*** harrymichal has quit IRC | 00:08 | |
*** harrymichal has joined #zuul | 00:08 | |
*** hamalq has quit IRC | 00:09 | |
*** rfolco has quit IRC | 00:26 | |
*** wuchunyang has joined #zuul | 00:33 | |
*** wuchunyang has quit IRC | 00:38 | |
corvus | tristanC: haha :) that's not me -- i open so many emacs sessions i'm constantly stealing locks from myself :) | 00:42 |
---|---|---|
*** gouthamr_ has quit IRC | 00:59 | |
*** gouthamr_ has joined #zuul | 01:05 | |
*** Goneri has quit IRC | 01:18 | |
*** gouthamr_ has quit IRC | 01:25 | |
*** gouthamr_ has joined #zuul | 01:37 | |
*** swest has quit IRC | 01:48 | |
*** rlandy|bbl is now known as rlandy | 01:49 | |
*** swest has joined #zuul | 02:02 | |
*** rlandy has quit IRC | 02:04 | |
mnaser | hmm | 02:49 |
*** gouthamr_ has quit IRC | 02:52 | |
*** bhavikdbavishi has joined #zuul | 02:54 | |
*** gouthamr_ has joined #zuul | 02:56 | |
*** bhavikdbavishi1 has joined #zuul | 03:01 | |
*** bhavikdbavishi has quit IRC | 03:03 | |
*** bhavikdbavishi1 is now known as bhavikdbavishi | 03:03 | |
*** wuchunyang has joined #zuul | 03:18 | |
*** gouthamr_ has quit IRC | 03:26 | |
*** gouthamr_ has joined #zuul | 03:32 | |
*** saneax has joined #zuul | 03:35 | |
*** saneax has quit IRC | 03:40 | |
*** saneax has joined #zuul | 03:41 | |
*** bhavikdbavishi has quit IRC | 03:52 | |
*** bhavikdbavishi has joined #zuul | 03:56 | |
*** bhavikdbavishi1 has joined #zuul | 03:58 | |
*** gouthamr_ has quit IRC | 04:00 | |
*** saneax has quit IRC | 04:00 | |
*** bhavikdbavishi has quit IRC | 04:00 | |
*** bhavikdbavishi1 is now known as bhavikdbavishi | 04:00 | |
*** saneax has joined #zuul | 04:01 | |
*** saneax has quit IRC | 04:03 | |
*** saneax has joined #zuul | 04:03 | |
*** wuchunyang has quit IRC | 04:04 | |
*** bhavikdbavishi has quit IRC | 04:05 | |
*** wuchunyang has joined #zuul | 04:10 | |
*** gouthamr_ has joined #zuul | 04:12 | |
*** wuchunyang has quit IRC | 04:16 | |
*** saneax has quit IRC | 04:17 | |
*** saneax has joined #zuul | 04:18 | |
*** saneax has quit IRC | 04:22 | |
*** saneax has joined #zuul | 04:22 | |
*** bhavikdbavishi has joined #zuul | 04:23 | |
*** saneax has quit IRC | 04:24 | |
*** saneax has joined #zuul | 04:25 | |
*** bhavikdbavishi has quit IRC | 04:26 | |
*** bhavikdbavishi has joined #zuul | 04:28 | |
*** gouthamr_ has quit IRC | 04:29 | |
*** gouthamr_ has joined #zuul | 04:30 | |
*** evrardjp has quit IRC | 04:33 | |
*** evrardjp has joined #zuul | 04:33 | |
*** gouthamr_ has quit IRC | 04:35 | |
*** gouthamr_ has joined #zuul | 04:40 | |
*** wuchunyang has joined #zuul | 04:44 | |
*** wuchunyang has quit IRC | 04:48 | |
*** saneax has quit IRC | 04:52 | |
*** saneax has joined #zuul | 04:53 | |
*** saneax has quit IRC | 04:55 | |
*** saneax has joined #zuul | 04:55 | |
*** ysandeep|away is now known as ysandeep|rover | 05:03 | |
*** saneax has quit IRC | 05:03 | |
*** saneax has joined #zuul | 05:03 | |
*** saneax has quit IRC | 05:17 | |
*** saneax has joined #zuul | 05:19 | |
*** saneax has quit IRC | 05:20 | |
*** wuchunyang has joined #zuul | 05:20 | |
*** marios has joined #zuul | 05:39 | |
*** reiterative has quit IRC | 05:42 | |
*** reiterative has joined #zuul | 05:42 | |
*** vishalmanchanda has joined #zuul | 06:05 | |
*** bhavikdbavishi1 has joined #zuul | 06:15 | |
*** saneax has joined #zuul | 06:15 | |
*** bhavikdbavishi has quit IRC | 06:17 | |
*** bhavikdbavishi1 is now known as bhavikdbavishi | 06:17 | |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: tutorial: Add "job dependencies" https://review.opendev.org/732071 | 06:29 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Rename quick-start to zuul-tutorial-quick-start https://review.opendev.org/737656 | 06:29 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST run zuul tutorials to test stream+callback (+ zuul-jobs change) https://review.opendev.org/735477 | 