| dmsimard | fungi: ah, yes, I had found the "executing local code is prohibited" but I'm not sure where it's coming from :) | 00:14 |
|---|---|---|
| dmsimard | ah, perhaps it's from use-cached-repos | 00:20 |
| dmsimard | oh, no, duh it's right there above the error mirror-workspace-git-repos | 00:21 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 00:24 |
| dmsimard | jeblair, mordred: seems like half of the roles contain something that runs on the executor and end up being prohibited because they're running out of the trusted context (congrats security) | 00:28 |
| dmsimard | so it looks like the best course of action is to go multinode with an ansible-playbook command :/ | 00:28 |
| jeblair | dmsimard: the mirror update role should not need to run in the trusted context | 00:29 |
| clarkb | jeblair: fungi Ithink I figured it out. The shallow copy of the nested default logging config means that there is a race in updating the shared handlers dict | 00:30 |
| clarkb | jeblair: I think a trivial fix for that is to use a deepcopy there. I'll push that up and we can discuss in gerrit | 00:30 |
| dmsimard | jeblair: I figured I might as well run everything that was included in base in one large integration job .. but nope. | 00:30 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 00:31 |
| jeblair | dmsimard: yeah, it may be best to just focus on the mirror role since that's the issue at the moment. constructing a pseudo executor is going to be a bit more work. | 00:32 |
| dmsimard | jeblair: I'm finding issues unrelated to mirrors with my attempts, such as opensuse not being able to run this module https://github.com/openstack-infra/zuul-jobs/blob/master/roles/validate-host/library/zuul_debug_info.py | 00:32 |
| jeblair | dmsimard: that one should also be okay in untrusted context | 00:33 |
| dmsimard | jeblair: apparently not | 00:33 |
| dmsimard | jeblair: http://logs.openstack.org/81/501281/9/check/base-integration-centos-7/5cccc7a/job-output.txt.gz#_2017-09-07_00_25_20_477442 | 00:33 |
| jeblair | dmsimard: i mean the test should function, not that it should pass | 00:33 |
| dmsimard | jeblair: oh, for the module it just spits a MODULE ERROR so it's probably a missing python dep or something. | 00:34 |
| jeblair | dmsimard: i mean to say that you can use the same approach as for the mirror config, you don't need to run that in the trusted context on the executor. | 00:34 |
| dmsimard | or just the command syntax on suse being different | 00:34 |
| jeblair | dmsimard: basically -- i think we're going to find that more or less by definition, the base roles that run on the nodes and don't need special access are the ones that are going to be suceptible to node differences and need more work+testing | 00:35 |
| dmsimard | oh, hum, the console output is not the same in the .txt.gz | 00:35 |
| dmsimard | than in the web console | 00:35 |
| jeblair | it's the same file | 00:35 |
| jeblair | i have to run; happy to help more tomorrow | 00:36 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul feature/zuulv3: Use deepcopy when copying ansible logging config https://review.openstack.org/501499 | 00:36 |
| clarkb | jeblair: fungi ^ thats the fix I think | 00:36 |
| clarkb | another alternative is to go from str to dict to str (json) | 00:38 |
| fungi | oh, yeah, classic mutable object reuse | 00:39 |
| fungi | er, i guess not. i misread. still, great find | 00:41 |
| *** harlowja has quit IRC | 00:48 | |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 01:01 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul feature/zuulv3: Use deepcopy when copying ansible logging config https://review.openstack.org/501499 | 01:03 |
| clarkb | I'm a derp | 01:03 |
| fungi | i was trying to figure out why that was failing | 01:04 |
| fungi | oh, hah | 01:05 |
| fungi | digging through 60k pages in lynx is kinda painful | 01:05 |
| jamielennox | is -infra putting tox in the image? i assume so | 01:16 |
| jamielennox | pabelanger: if you pip install --user tox here: http://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/ensure-tox/tasks/main.yaml#n2 | 01:17 |
| jamielennox | then the $PATH is wrong in command, here: http://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/tox/tasks/main.yaml#n14 | 01:18 |
| clarkb | jamielennox: yes we are | 01:18 |
| clarkb | it gets pulled in by our infra-package-needs dib element iirc | 01:18 |
| jamielennox | clarkb: yea, i found it, it's interesting that it's just pip installed into root as opposed to like bindeps which is into a venv | 01:18 |
| jamielennox | just found it | 01:19 |
| jamielennox | i was expecting tox_executable to be set somewhere | 01:19 |
| fungi | we'd been installing tox into the system environment for far longer | 01:20 |
| fungi | i'm not personally keen on installing python packages that way given how they can drag in dependencies and the notorious conflicts between distro packages on some platforms | 01:21 |
| jamielennox | yea, installing into root is always bad, but ok, it's just something that hasn't been updated | 01:22 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 01:35 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 01:45 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Handle floating ips in multi-node-known-hosts https://review.openstack.org/501459 | 01:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 01:53 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 01:57 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: WIP: Make the base playbooks/roles work for every supported distro https://review.openstack.org/501281 | 02:02 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration test jobs for the configure-mirror role https://review.openstack.org/501534 | 02:46 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 02:46 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 02:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for debian in configure-mirrors https://review.openstack.org/501537 | 02:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for fedora in configure-mirrors https://review.openstack.org/501538 | 02:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for opensuse in configure-mirrors https://review.openstack.org/501539 | 02:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration test jobs for the configure-mirror role https://review.openstack.org/501534 | 02:54 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Fix lint issue https://review.openstack.org/501540 | 02:54 |
| Shrews | jeblair: mordred: pabelanger: before I fall asleep and forget, we should probably consider upgrading the nodepool builders to the v3 branch code as part of the upgrade. | 02:58 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 02:59 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 02:59 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for debian in configure-mirrors https://review.openstack.org/501537 | 02:59 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for fedora in configure-mirrors https://review.openstack.org/501538 | 02:59 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for opensuse in configure-mirrors https://review.openstack.org/501539 | 02:59 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration tests for validate-host https://review.openstack.org/501543 | 03:12 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration tests for validate-host https://review.openstack.org/501543 | 03:13 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 03:13 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 03:14 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for debian in configure-mirrors https://review.openstack.org/501537 | 03:14 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for fedora in configure-mirrors https://review.openstack.org/501538 | 03:14 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for opensuse in configure-mirrors https://review.openstack.org/501539 | 03:14 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration tests for validate-host https://review.openstack.org/501543 | 03:14 |
| dmsimard | Sorry for the spam, screwed up a rebase twice | 03:15 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Use deepcopy when copying ansible logging config https://review.openstack.org/501499 | 03:21 |
| dmsimard | clarkb, jeblair, mordred, pabelanger: There's a patch tree to add support for configure-mirrors with the supported distros here: https://review.openstack.org/#/q/status:open+topic:zuulv3+project:openstack-infra/zuul-jobs .. first change is 501534, last change is 501539 | 03:21 |
| dmsimard | I'll work on adding the /etc/ci/mirror_info file tomorrow and I've also started looking at validate-host which is challenging because it runs a lot of stuff on the executor side https://review.openstack.org/#/c/501543/ | 03:22 |
| dmsimard | I know so far that the custom zuul_debug_info module doesn't work on suse: http://logs.openstack.org/43/501543/3/check/base-integration-opensuse423/9375b75/job-output.txt.gz#_2017-09-07_03_18_35_776512 (could we just run command/shell tasks? we could also consider using ara_record I guess) | 03:24 |
