabelanger: left you a comment on PS3. I think we should rebase my 'list' change on top of yours, too. | 20:27 |
| pabelanger | Shrews: sure, just testing the patch with dib_pause tests, ATM | 20:30 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Re-enable test_dib_image_pause / test_dib_image_upload_pause https://review.openstack.org/432447 | 20:32 |
| pabelanger | Shrews: left a comment, also added some more code in ps4 | 20:34 |
| Shrews | pabelanger: yes, it would be a bit problematic if we have configs that launch more than one node of that type. I'm not sure how else to handle it, TBH. | 20:35 |
| pabelanger | I think ps4 adds the 2nd check you are looking for | 20:36 |
| pabelanger | this lets us limit things to a specific label | 20:36 |
| pabelanger | 432447 is a follow up to see it in action | 20:36 |
| Shrews | pabelanger: your PS4 is doing pretty much the exact same thing as what I suggested. In fact, you could reduce PS4 to just the 2nd while-loop | 20:39 |
| Shrews | all of our test .yaml files have min-ready = 1, fwiw. except for the subnode.yaml which is useless now | 20:41 |
| pabelanger | okay | 20:42 |
| pabelanger | let me do that | 20:42 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNodes() to support zookeeper https://review.openstack.org/432429 | 20:43 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update waitForNodes() to support zookeeper https://review.openstack.org/432429 | 20:48 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Update nodepool 'list' command https://review.openstack.org/431647 | 20:48 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: WIP: Implement node launching https://review.openstack.org/431523 | 20:48 |
| Shrews | jeblair: So, we right now assume that all of the node types given in a NodeRequest can be handled by a single provider. If that's not what we wanted, I might have designed us into a corner here. | 20:48 |
| jeblair | Shrews: nope, that's correct. we want that (and further, we actually want them in the same az, but like i said earlier, that can wait) | 20:49 |
| pabelanger | okay, I think I rebased correctly | 20:49 |
| jeblair | Shrews: that's desirable so that, for example, devstack multinode tests which require close network proximity will work | 20:49 |
| Shrews | pabelanger: uh oh. no, you undid my last PS to 431523 (which had two +2's and a +1) :( | 20:50 |
| pabelanger | grr | 20:50 |
| pabelanger | not sure why that happened | 20:50 |
| Shrews | jeblair: oh *phew* | 20:50 |
| pabelanger | Oh: https://review.openstack.org/#/c/431523/13..14/nodepool/nodepool.py | 20:50 |
| pabelanger | I have no idea how I pulled that in | 20:50 |
| Shrews | pabelanger: perhaps we should undo that and get it merged to prevent this again ;) | 20:51 |
| jeblair | Shrews: (however, we'll probably want to tweak the min-ready algorithm once we do that so we don't end up making giant requests that all get filled from the same place) | 20:51 |
| pabelanger | Shrews: ya | 20:51 |
| pabelanger | lets do that | 20:51 |
| Shrews | jeblair: same place? it should be random which provider satisfies the request | 20:52 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool feature/zuulv3: Implement node launching https://review.openstack.org/431523 | 20:52 |
| pabelanger | reverts to PS13 | 20:52 |
| Shrews | jeblair: oh, you mean if like min-ready is 50 or something | 20:52 |
| jeblair | Shrews: exactly | 20:52 |
| Shrews | didn't think of that. we could break up the requests later | 20:53 |
| jeblair | Shrews: ours are on the order of about 20 | 20:53 |
| jeblair | Shrews: yeah. it's not urgent. | 20:53 |
| Shrews | pabelanger: thx again | 20:53 |
| pabelanger | I'll learn git one day | 20:54 |
| Shrews | pabelanger: feel free to +A it :) | 20:54 |
| pabelanger | jeblair: have a few minutes to look at 432429, Shrews and I have been going back and forth on that. I think we have it right now, but side affect is we'd potentially only launch 1 ready node, if we had more then one | 20:54 |
| jeblair | pabelanger: i'll look at it a bit later, i have to afk for a bit | 20:55 |
| pabelanger | ack | 20:55 |
| Shrews | pabelanger: i think your test is fooling you | 21:00 |
| Shrews | it's not actually being called | 21:00 |
| Shrews | left comments | 21:00 |
| pabelanger | Shrews: ya, i'll squash it | 21:02 |
| openstackgerrit | Antoine Musso proposed openstack-infra/zuul master: (WIP) mimic clone when directory exists https://review.openstack.org/432477 | 22:03 |
| openstackgerrit | Antoine Musso proposed openstack-infra/zuul master: (WIP) mimic clone when directory exists https://review.openstack.org/432477 | 22:06 |
| hashar | ^ ignore my patches :) | 22:10 |
| *** Guest34532 has joined #zuul | 22:30 | |
| *** Guest34532 has quit IRC | 22:33 | |
| *** hashar has quit IRC | 22:38 | |
| *** jamielennox is now known as jamielennox|away | 22:58 | |
| *** nt has quit IRC | 23:06 | |
| *** nt has joined #zuul | 23:10 | |
| *** jamielennox|away is now known as jamielennox | 23:50 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!