| *** harlowja has joined #zuul | 04:06 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Handle some common yaml syntax errors https://review.openstack.org/501486 | 04:09 |
| *** yolanda has quit IRC | 04:15 | |
| *** timrc has quit IRC | 04:20 | |
| *** timrc has joined #zuul | 04:21 | |
| *** xinliang has quit IRC | 04:21 | |
| *** cinerama has quit IRC | 04:21 | |
| *** cinerama has joined #zuul | 04:21 | |
| *** xinliang has joined #zuul | 04:21 | |
| *** xinliang has quit IRC | 04:21 | |
| *** xinliang has joined #zuul | 04:21 | |
| *** tflink has quit IRC | 04:28 | |
| *** tflink has joined #zuul | 04:35 | |
| *** tflink has quit IRC | 04:53 | |
| *** tflink has joined #zuul | 04:59 | |
| ianw | dmsimard: current we don't have anything making mirror_info? | 05:04 |
| ianw | currently? | 05:04 |
| *** harlowja has quit IRC | 05:11 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Delete unused run-cover role https://review.openstack.org/501244 | 05:51 |
| openstackgerrit | Jamie Lennox proposed openstack-infra/zuul-jobs master: Fix fetching tox output for empty dirs https://review.openstack.org/501572 | 05:57 |
| *** jianghuaw has joined #zuul | 06:17 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: launcher: add cancel job function https://review.openstack.org/501581 | 06:35 |
| *** hashar has joined #zuul | 07:38 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: Use password supplied from nodepool https://review.openstack.org/500823 | 08:20 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: Use username from node information if available https://review.openstack.org/453983 | 08:20 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: Rename ssh_port to connection_port https://review.openstack.org/500799 | 08:20 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Fix nodepool's docs about config-drive https://review.openstack.org/500391 | 09:13 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test https://review.openstack.org/501643 | 09:20 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul feature/zuulv3: DNM: test https://review.openstack.org/501643 | 09:25 |
| *** jkilpatr has quit IRC | 10:40 | |
| *** electrofelix has quit IRC | 11:07 | |
| *** jkilpatr has joined #zuul | 11:15 | |
| dmsimard | ianw: no, it's on the to-do | 11:41 |
| dmsimard | there's a tox-pep8 job that has been stuck for a good while now, probably worth looking into: http://zuulv3.openstack.org/ | 12:48 |
| dmsimard | http://zuulv3.openstack.org/static/stream.html?uuid=25439f4aee1a47ff8f3bf791fd58eeb9&logfile=console.log | 12:48 |
| mordred | 2017-09-07 04:10:06,748 DEBUG zuul.AnsibleJob: [build: 71489da8e8244e4eb813b2d56893d4a3] Sending result: {"result": "ERROR", "error_detail": "Unable to find playbook /var/lib/zuul/builds/71489da8e8244e4eb813b2d56893d4a3/trusted/project_0/git.openstack.org/openstack-infra/project-config/base/ensure-tox"} | 13:03 |
| mordred | followed by | 13:03 |
| mordred | 2017-09-07 04:10:06,904 ERROR zuul.AnsibleJob: [build: 71489da8e8244e4eb813b2d56893d4a3] Error cleaning up jobdir: | 13:03 |
| mordred | Traceback (most recent call last): | 13:04 |
| mordred | File "/usr/local/lib/python3.5/dist-packages/zuul/executor/server.py", line 912, in execute | 13:04 |
| mordred | self.jobdir.cleanup() | 13:04 |
| mordred | ... | 13:04 |
| mordred | FileNotFoundError: [Errno 2] No such file or directory: 'ae4368a4459f901e6b3d6e6e1ad544ba4652cd' | 13:04 |
| dmsimard | mordred: for that tox-pep8 job thing ? | 13:08 |
| mordred | yah | 13:14 |
| dmsimard | looks like we need to do a better job at cleanly catching that | 13:16 |
| mordred | yah - well - | 13:41 |
| mordred | there are two things I don't understand | 13:41 |
| openstackgerrit | Dirk Mueller proposed openstack-infra/zuul-jobs master: Expand PATH for SUSE systems to include /usr/sbin https://review.openstack.org/501737 | 13:42 |
| *** electrofelix has joined #zuul | 13:42 | |
| openstackgerrit | Dirk Mueller proposed openstack-infra/zuul-jobs master: Expand PATH for SUSE systems to include /usr/sbin https://review.openstack.org/501737 | 13:45 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Make validate-host read from site-variables https://review.openstack.org/500592 | 13:57 |
| openstackgerrit | Dirk Mueller proposed openstack-infra/zuul-jobs master: Expand PATH for SUSE systems to include /usr/sbin https://review.openstack.org/501737 | 14:06 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul-jobs master: Create fetch-zuul-cloner role. https://review.openstack.org/501749 | 14:09 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul feature/zuulv3: Remove zuul-cloner shim https://review.openstack.org/501750 | 14:09 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul feature/zuulv3: Remove zuul-cloner shim https://review.openstack.org/501750 | 14:10 |
| Shrews | jeblair: mordred: shim moved to a role ^^^ Is there a zuul variable that specifies the zuul source repo on the nodes? I created a role var for it because I couldn't really find one to use (closest was zuul.project.src_dir) | 14:12 |
| Shrews | mordred: yoo iz fast | 14:14 |
| *** dkranz has joined #zuul | 14:14 | |
| mordred | Shrews: it's just '~/src' | 14:15 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix lint issue https://review.openstack.org/501540 | 14:15 |
| Shrews | mordred: should I add that as a default for the variable? | 14:15 |
| mordred | oh - wait - I guess for us in this context it's '~/src/git.openstack.org' right? because you're looking for 'openstack-infra/shade' | 14:16 |
| Shrews | mordred: oh actually, it would need the "git.openstack.org" part | 14:16 |
| Shrews | yah that | 14:16 |
| mordred | yah | 14:16 |
| Shrews | yah yah | 14:16 |
| mordred | yah- let's just set the default for it to that | 14:18 |
| mordred | Shrews: we can make it more fancy later if we need to | 14:18 |
| Shrews | mordred: can't use that as the default... it's openstack specific | 14:19 |
| Shrews | (no-no for zuul-jobs) | 14:19 |
| Shrews | i'll just leave it as-is i guess | 14:20 |
| jeblair | we could just not have a default, and require that users supply it for the role; so when we add that to the base job, we put ~zuul/src/git.openstack.org | 14:20 |
| mordred | ah. good point. SO ... we could set the default to "{{ zuul_site_legacy_cloner_src_dir }}" | 14:20 |
| mordred | and set zuul_site_legacy_cloner_src_dir: ~zuul/src/git.openstack.org in site-variables | 14:20 |
| mordred | but we can also just do jeblair's thing | 14:21 |
| Shrews | jeblair's thing is already done, so that gets my vote :) | 14:21 |
| mordred | awesome | 14:21 |
| jeblair | mordred: didn't we use apt-get update in https://review.openstack.org/501540 because the apt python module isn't installable until we do that? | 14:22 |
| clarkb | mordred that no such file or directpry thing looks similar to the problem I in logging.json config I pushed patch to fix yesterday. Basically crossed streams via shared dict. Though I dont think the work dir itself could be shared that way | 14:23 |
| mordred | jeblair: oh! yes - I'd forgotten abot that - maybe we need to put in a comment there | 14:23 |
| jeblair | https://review.openstack.org/492716 | 14:23 |
| jeblair | yep | 14:23 |
| jeblair | mordred: i will revert with comment | 14:23 |
| jeblair | dmsimard: ^ | 14:23 |
| mordred | jeblair: ++ | 14:24 |
| dmsimard | jeblair: ah eh.. then we need to add a skip lint tag | 14:24 |
| dmsimard | jeblair: I had to send that in because the tox-linters job was failing | 14:24 |
| dmsimard | either add a skip lint tag or add an exception for that ansible-lint error | 14:24 |
| jeblair | dmsimard: let's add an exception | 14:24 |
| dmsimard | jeblair: https://github.com/willthames/ansible-lint#false-positives | 14:24 |
| jeblair | that's not linting, that's a computer telling a person how to write a program and getting it wrong | 14:25 |
| dmsimard | let me see what error code it was ... | 14:25 |
| dmsimard | jeblair: eh, it's a best practice to use ansible modules where possible :) | 14:25 |
| jeblair | dmsimard: yep, and where not possible is a surprisingly large amount of time | 14:25 |
| dmsimard | jeblair: looks like ANSIBLE0006 | 14:27 |
| dmsimard | http://logs.openstack.org/34/501534/1/check/tox-linters/3592e80/job-output.txt.gz#_2017-09-07_02_50_17_484062 | 14:27 |
| dmsimard | I'll send a revert with an exception added in | 14:27 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Revert "Fix lint issue" https://review.openstack.org/501753 | 14:27 |
| mordred | jeblair: http://logs.openstack.org/62/500162/1/gate/tox-py35/213272a/ been sporadically hitting this test failure - I tried writing https://review.openstack.org/#/c/501266/ - but that doesn't seem to work :) | 14:28 |
| mordred | jeblair: any thoughts on what I should look at? | 14:28 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Revert "Fix lint issue" https://review.openstack.org/501754 | 14:28 |
| jeblair | dmsimard: ^ feel free to amend that if necessary | 14:29 |
| dmsimard | jeblair: oops, had already sent in a revert | 14:29 |
| jeblair | dmsimard: mine has a comment | 14:30 |
| dmsimard | jeblair: okay :P | 14:30 |
| Shrews | mordred: oh, ewww | 14:30 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Revert "Fix lint issue" https://review.openstack.org/501754 | 14:30 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Expand PATH for SUSE systems to include /usr/sbin https://review.openstack.org/501737 | 14:32 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration tests for validate-host https://review.openstack.org/501543 | 14:32 |
| jeblair | mordred: part of your change uses create_connection, but part of it just switches to inet6 | 14:32 |
| jeblair | mordred: what happens when you run test_websocket_streaming locally? | 14:33 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Make validate-host read from site-variables https://review.openstack.org/500592 | 14:37 |
| dmsimard | mordred: btw when troubleshooting a suse issue I found this: http://logs.openstack.org/43/501543/3/check/base-integration-opensuse423/9375b75/job-output.txt.gz#_2017-09-07_03_18_35_776512 but in the json there's the real OSError exception: http://logs.openstack.org/43/501543/3/check/base-integration-opensuse423/9375b75/job-output.json | 14:37 |
| dmsimard | is there a particular reason why the OSError isn't fed back to the console/job output ? | 14:37 |
| mordred | jeblair: yah - I just switched to AF_INET6 for server connections, and to create_connection for clients | 14:38 |
| mordred | dmsimard: nope. that is a bug - it shuold, in fact, print the exception - and we have integration tests for that :) | 14:38 |
| dmsimard | well isn't that unfortunate | 14:39 |
| mordred | yah | 14:39 |
| dmsimard | mordred: it's reproducible while running validate-host without https://review.openstack.org/#/c/501737/ -- I'll write it down somewhere so we can look at it later | 14:40 |
| mordred | dmsimard: cool | 14:40 |
| mordred | dmsimard: http://logs.openstack.org/62/500162/1/check/zuul-stream-functional/e878cc7/stream-files/stream-job-output.txt if you search for MODULE FAILURE in there you'll see it | 14:40 |
| jeblair | mordred: i think i see the problem. you changed one of the clients to inet6 instead of using create connection. left comment. | 14:40 |
| mordred | jeblair: oh - piddle. thanks | 14:41 |
| dmsimard | mordred: maybe something specific to SUSE? but that'd be weird, I guess we can easily test it, let me send a test patch | 14:41 |
| jeblair | dmsimard, clarkb, mordred: so what's going on with jobdirs? | 14:41 |
| dmsimard | jeblair: uncaught exception afaict? mordred pulled an exception out: http://eavesdrop.openstack.org/irclogs/%23zuul/%23zuul.2017-09-07.log.html#t2017-09-07T13:03:31 | 14:43 |
| dmsimard | looks like he's currently fixing that exception by getting rid of ensure-tox :P | 14:43 |
| jeblair | okay, i'll look at those logs and see if i can learn more | 14:43 |
| dmsimard | I haven't had the time to take a look at it yet | 14:43 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Update tests to use AF_INET6 https://review.openstack.org/501266 | 14:45 |
| mordred | jeblair: yah - unsure - there's an exception in the logs related to jobdir - but what's more troubling is the job that's just hung | 14:45 |
| mordred | jeblair: the exception in the logs looks to me like a thing that would just cause the job to abort | 14:46 |
| *** hashar is now known as hasharAway | 14:47 | |
| jeblair | that error is pretty disturbing on its own | 14:51 |
| jeblair | clarkb: did you have a theory that tied it to that test failure? | 14:51 |
| clarkb | jeblair: https://review.openstack.org/#/c/501499/ was a similar issue I fixed | 14:51 |
| clarkb | thinking about it I dont think the streams could cross like that outside the logging confs as it relied on the shared dict refs being used | 14:52 |
| clarkb | the top level JobDir attributes shouldnt be shared like that | 14:52 |
| dmsimard | need +5 on https://review.openstack.org/#/c/501754/ | 14:53 |
| jeblair | clarkb: yeah, we rmtree jobdir.root which is a plain string; i think the only possibility for racing there is if we have two builds with the same uuid | 14:54 |
| jeblair | and the executor claims only to have gotten one job with that uuid | 14:55 |
| jeblair | mordred: the item that job ran for was removed; the stuck item looks like something else | 14:57 |
| mordred | jeblair: oh - yah - sorry - I forgot that I poked around a little bit on item one - that playbook it's referencing legitimately did not exist | 14:59 |
| mordred | and I have a patch up to project-config to fix that error | 15:00 |
| mordred | https://review.openstack.org/#/c/501742/ | 15:00 |
| jeblair | we have 3 zuul-executor processes running | 15:00 |
| jeblair | normally there are 2 | 15:00 |
| mordred | jeblair: oh! so potentially an orphan subprocess from a restart, and thus two zuul-executors stepping on each other? | 15:01 |
| * SpamapS awakens | 15:03 | |
| dmsimard | Anyone else seen a bit of flapping on the "tox-py35-on-zuul" job ? | 15:03 |
| mordred | dmsimard: yah - it's the websocket streaming test which seems to be less happy on ipv6 hosts | 15:04 |
| mordred | dmsimard: https://review.openstack.org/501266 is a stab at fixing | 15:04 |
| dmsimard | mordred: ok, thought it was just me http://logs.openstack.org/43/501543/4/check/tox-py35-on-zuul/86de246/testr_results.html.gz | 15:04 |
| dmsimard | do we have v6 nodes ? I thought OSIC was the only cloud with v6 | 15:05 |
| mordred | nope - it started showing up when we added some clouds with ipv6 | 15:05 |
| mordred | dmsimard: there are many clouds with ipv6 | 15:05 |
| mordred | dmsimard: vexxhost and rackspace both spring to mind - and I think ovh too? | 15:05 |
| mordred | it's an area in which openstack is demostrably WAY better than the non-free clouds that we can't seem to ever get anyone to shout loudly enough about | 15:06 |
| dmsimard | ok then :) | 15:06 |
| jeblair | mordred: yeah. i can't explain how two of them would interfere in a way that would cause this, but then, i was prompted to check because the logs didn't make sense. i'm kind of inclined to throw up my hands and say regardless of whether two instances running is causing the weirdness we're seeing, it's at least too difficult to eliminate it as a possibility. | 15:06 |
| jeblair | i'm going to restart ze01 | 15:06 |
| mordred | jeblair: I agree | 15:06 |
| mordred | jeblair: running two executors on the same dirs is not a thing we've designed for | 15:07 |
| mordred | jeblair: btw - https://review.openstack.org/#/c/501266/ passed tests | 15:07 |
| Shrews | i just saw the tox-py35 job fail, but for different reasons on test_dynamic_dependent_pipeline_failure | 15:08 |
| jeblair | Shrews: that's a new test i just wrote; i may have not written it well enough | 15:08 |
| jeblair | Shrews: link? | 15:08 |
| Shrews | jeblair: http://logs.openstack.org/50/501750/2/check/tox-py35/255c26e/testr_results.html.gz | 15:08 |
| jeblair | mordred: +2. Shrews want to +3 501266 ? | 15:08 |
| jeblair | ze01 restarted | 15:09 |
| Shrews | looking | 15:09 |
| jeblair | ze02 has *one* executor running | 15:09 |
| jeblair | which probably means its streaming thing is inop | 15:09 |
| jeblair | i will restart it as well | 15:09 |
| jeblair | ze02 restarted | 15:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add UPPER_CONSTRAINTS_FILE file if it exists https://review.openstack.org/500320 | 15:10 |
| dmsimard | What makes it so a role goes in project-config, zuul-jobs or openstack-zuul-jobs ? | 15:15 |
| dmsimard | I guess zuul-jobs are for what would make for "generic" zuul jobs, openstack zuul jobs specific to our zuul instance.. but then project-config ? | 15:16 |
| mordred | dmsimard: trusted openstack-specific jobs | 15:17 |
| dmsimard | mordred: ah, okay. | 15:17 |
| mordred | dmsimard: by and large we try to put the majority of the roles in zuul-jobs and make them parameterizable | 15:18 |
| mordred | some things, like roles/legacy-copy-project-config-scripts are very legimitately openstack-only and intended to only exist for transition | 15:18 |
| Shrews | remote: https://review.openstack.org/501769 Add fetch-zuul-cloner role to base job | 15:21 |
| Shrews | i *think* those var values are accurate | 15:21 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Revert "Fix lint issue" https://review.openstack.org/501754 | 15:22 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Update tests to use AF_INET6 https://review.openstack.org/501266 | 15:23 |
| mordred | Shrews: I agree with your var values | 15:24 |
| Shrews | \o/ | 15:24 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Make test_dynamic_dependent_pipeline_failure more robust https://review.openstack.org/501778 | 15:34 |
| jeblair | Shrews: ^ i think that should tidy that up | 15:34 |
| dmsimard | jeblair, mordred, Shrews: there's a tree of patches for configure-mirrors and validate-host for cross-distro support ready for final +3 -- starts at https://review.openstack.org/#/c/501534/ and works it's way up a couple patches deep | 15:34 |
| dmsimard | jeblair: if I hijack your devstack-legacy patch and add a depends-on to the top of that zuul-jobs patch tree, would I be able to test devstack on centos and suse ? | 15:36 |
| jeblair | Shrews: for the zuul-cloner change, can you add it to the base-test job first, then after that lands, pick some job (like py35) and set parent to base-test. the second change doesn't have to land, but just running the tests for it should exercise that role without us having to land a change to 'base' untested | 15:36 |
| jeblair | Shrews: then if that works, you can copy the role over to base | 15:36 |
| dmsimard | jeblair: and by hijack I mean add a review rebased on your patch | 15:36 |
| Shrews | jeblair: oh sure | 15:37 |
| jeblair | dmsimard: i think if you wrote a patch to devstack-gate that reparented devstack-legacy on your job which used base-minimal + the new configure mirror role, that should work yes. | 15:38 |
| dmsimard | jeblair: oh, right, because yours inherits base. Okay, well, at least I can start testing stuff. | 15:39 |
| jeblair | it looks like devstack-legacy has enough +2s, so i +w'd it. but yeah, you can build on top of that while it's landing. | 15:39 |
| jeblair | Shrews: +3d both | 15:41 |
| Shrews | jeblair: thx. and now i understand why we have a base-test :) | 15:42 |
| dmsimard | jeblair: where are the "playbooks/legacy-pre" and "playbooks/legacy-post" playbooks ? I don't see those ? | 15:43 |
| dmsimard | jeblair: nevermind.. code editor didn't pick up new git checkout | 15:43 |
| * dmsimard curses at pycharm | 15:43 | |
| dmsimard | ah, I'm not going to be able to test with minimal -- mirror-workspace-git-repos requires to be run from the executor | 15:45 |
| dmsimard | so I guess I'll need to wait until the zuul-jobs work lands | 15:46 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Fix spacer-lines to work with multi-node and items https://review.openstack.org/500162 | 15:46 |
| Shrews | looks like open source version of Tower was announced: https://www.ansible.com/awx-project-faq | 15:47 |
| dmsimard | \o/ | 15:48 |
| dmsimard | Shrews: https://twitter.com/dmsimard/status/905815117870313473 | 15:49 |
| Shrews | oops. not sure that's supposed to be public just yet.... ^D all the things | 15:50 |
| jlk | AWX again? | 15:50 |
| jlk | hahaa | 15:50 |
| dmsimard | Shrews: it's leaked all over the place already | 15:50 |
| Shrews | oh, yeah. that was from yesterday's email phew | 15:51 |
| dmsimard | Michael and Robyn have tweeted about it, amongst other leaks | 15:51 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Create fetch-zuul-cloner role. https://review.openstack.org/501749 | 15:52 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Make test_dynamic_dependent_pipeline_failure more robust https://review.openstack.org/501778 | 15:57 |
| pabelanger | Shrews: mordred jeblair: re: nodepool feature/zuulv3, should we merge master into feature/zuulv3 and prepare to merge feature/zuulv3 back to master? | 16:00 |
| mordred | pabelanger: yes - I think I did a recent master->feature/zuulv3 so it shouldn't be too hard | 16:01 |
| Shrews | pabelanger: mordred: jeblair: oh that reminds me... see my note about migrating the builders to v3 code? | 16:02 |
| Shrews | something we hadn't considered | 16:02 |
| Shrews | 2.5 should be very similar to v3, but still.... | 16:02 |
| pabelanger | Ya, maybe we should just stand up nb01.o.o as feature/zuulv3 and leave nb03 / nb04 for now | 16:03 |
| Shrews | ++ | 16:04 |
| pabelanger | let me see if I can merge master to feature/zuulv3, without paste bombing 1000s patches | 16:04 |
| *** persia has quit IRC | 16:07 | |
| SpamapS | What's AWX? | 16:08 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul-jobs master: DNM: testing cloner shim https://review.openstack.org/501789 | 16:10 |
| mordred | SpamapS: it's the new project that is the open source version of tower | 16:10 |
| mordred | SpamapS: that is being announced this week but seems to have been leaked already | 16:10 |
| SpamapS | Ah | 16:11 |
| SpamapS | never heard of it :-P | 16:11 |
| mordred | think AWX -> Tower like Fedora -> RHEL fwiw - same general approach/setup/thinking | 16:11 |
| mordred | SpamapS: well, that's not surprising, as it hasn't actually been announced yet :) | 16:11 |
| mordred | or, rather, perhaps was only just announced today | 16:11 |
| SpamapS | I heard THOSE words uttered last night | 16:11 |
| SpamapS | but never WHAT hadn't been announced :-P | 16:14 |
| Shrews | jeblair: seems /usr/zuul-env/bin does not exist on the nodes. I guess I should change the fetch-zuul-cloner role to just create it? | 16:16 |
| dmsimard | mordred: yes, it was leaked all over the place today, afaik it's announced at ansiblefest (if it hasn't been already) | 16:16 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Make validate-host read from site-variables https://review.openstack.org/500592 | 16:18 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: Remove opensuse-42.2 and fedora-25 from nodepool dsvm jobs https://review.openstack.org/501793 | 16:19 |
| mordred | Shrews: oh - that's very strange - it should exist | 16:19 |
| pabelanger | Shrews: mordred: ^ something we can skip for merging master to feature/zuulv3 branch | 16:19 |
| mordred | Shrews: we should be baking that in to the nodes | 16:19 |
| mordred | Shrews: in dib | 16:20 |
| mordred | Shrews: nodepool/elements/nodepool-base/install.d/89-install-zuul in project-config | 16:20 |
| mordred | pabelanger: +A | 16:21 |
| Shrews | mordred: umm, i'm baffled then | 16:21 |
| mordred | Shrews: where is the fail? | 16:22 |
| Shrews | mordred: failed on https://review.openstack.org/501789 I would paste the log, but the jobs kept failing over and over for some reason. I was watching live logs | 16:22 |
| Shrews | mordred: 2017-09-07 16:16:59.861230 | TASK [fetch-zuul-cloner : Install zuul-cloner shim] | 16:23 |
| Shrews | 2017-09-07 16:17:00.772532 | ubuntu-xenial -> localhost | ERROR: Destination directory /usr/zuul-env/bin does not exist | 16:23 |
| mordred | Shrews: let's add an autohold on a job and go poke on a node | 16:24 |
| mordred | also - why did those retry_limit | 16:26 |
| SpamapS | ok, python SRU re-uploaded, including to zesty now | 16:28 |
| pabelanger | nice | 16:29 |
| SpamapS | weird not having a linux laptop.. but actually I have a big cloud so I can do my work there... | 16:30 |
| SpamapS | only thing I don't like about packaging in the cloud is that I still have to download stuff to sign it. | 16:30 |
| SpamapS | aaaaand now I see zesty has python 3.6.1 which also needs that fix. Arggghh | 16:34 |
| mordred | SpamapS: ugh. | 16:35 |
| mordred | SpamapS: feels like you should be able to figure out something using GPG_AGENT and ssh to let you sign things remotely over your ssh connection | 16:36 |
| jlk | SpamapS: why package in the cloud instead of in a local container or VM? | 16:36 |
| SpamapS | mordred: I don't trust that :) | 16:37 |
| SpamapS | jlk: cause I don't really know how to do that yet :) | 16:37 |
| Shrews | mordred: i was assuming they kept retrying b/c the base pre-run playbook kept failing, but i'm not sure. let me set the autohold... | 16:37 |
| SpamapS | still re-learning mac | 16:37 |
| jlk | yeah that's fiar. | 16:37 |
| jlk | fair | 16:37 |
| SpamapS | dockering isn't a bad idea actually | 16:37 |
| SpamapS | I do kinda know that | 16:37 |
| jlk | Docker for Mac has gotten fairly good. It still uses a little Linux VM to do the dockering, but it's nearly seamless | 16:38 |
| SpamapS | That's totally fine | 16:38 |
| SpamapS | just don't make me learn all the vm things :-P | 16:38 |
| SpamapS | aaaand the patch doesn't cleanly apply | 16:38 |
| mordred | SpamapS: of course it doesn't | 16:41 |
| mordred | SpamapS: oh - although there is probably an upstream 3.6 version of the patch | 16:41 |
| mordred | SpamapS: since upstream ported it to 3.6 too | 16:41 |
| mordred | SpamapS: so rather than strictly porting the quilt patch I made, you might be better off refreshing that patch in the 3.6 branch wiht a clean applicatoin of the upstream version for 3.6 | 16:42 |
| jlk | SpamapS: yeah the docker app hides it. You just launch "docker" app and you get a task bar icon to interact with, and you can then just use "docker" commands on the CLI | 16:42 |
| SpamapS | Yeah I just have to find that. :-P | 16:42 |
| Shrews | mordred: 198.72.124.220 .... and that path exists | 16:43 |
| Shrews | jeblair: also, you have a held node (in case you're done with it) | 16:43 |
| mordred | Shrews: got it | 16:46 |
| mordred | Shrews: https://review.openstack.org/#/c/501749/1/roles/fetch-zuul-cloner/tasks/main.yaml | 16:46 |
| mordred | Shrews: I believe those both need "become: yes" | 16:46 |
| Shrews | ah ha | 16:46 |
| mordred | Shrews: I think that error message ansible is giving us is total bong | 16:47 |
| mordred | Shrews: OH wait | 16:49 |
| mordred | Shrews: delegate_to: localhost | 16:49 |
| mordred | Shrews: so - multiple things | 16:49 |
| mordred | Shrews: remove delegate_to: localhost | 16:49 |
| mordred | Shrews: add become: yes to both tasks | 16:49 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul-jobs master: Use elevated privs for fetch-zuul-cloner https://review.openstack.org/501807 | 16:50 |
| mordred | Shrews: and add a file: path: "{{ destination | dirname }}" state: directory become: yes as a first task | 16:50 |
| Shrews | k. 1 sec | 16:50 |
| mordred | Shrews: so that we can stop installing that venv once v3 is out | 16:50 |
| Shrews | mordred: should that first task use become: yes also? probably so | 16:52 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul-jobs master: Fixes for fetch-zuul-cloner https://review.openstack.org/501807 | 16:53 |
| Shrews | mordred: i think ^^^ covers all that | 16:54 |
| mordred | Shrews: ++ | 16:54 |
| mordred | clarkb: ^^ could you +A that real quick? | 16:54 |
| clarkb | we are installing a python script that is a jinja2 template? fun | 16:55 |
| clarkb | is it actually a template or is that how ansible looks for the content? | 16:56 |
| Shrews | clarkb: actual template for specifying the source directory for the on-node repos | 16:56 |
| Shrews | clarkb: https://review.openstack.org/#/c/501769/2/playbooks/base-test/pre.yaml | 16:57 |
| clarkb | couldn't that be an argument when caling the script? | 16:57 |
| Shrews | clarkb: it has to be backward compat with how current jobs are using it | 16:58 |
| clarkb | oh right | 16:58 |
| clarkb | fun | 16:58 |
| openstackgerrit | Merged openstack-infra/nodepool master: Remove opensuse-42.2 and fedora-25 from nodepool dsvm jobs https://review.openstack.org/501793 | 17:13 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Fixes for fetch-zuul-cloner https://review.openstack.org/501807 | 17:15 |
| *** persia has joined #zuul | 17:16 | |
| Shrews | mordred: yay, looks like those fixes worked | 17:17 |
| Shrews | gotta step away for a bit | 17:17 |
| *** harlowja has joined #zuul | 17:19 | |
| dmsimard | clarkb: should there be a reminder on openstack-dev about the zuul migration ? | 17:49 |
| clarkb | dmsimard: probably. But I'll defer to mordred and jeblair on that as far as preparedness goes | 17:50 |
| dmsimard | aye, my next question was when/where this migration was going to occur -- I'm landing in denver sunday around noon and have nothing special planned so I can probably lend a hand | 17:51 |
| clarkb | plan was to do first attempt saturday evening then roll back and use what that teaches us to make proper rollout sunday/monday | 17:51 |
| clarkb | probably not a bad idea to have a concreter plan here soon | 17:52 |
| clarkb | I know mordred and fungi has tc+board meeting things all day sunday but other than that I expect we'll be flexible? I land saturday at 4pm ish | 17:52 |
| dmsimard | ah, I'm probably not going to be able to help saturday evening, my flight is at 6AM sunday morning so I'll need the sleep.. waking up at 3AM is gonna be brutal | 17:52 |
| *** electrofelix has quit IRC | 17:53 | |
| jlk | I can be around Saturday evening depending on time. But I'm out Sunday due to memorial service, and I'm traveling to DEN on Monday. | 17:54 |
| dmsimard | still need a +3 on a series of patches to unblock my devstack tests on suse/centos/fedora when anyone has a chance, starts here: https://review.openstack.org/#/c/501534/ | 17:54 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Don't install bindep if there's no bindep file https://review.openstack.org/501018 | 18:00 |
| jeblair | let's regroup tomorrow and figure out what we're doing and when | 18:07 |
| jeblair | dmsimard: jlk has questions on 501534 | 18:09 |
| dmsimard | looking | 18:11 |
| jeblair | dmsimard: do you want to land that change, or are you just using it to drive testing? | 18:13 |
| dmsimard | jeblair: yeah I'm looking to land that whole thing | 18:13 |
| dmsimard | unless there's a reason not to | 18:13 |
| dmsimard | jlk: replied to your comments | 18:13 |
| jlk | okay | 18:16 |
| jlk | will look again | 18:16 |
| jeblair | dmsimard: okay, left some suggestions on where i think we sholud move those jobs. | 18:17 |
| dmsimard | jeblair: the plan is to prevent a commit to zuul-jobs from introducing a regression in base, so the jobs *have* to be in zuul-jobs ? | 18:19 |
| dmsimard | jeblair: unless you mean define the jobs in openstack-zuul-jobs and then add it to the zuul-jobs pipeline | 18:19 |
| jeblair | dmsimard: yep the second thing is what i meant | 18:19 |
| dmsimard | ok, sure | 18:19 |
| jeblair | dmsimard: as a config-repo, project-config is permitted to add jobs to any project's pipeline. and since they're actually defined in openstack-zuul-jobs, they'll still work for pre-merge testing in the same way | 18:20 |
| jeblair | dmsimard: you should find an existing project stanza for zuul-jobs in project-config | 18:20 |
| dmsimard | jeblair: wait -- so the jobs need to go in openstack-zuul-jobs but added to project-config for the zuul-jobs pipeline? | 18:21 |
| jeblair | dmsimard: "job:" stanzas in openstack-zuul-jobs; "project:" stanza in project-config | 18:21 |
| dmsimard | jeblair: ok so that means I'd need to squash the configure-mirrors patch so that they pass the jobs that would be merged in project-config (cause I'm not going to send 5 patches to project-config to progressively add each distro like I did for zuul-jobs) | 18:24 |
| dmsimard | jobs that would be added to the pipeline in project-config* | 18:24 |
| dmsimard | in zuul-jobs, I made sure the jobs would pass for each patch (each new distro, adding a distro job on each patch) | 18:26 |
| jeblair | dmsimard: if you want, we can land the whole series and forklift it over post facto | 18:28 |
| jeblair | (zuul-jobs doesn't have many consumers yet; it won't be broken for folks for long) | 18:29 |
| dmsimard | I don't mind either way, I iterated that series as a single patch ( https://review.openstack.org/#/c/501281/ ) and then split it so it would be easier to review and test | 18:29 |
| dmsimard | forklifting would unblock me immediately because base-minimal doesn't run trusted tasks that are required | 18:30 |
| jeblair | dmsimard: not sure what you mean by that? | 18:32 |
| dmsimard | configure-mirrors is broken for every distro but ubuntu, the fixes for that are in this series of patches | 18:32 |
| dmsimard | I'm unable to leverage the devstack-legacy work because I'd need to run from base-minimal which doesn't include the cached-repos/git-workspace roles which I can't run "manually" due to the local executor commands | 18:33 |
| jeblair | dmsimard: ah gotcha | 18:34 |
| dmsimard | It's confusing, sorry :( | 18:34 |
| jeblair | dmsimard: want to write the forklift patches while i go through the stack? | 18:34 |
| dmsimard | aye, I had already started working on it | 18:34 |
| dmsimard | BTW mirror_info.sh is still on my todo | 18:35 |
| dmsimard | jeblair: ah I suspect there might be a merge conflict hidden in there about the apt update thing.. let me go ahead and fix that | 18:36 |
| jeblair | dmsimard: yeah i was just looking at that. i think you're right | 18:36 |
| dmsimard | I'll need to rebase the whole thing, blergh | 18:38 |
| *** olaph has quit IRC | 18:40 | |
| *** olaph has joined #zuul | 18:41 | |
| jeblair | okay i have a ridiculous problem: http://paste.openstack.org/show/620657/ | 18:43 |
| jeblair | when i use a shell task with a heredoc, everything is indented one space | 18:43 |
| jeblair | er, i should rephrase that as: everything in that shell block is indented one space so the heredoc doesn't work (since " EOF" != "EOF") | 18:44 |
| pabelanger | jeblair: did you make it to ansiblefest? | 18:44 |
| jeblair | pabelanger: yes, im hiding in a room behind the tshirts | 18:44 |
| dmsimard | jlk: btw, on ignore_errors vs failed_when | 18:45 |
| pabelanger | jeblair: any room for another body? | 18:45 |
| jeblair | pabelanger: yep | 18:45 |
| pabelanger | k, making my way over | 18:45 |
| dmsimard | jlk: the behavior is different afaik, ignore_errors will count the failures but make them non-fatal but failed_when: false makes the task 'ok' while in reality they're not | 18:45 |
| dmsimard | do I have that right ? | 18:45 |
| jlk | jeblair: do it as "script:" instead of shell? | 18:45 |
| jlk | dmsimard: let me confirm some testing. Mostly what it does is visually different. ignore_errors still makes the output red, even though it doesn't stop execution. failed_when: false does not. | 18:46 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for debian in configure-mirrors https://review.openstack.org/501537 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for opensuse in configure-mirrors https://review.openstack.org/501539 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for fedora in configure-mirrors https://review.openstack.org/501538 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Expand PATH for SUSE systems to include /usr/sbin https://review.openstack.org/501737 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration tests for validate-host https://review.openstack.org/501543 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add integration test jobs for the configure-mirror role https://review.openstack.org/501534 | 18:47 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 18:47 |
| dmsimard | jeblair: ^ rebased to resolve the apt cache thing | 18:47 |
| jeblair | jlk: yeah, i'd have to make it an external file then; i was hoping to just do something really quick and simple | 18:48 |
| dmsimard | jeblair: if it can make you feel better I planned on take on mirror_info and hit that kind of problem I guess :) | 18:49 |
| jeblair | mostly, i don't understand why this doesn't work with shell | 18:49 |
| dmsimard | I don't remember seeing this problem | 18:50 |
| jeblair | (and if i'm going to have to eat crow after telling people "this is how you make ansible do a shell thing like in jjb") | 18:50 |
| dmsimard | then again, when copying a file, I'll typically use a copy or a template task | 18:50 |
| dmsimard | s/copying/writing/ | 18:50 |
| dmsimard | copy allows you to source the content directly without needing to use a template file if it's not appropriate | 18:51 |
| jlk | dmsimard: http://paste.openstack.org/show/620660/ | 18:51 |
| jlk | dmsimard: at least with the command module, piping through |failed works for both cases. | 18:51 |
| jeblair | yeah, there are better ways of doing this if *this* were the task. instead, this is a radically simplified version to illustrate the problem | 18:51 |
| dmsimard | jlk: but it shows up as changed instead of failed | 18:52 |
| dmsimard | jlk: which bothers me a bit when the failures are somewhat important but not fatal | 18:52 |
| jlk | yeah. I'm likewise bothered by red output that's not fatal, it makes it difficult to find the real error | 18:52 |
| jlk | I won't fight hard though if the Zuul project prefers ignore_errors. I'd just want it to be consistent | 18:53 |
| dmsimard | jlk: then again, I'm not convinced by https://review.openstack.org/#/c/501543/5/roles/validate-host/tasks/main.yaml anyway | 18:53 |
| dmsimard | jlk: I need to think about how these roles might behave when outside of a zuul context (i.e, third party ci) | 18:54 |
| dmsimard | the block is basically a hack in order to try and run the role without failing because some tasks need to run on the executor (and fail because they can't) | 18:55 |
| dmsimard | it did allow me to troubleshoot that SUSE validate-host issue, though, so it wasn't in vain | 18:56 |
| clarkb | dmsimard: https://review.openstack.org/#/c/501536/4/roles/configure-mirrors/handlers/main.yaml does that require apt and yum be installed on both ubuntu and centos? | 19:02 |
| clarkb | or handlers only run when they are notified? | 19:02 |
| dmsimard | clarkb: handlers only run when they are notified. | 19:02 |
| dmsimard | clarkb: those reviews are self-tested FWIW, look at the v3 jobs | 19:03 |
| clarkb | ya I wasn't sure if it was passing beacuse we somehow had yum on ubuntu or not | 19:03 |
| jeblair | here's my issue: https://github.com/ansible/ansible/issues/12856 | 19:03 |
| pabelanger | stepping away for some lunch, but puppet changes are up to allow us to bring nb01 / nb02 back online as feature/zuulv3 is we want to | 19:04 |
| jeblair | so, wow, _raw_params fixes that. fascinating | 19:05 |
| dmsimard | clarkb: you would see if these tasks had ran | 19:05 |
| dmsimard | jeblair: we just learned that together | 19:05 |
| dmsimard | larsks (bug submitter) is quite awesome :) | 19:06 |
| dmsimard | he helped me jumpstart ara in the beginning of the project | 19:06 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Enable maintainConnectionCache https://review.openstack.org/501144 | 19:07 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Close command socket after sending stop https://review.openstack.org/500755 | 19:11 |
| clarkb | dmsimard: I left a comment on https://review.openstack.org/#/c/501534/3 and looks like jeblair's -1 still applies from ps2 | 19:12 |
| Shrews | jeblair: looks like the base-test stuff worked: http://logs.openstack.org/89/501789/1/check/tox-py35-on-zuul/ef1c4e2/job-output.txt.gz#_2017-09-07_17_17_08_239257 | 19:12 |
| Shrews | jeblair: submitted https://review.openstack.org/501843 for the base job | 19:12 |
| Shrews | anyone care to +3 https://review.openstack.org/496251 ? | 19:13 |
| dmsimard | clarkb: ok, I'll fix your comment in a follow up patch so I don't have to rebase the stack again, I think jeblair said he'd look at the tree soon. I'm working on patches to address his comment and move stuff out to openstack-zuul-jobs and project-config. | 19:14 |
| jlk | mordred: re your project-config change to emit messages about what change triggered the job, yeah you'd think that'd be a variable generated by the driver and passed through. | 19:15 |
| *** jkilpatr has quit IRC | 19:17 | |
| jlk | mordred: and in a way that has the full URL so that multiple connections in the same install would work (multiple gerrits for example) | 19:19 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Set base environment as python3 https://review.openstack.org/496251 | 19:19 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Remove project pipeline definition from zuul-jobs https://review.openstack.org/501849 | 19:23 |
| jeblair | Shrews: perfect thanks! usually that catches something. ;) | 19:28 |
| Shrews | \o/ one less thing on the pre-ptg todo list | 19:29 |
| dmsimard | Shrews: btw was it you who worked on the improved integration tests for zuul stream | 19:29 |
| dmsimard | I forget | 19:29 |
| Shrews | dmsimard: not really. i just added some output validation | 19:30 |
| jeblair | dmsimard: what happened to my comment about apt-get update ? | 19:30 |
| dmsimard | Shrews: ah, k -- caught a bug earlier in zuul stream figured you might know where to look | 19:30 |
| jeblair | dmsimard: you dropped it in your update to 501754 | 19:31 |
| dmsimard | jeblair: I rebased on top of the revert and it's moved here https://review.openstack.org/#/c/501535/4/roles/configure-mirrors/handlers/main.yaml | 19:31 |
| dmsimard | it's used as a handler instead | 19:31 |
| jeblair | dmsimard: yeah, i'm looking for my *commentZ* | 19:31 |
| dmsimard | jeblair: oh, hm, let me check my emails | 19:32 |
| jeblair | dmsimard: it was in PS1 of my revert change, but you pushed ps2 on top of it and dropped the comment | 19:32 |
| jeblair | i really don't want to go through another revert cycle on that | 19:32 |
| dmsimard | jeblair: doh, sorry, let me add that back | 19:33 |
| jeblair | thx | 19:33 |
| jeblair | dmsimard: that and one other comment on 501535 | 19:34 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 19:36 |
| dmsimard | jeblair: there's the commment fix ^ | 19:36 |
| dmsimard | checking 501535 | 19:36 |
| dmsimard | ah doh, that's the same patch :D | 19:36 |
| mordred | jlk: well ... it is! https://review.openstack.org/#/c/501492/ | 19:37 |
| *** dkranz has quit IRC | 19:37 | |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 19:37 |
| dmsimard | jeblair: debug message fixed ^ | 19:37 |
| dmsimard | brb | 19:37 |
| *** jkilpatr has joined #zuul | 19:38 | |
| jeblair | tobiash, mordred: 501492 needs docs. i had to -2 to stop merge. happy to flip when change has docs. | 19:40 |
| jeblair | pabelanger: ^ | 19:40 |
| *** olaph1 has joined #zuul | 19:40 | |
| mordred | jeblair: oh - piddle - sorry ... one sec | 19:41 |
| jeblair | dmsimard: your stack has +3s all the way but you'll need to rebase | 19:41 |
| *** olaph has quit IRC | 19:42 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Be more permissive in configuring nodes https://review.openstack.org/501853 | 19:47 |
| mordred | jeblair, dmsimard: ^^ inspired by yesterday's conversation about node specifications | 19:48 |
| jeblair | mordred, dmsimard: i was thinking about that further, and i'm starting to lean against it | 19:48 |
| mordred | jeblair: yeah? whyso? | 19:48 |
| jeblair | mordred: mostly because if we go with dmsimard's idea of making a bunch of convenience nodesets, then we get syntax errors if someone tries to use one that doesn't exist. but with that change, we don't. | 19:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul feature/zuulv3: Add change_url to zuul dict passed into inventory https://review.openstack.org/501492 | 19:54 |
| jeblair | mordred: also, does your change eliminate the ability to specify a nodeset? | 19:54 |
| jeblair | mordred: oh, no i see it now. nm. | 19:54 |
| mordred | jeblair: not at all - it only treats a string in a place where a dict used to be as the name of a node with both label and name the same | 19:54 |
| mordred | jeblair: the nodeset idea doesn't bother me / sounds good - it's still slightly repetitive typing when defining multi-node nodesets, but there are other improvements I think we shoudl consder there as well, so happyto shelve this one | 19:57 |
| jeblair | mordred: okay, i'd still like us to mull over whether we want to add the ambiguity here -- i think this may be the first place where a string could mean two differet things. | 19:57 |
| jeblair | mordred: yeah, i'm in favor of defining convenince nodesets in project-config for now and discussing this later | 19:58 |
| mordred | jeblair: ++ | 19:58 |
| mordred | jeblair: also,it' snot *quite* two different meanings for string, - since one is 'nodesetname' and one is ['nodename'] - but I take your meaning, the leading - or not could certainly lead to confusion | 19:59 |
| mordred | jeblair: I've marked the nodes one WIP, https://review.openstack.org/#/c/501492/ now had docs | 20:00 |
| jeblair | mordred: gotcha. that may help a bit. it's at least not ambiguous for the software, may still be for the wetware :) | 20:01 |
| mordred | jeblair: silly wetware :) | 20:02 |
| jeblair | mordred: okay, i carried through the existing +2s for a +3 on change_url | 20:02 |
| mordred | woot. thanks | 20:05 |
| dmsimard | jeblair: not sure if I need rebase, I think it's https://review.openstack.org/#/c/501534/ that is missing, it's the first of the tree | 20:13 |
| dmsimard | I'll address clarkb and pabelanger's comments in a follow up patch | 20:14 |
| jeblair | dmsimard: 501536 is based on outdated ps of 501535 | 20:14 |
| dmsimard | jeblair: does that matter with "merge if necessary" ? | 20:15 |
| dmsimard | zuul didn't fail the merge check | 20:15 |
| jeblair | i believe so. however, if you would like to test that, i'll indulge. | 20:16 |
| clarkb | ya gerrit won't let you merge if you are based on the old thing | 20:17 |
| dmsimard | clarkb: hmm, I overestimated the merge if necessary :) | 20:19 |
| jeblair | clearly needs a 'merge if convenient' middle-ground option :) | 20:19 |
| dmsimard | lol | 20:20 |
| dmsimard | ok let me rebase the darn thing | 20:20 |
| dmsimard | might as well address clark and pabelanger's comments | 20:20 |
| dmsimard | spam inc :( | 20:21 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 20:22 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for debian in configure-mirrors https://review.openstack.org/501537 | 20:22 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for fedora in configure-mirrors https://review.openstack.org/501538 | 20:22 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Add support for opensuse in configure-mirrors https://review.openstack.org/501539 | 20:22 |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Remove project pipeline definition from zuul-jobs https://review.openstack.org/501849 | 20:22 |
| dmsimard | ok, I didn't need to touch the base, just from 501536 and up, hopefully that should do it | 20:23 |
| * dmsimard sends new patch to fix 501536's comments | 20:25 | |
| jeblair | whole stack approved | 20:27 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Stop output start and end times for each task https://review.openstack.org/500163 | 20:28 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Handle debug messages cleanly https://review.openstack.org/501490 | 20:31 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add a role to emit an informative header for logs https://review.openstack.org/501495 | 20:34 |
| mordred | jlk: ^^ thank you - the prevoius version of that did not copy-pasta properly | 20:35 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add integration test jobs for the configure-mirror role https://review.openstack.org/501534 | 20:35 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Streamline configure-mirrors and Ubuntu repository setup https://review.openstack.org/501535 | 20:36 |
| ianw | dmsimard: i started on the /etc/ci/ script in 501578 | 20:43 |
| ianw | seemed too infra specific for zuul-jobs | 20:43 |
| mordred | ianw: in its current form it is, but eventual goal is for configure-mirrors to properly set up any mirrors a zuul operator indicates that they have, and no-opping if they don't configure a mirror-host | 20:49 |
| dmsimard | ianw: arguably configure-mirrors is very openstack specific right now | 20:49 |
| mordred | ianw: I imagine it'll take us a while to get it fully solid and legit re-usable of course | 20:49 |
| jeblair | is it? | 20:50 |
| mordred | I mena, I know that when settingup 3rd party cI things, it's also quite important to set up local mirrors of things, else jobs fail frequently | 20:51 |
| jeblair | at any rate, /etc/ci definitely is openstack specific, so not putting it in zuul-jobs seems prudent | 20:51 |
| jeblair | the goal for configure-mirrors is to be reusable, so in zuul-jobs seems appropriate (regardless of whether we have perfectly achived that yet) | 20:51 |
| mordred | agree | 20:52 |
| dmsimard | we meant to deprecate/get rid of /etc/ci anyway right ? | 20:52 |
| ianw | at least the /etc/ci/mirror_info.sh script is full of our http proxies and stuff | 20:52 |
| dmsimard | like we're only putting it in because people are relying on it right now ? | 20:52 |
| pabelanger | eventually, but I think we need to support /etc/ci/mirror_info.sh for a bit | 20:53 |
| jeblair | dmsimard: ideally; we haven't really thought about that more than agreeing that we should think about it. :) | 20:53 |
| ianw | yeah, but i still think it's useful for things that aren't 100% ansible driven and want access to the variables | 20:53 |
| mordred | right. we don't konw what it wants to be in the future - we just know it'snecessary today | 20:53 |
| pabelanger | I know about 5 projects that depend on it | 20:53 |
| dmsimard | did we have success in fixing that zuul stream ipv6 thing ? | 20:55 |
| ianw | anyway now the integration tests are there i'll try working it in | 20:56 |
| pabelanger | dmsimard: yes | 20:56 |
| jeblair | dmsimard: mordred's change to fix it merged. | 20:56 |
| jeblair | mordred: http://logs.openstack.org/36/501536/5/gate/tox-py35-on-zuul/1154233/testr_results.html.gz | 20:56 |
| jeblair | dmsimard, mordred: however, it seems there may still be a problem ^ | 20:56 |
| dmsimard | ianw: fyi I'm currently waiting for the configure-mirror/validate-host tree to land (it's merged) and then I'd like to test centos/fedora devstack things | 20:57 |
| jeblair | ConnectionRefusedError: [Errno 111] Connect call failed ('2604:e100:1:0:f816:3eff:fe3b:7930', 7902) | 20:57 |
| dmsimard | ianw: the plan is to leverage jeblair's work here https://github.com/openstack-infra/devstack-gate/blob/master/.zuul.yaml#L27 and work from there but is there a lot of custom stuff involved for other distros or is it mostly copypasta ? | 20:57 |
| dmsimard | jeblair: let me see when that merged because I just recheck'd two fairly recent failures | 20:58 |
| ianw | dmsimard: hmm, i want to say almost nothing would need to be done, but that may not be true :) d-g has an is_fedora function used in just a couple of places | 20:59 |
| ianw | it wouldn't be more than those things | 21:00 |
| dmsimard | ianw: we're mostly running d-g from v3 | 21:00 |
| dmsimard | I might request your help :) | 21:01 |
| ianw | sure. this is going to break http://people.redhat.com/~iwienand/devstack-status/ again :) | 21:01 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Split the log_path creation into its own role https://review.openstack.org/501494 | 21:04 |
| jlk | jeblair: has zuul been restarted since yesterday afternoon? Are we seeing improvement in github responses? | 21:04 |
| mordred | jeblair: do we still want/need merge-check ? | 21:05 |
| mordred | jeblair: and yah - my change to fix it does not seem ot have fixed it | 21:05 |
| jeblair | mordred: let's keep it for now; retiring it isn't really tied to v3, more tied to gerrit features/settings. | 21:06 |
| mordred | jeblair: gotcha. | 21:06 |
| mordred | jeblair: and noop job still exists? | 21:06 |
| jeblair | mordred: yep | 21:06 |
| mordred | jeblair: cool | 21:07 |
| jeblair | Sep 05 21:30:29 zuulv3 systemd[1]: Started LSB: Zuul scheduler. | 21:07 |
| jeblair | jlk: that looks like a negative; let's do that when things settle down a bit (running a lot of jobs now) | 21:07 |
| jlk | kk | 21:07 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Add change_url to zuul dict passed into inventory https://review.openstack.org/501492 | 21:23 |
| mordred | jeblair, jlk, clarkb, pabelanger, Shrews, ianw: I'd like to restart the zuul-scheduler to pick ip thatchange ^^ - that ok with folks? | 21:27 |
| jeblair | mordred: wfm; please recheck 500202 when you do | 21:28 |
| mordred | jeblair: will do | 21:31 |
| pabelanger | ++ | 22:34 |
| mordred | jeblair, pabelanger, jlk: zuul scheduler as well as ze01 and ze02 have been restarted | 22:36 |
| pabelanger | great | 22:39 |
| jlk | woo | 22:44 |
| *** hasharAway has quit IRC | 22:54 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add support for centos in configure-mirrors https://review.openstack.org/501536 | 23:07 |
| SpamapS | anybody know how to get zookeeper on centos? | 23:20 |
| SpamapS | pabelanger: ? | 23:21 |
| mordred | SpamapS: I think dmsimard and tristanC have a something somewhere | 23:24 |
| SpamapS | I'm sure they do.. but they're not responding to me and I"m racing them to update BonnyCI before they just throw the whole software factory at me. ;-) | 23:24 |
| dmsimard | nooooo that's a lie | 23:24 |
| clarkb | SpamapS: I install from zk upstream tarball works fine | 23:24 |
| dmsimard | SpamapS: sorry I went to eat while this cloud thing was down | 23:24 |
| SpamapS | clarkb: but then I have to like, write a systemd unit file | 23:24 |
| mordred | SpamapS: oh - well, as far as that race goes, tristanC has already updated software factory to include v3 support ;) | 23:24 |
| clarkb | SpamapS: it comes with a start stop etc script you can just copy over | 23:25 |
| SpamapS | dmsimard: clouds are always down .. a bit ;) | 23:25 |
| dmsimard | SpamapS: this cloud happens to host software factory | 23:25 |
| clarkb | its sysvy but systemd knows how to talk to that | 23:25 |
| dmsimard | SpamapS: so there ya go | 23:25 |
| mordred | dmsimard: \o/ | 23:25 |
| dmsimard | SpamapS: https://github.com/softwarefactory-project | 23:25 |
| dmsimard | you want just zookeeper or the whole thing ? | 23:25 |
| dmsimard | mnaser tried out software factory and was pretty happy with it :D | 23:25 |
| SpamapS | thats um | 23:26 |
| SpamapS | a lot of.. | 23:26 |
| SpamapS | repos | 23:26 |
| clarkb | SpamapS: seriously just use upstream tarball... | 23:26 |
| dmsimard | SpamapS: https://github.com/softwarefactory-project/zookeeper-lite-distgit | 23:26 |
| dmsimard | SpamapS: I could point you to the RPM repository but hah, it's kinda down right now | 23:26 |
| SpamapS | dmsimard: well if there's a like.. "here's how to try software factory" ... I'll take that too :) | 23:27 |
| SpamapS | clarkb: I belieb you.. but.. bah ;) | 23:27 |
| dmsimard | SpamapS: the docs are self hosted on SF, it's really.. yeah | 23:28 |
| dmsimard | embarassing | 23:28 |
| dmsimard | SpamapS: hang on | 23:28 |
| dmsimard | SpamapS: https://github.com/softwarefactory-project/sf-docs | 23:29 |
| clarkb | the nice thing for me is it runs entirely in a corner as a non privileged user | 23:29 |
| dmsimard | SpamapS: SF provides either heat templates to deploy the whole thing for you automatically (really) or you can nova boot your stuff manually like it's 2015 | 23:30 |
| clarkb | I think the script might need a slight modification to work properly if you want to symlink it into /etc/init.d/ though | 23:30 |
| clarkb | (so that it properly looks up its real path) | 23:30 |
| clarkb | dmsimard: if only heat existed in any cluods | 23:31 |
| mordred | clarkb: it's actually in most of them now I believe | 23:31 |
| clarkb | mordred: I think of the ones we use its in 2? | 23:31 |
| clarkb | vexxhost and rax | 23:31 |
| mordred | hrm. I though it was more than that | 23:32 |
| SpamapS | dmsimard: As a former heat core.. I don't really trust Heat. Also we're about to remove Heat from our clouds so........ | 23:32 |
| mordred | oh well | 23:32 |
| clarkb | oh citycloud might have it too | 23:32 |
| SpamapS | since nobody uses it here | 23:32 |
| SpamapS | (to be fair, it never breaks when nobody uses it) | 23:32 |
| dmsimard | SpamapS: oh don't get me wrong, I hate heat | 23:33 |
| dmsimard | I also deployed it at $oldjob and no one used it | 23:33 |
| dmsimard | just saying it's there and it's kinda like that "docker run" kind of trying it out thingy | 23:33 |
| SpamapS | ACK | 23:34 |
| clarkb | mordred: ya citycloud seems to have it | 23:34 |
| SpamapS | well I'm nearly done with BonnyCI-on-centos so I may as well finish that before I play with SF | 23:34 |
| clarkb | also til heat uses two catalog entries | 23:34 |
| mordred | clarkb: orly? | 23:34 |
| clarkb | mordred: yup and in some clouds they have different names but in others they don't | 23:35 |
| dmsimard | SpamapS: FWIW SF is a bit more than just nodepool/zuul/gerrit, it also bundles storyboard, ELK, grafana/influx | 23:35 |
| mordred | clarkb: what entries did you see? | 23:35 |
| dmsimard | they're on-demand enabled | 23:35 |
| clarkb | mordred: https://wiki.openstack.org/wiki/API_Working_Group/Current_Design/Service_Catalog sometimes its heat-cfn for cloudformations and sometimes its just heat for cloud formations | 23:35 |
| SpamapS | yeah I don't even need gerrit | 23:35 |
| clarkb | mordred: and then 'heat' for orchestration | 23:35 |
| dmsimard | SpamapS: oh, SF kinda assumes gerrit right now though | 23:36 |
| mordred | clarkb: that's service-name - did you see 'orchestration' as service-type for orchestration? | 23:36 |
| SpamapS | well I could fix that :) | 23:36 |
| SpamapS | or | 23:36 |
| dmsimard | SpamapS: it's going to remove that assumption actually, but not right now | 23:36 |
| SpamapS | keep the gerrit | 23:36 |
| SpamapS | and get us on to it ;) | 23:36 |
| dmsimard | SpamapS: where's your new gig? :D | 23:36 |
| SpamapS | dmsimard: GoDaddy | 23:36 |
| clarkb | mordred: ya the two types are orchestration and cloudformations | 23:37 |
| dmsimard | oh, with harlowja and klindgren ? | 23:37 |
| mordred | awesome | 23:37 |
| harlowja | sup | 23:37 |
| dmsimard | harlowja: ohai | 23:37 |
| SpamapS | if ZK is already in Fedora (since F22??) why isn't it in EPEL yet? | 23:37 |
| harlowja | and now SpamapS | 23:37 |
| mordred | clarkb: we do not have an entry in service-types-authority for cloudformations - but that's ok, because we also don't care about it :) | 23:37 |
| dmsimard | SpamapS: grats :) | 23:37 |
| mordred | SpamapS: there are bugs open for that | 23:37 |
| clarkb | mordred: its apparently common in those example catalogs though | 23:38 |
| mordred | oh - sure | 23:38 |
| dmsimard | SpamapS: f22 is way too bleeding edge for rhel :) | 23:38 |
| harlowja | dmsimard whats up, ha | 23:38 |
| harlowja | oh, nm, reading backlog, lol | 23:38 |
| mordred | clarkb: I'm more just making sure that there isn't variation I didn't know about on the orchstation service-type front | 23:38 |
| dmsimard | harlowja: you end up going to the ptg after all or no ? | 23:38 |
| clarkb | mordred: there is variation in the 'name' field | 23:38 |
| ianw | SpamapS: http://lists.openstack.org/pipermail/openstack-infra/2016-December/004963.html ... guess what, it's almost december again! | 23:39 |
| clarkb | mordred: not sure if that matters | 23:39 |
| mordred | clarkb: that doesn't matter - name should be ignored | 23:39 |
| ianw | i haven't updated that, but zookeeper is like the little top of a big iceberg | 23:39 |
| mordred | clarkb: it is only looked at during service discovery as an absolute last resort (such as when our rax account had 2 different compute entries) | 23:39 |
| mordred | clarkb: but situations such as that are considered bong, and the recommendation is for folks to largely ignore service_name | 23:40 |
| clarkb | mordred: got it | 23:40 |
| clarkb | seriously though zookeeper tarball has worked great on tumbleweed | 23:40 |
| clarkb | I run the init script and it starts, I run it again and it stops | 23:40 |
| SpamapS | ianw: apparently this mountain moves slowly ;) | 23:42 |
| ianw | i think it's because it's java | 23:42 |
| dmsimard | oh man ianw beat me to it | 23:42 |
| dmsimard | no one wants to package zookeeper in RDO | 23:43 |
| dmsimard | no one | 23:43 |
| clarkb | also you could totally run an ubuntu vm or container on centos >_> | 23:43 |
| dmsimard | it's like elasticsearch, we stay away from them like the plague | 23:43 |
| clarkb | dmsimard: doesn't elasticsearch publish packages though? so at least there is an easy way to consume that ste of tools on $distro | 23:43 |
| dmsimard | otoh, if we were to adopt etcd3 instead... :) | 23:43 |
| harlowja | dmsimard nah, don't think i can, due to a bachloer party i gotta go to | 23:43 |
| dmsimard | harlowja: ok :( | 23:44 |
| SpamapS | yeah but elasticsearch actually makes their own centos packages | 23:44 |
| SpamapS | so it's NBD | 23:44 |
| harlowja | u can come to bay area dmsimard | 23:44 |
| ianw | it's just a huge time sink to context switch into the java world | 23:44 |
| clarkb | oh ya zk tarballs says symlink is followed automatically | 23:44 |
| clarkb | so that should totally just work, grab tarball, unpack, set symlink in /etc/init.d/ then systemctl reload-units | 23:44 |
| dmsimard | Is zuul eventually going to move to etcd3 ? I know zuul isn't impacted by TC decisions, just curious | 23:45 |
| clarkb | dmsimard: considering how even more painful it is to use etcd3 than zk right now in devstack I hope not | 23:45 |
| SpamapS | grab tarball, embed gpg key.. verify ... :-P | 23:45 |
| ianw | SpamapS / dmsimard : so yeah, the conclusion i basically came to was that centos 8 will appear before anyone does the work to get it all done properly on centos 7 | 23:46 |
| SpamapS | ianw: ACK | 23:46 |
| dmsimard | ianw: so just like py3 :P | 23:46 |
| SpamapS | Almost done fetching the tarball in ansible :-P | 23:46 |
| clarkb | SpamapS: path to script is bin/zkServer.sh in the tarball | 23:46 |
| clarkb | should work as a sys v init script | 23:47 |
| SpamapS | we'll see | 23:47 |
| SpamapS | I'm mostly trying to figure out how to verify the sig easily | 23:47 |
| SpamapS | can't find the GPG keys in any verifiable place at the moment | 23:48 |
| clarkb | SpamapS: according to their docs they want you to grab them from http://apache.org/dist/zookeeper/KEYS | 23:49 |
| clarkb | http://www.apache.org/dyn/closer.cgi#verify <- the docs | 23:49 |
| * clarkb notes that is http not https | 23:49 | |
| clarkb | oh https works too they just don't link/redirect to it | 23:49 |
| SpamapS | yeah as usual :-P | 23:50 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!