06:29 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: [DNM] TEST run zuul tutorials to test stream+callback and multiple nodes https://review.opendev.org/735483 | 06:29 |
*** sanjayu_ has joined #zuul | 06:49 | |
*** gouthamr_ has quit IRC | 06:50 | |
*** saneax has quit IRC | 06:52 | |
*** gouthamr_ has joined #zuul | 06:52 | |
*** iamweswilson has quit IRC | 06:54 | |
*** dcastellani has quit IRC | 06:54 | |
*** PrinzElvis has quit IRC | 06:55 | |
*** dcastellani has joined #zuul | 06:56 | |
*** bhavikdbavishi has quit IRC | 06:56 | |
*** iamweswilson has joined #zuul | 06:56 | |
*** PrinzElvis has joined #zuul | 06:56 | |
*** gouthamr_ has quit IRC | 07:04 | |
*** gouthamr_ has joined #zuul | 07:05 | |
*** gouthamr_ has quit IRC | 07:15 | |
*** jcapitao has joined #zuul | 07:20 | |
*** gouthamr_ has joined #zuul | 07:22 | |
avass | tobiash: I added the changes we discussed yesterday on top of your change: https://review.opendev.org/#/c/740110/1 | 07:23 |
*** bhavikdbavishi has joined #zuul | 07:23 | |
*** zbr7 is now known as zbr|ruck | 07:23 | |
*** sugaar has joined #zuul | 07:26 | |
tobiash | avass: cool :) | 07:31 |
tobiash | zuul-maint: I have two test stability fixes for unstable tests that I observed recently: https://review.opendev.org/#/q/status:open+project:zuul/zuul+branch:master+topic:test-fixes | 07:32 |
*** sanjayu__ has joined #zuul | 07:34 | |
*** sanjayu_ has quit IRC | 07:36 | |
*** sanjayu__ has quit IRC | 07:40 | |
*** saneax has joined #zuul | 07:43 | |
*** hashar has joined #zuul | 07:45 | |
*** gouthamr_ has quit IRC | 07:46 | |
*** sanjayu_ has joined #zuul | 07:47 | |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Update synchronize-repos https://review.opendev.org/740110 | 07:47 |
*** saneax has quit IRC | 07:50 | |
*** sanjayu__ has joined #zuul | 07:51 | |
*** ysandeep|rover is now known as ysandeep|lunch | 07:52 | |
*** gouthamr_ has joined #zuul | 07:52 | |
*** ysandeep|lunch is now known as ysandeep | 07:53 | |
*** ysandeep is now known as ysandeep|lunch | 07:53 | |
*** sanjayu_ has quit IRC | 07:54 | |
*** sshnaidm has joined #zuul | 07:55 | |
*** sanjayu_ has joined #zuul | 07:55 | |
*** gouthamr_ has quit IRC | 07:57 | |
*** sanjayu__ has quit IRC | 07:59 | |
*** gmann has quit IRC | 08:03 | |
*** erbarr has quit IRC | 08:03 | |
*** erbarr has joined #zuul | 08:04 | |
*** gmann has joined #zuul | 08:06 | |
*** nils has joined #zuul | 08:09 | |
*** gouthamr_ has joined #zuul | 08:09 | |
*** iurygregory has quit IRC | 08:10 | |
*** sanjayu_ has quit IRC | 08:19 | |
*** iurygregory has joined #zuul | 08:24 | |
*** tumble has joined #zuul | 08:24 | |
*** tobiash has quit IRC | 08:26 | |
*** tobiash has joined #zuul | 08:28 | |
*** tobiash has quit IRC | 08:30 | |
*** ysandeep|lunch is now known as ysandeep|rover | 08:34 | |
*** tobiash has joined #zuul | 09:03 | |
*** tosky has joined #zuul | 09:10 | |
*** wuchunyang has quit IRC | 09:12 | |
*** saneax has joined #zuul | 09:13 | |
*** wuchunyang has joined #zuul | 09:17 | |
*** bhavikdbavishi has quit IRC | 09:19 | |
*** wuchunyang has quit IRC | 09:34 | |
*** bhavikdbavishi has joined #zuul | 09:42 | |
*** wuchunyang has joined #zuul | 09:48 | |
openstackgerrit | Tobias Henkel proposed zuul/zuul master: Make test_playbook more stable https://review.opendev.org/739835 | 09:54 |
openstackgerrit | Tobias Henkel proposed zuul/zuul master: Join command thread on exit https://review.opendev.org/739838 | 09:55 |
*** wuchunyang has quit IRC | 10:18 | |
*** wuchunyang has joined #zuul | 10:18 | |
*** wuchunyang has quit IRC | 10:20 | |
*** hashar has quit IRC | 10:27 | |
*** ysandeep|rover is now known as ysandeep|afk | 10:42 | |
*** ysandeep|afk is now known as ysandeep|rover | 10:58 | |
*** wuchunyang has joined #zuul | 11:01 | |
*** wuchunyang has quit IRC | 11:01 | |
*** jcapitao is now known as jcapitao_lunch | 11:06 | |
*** harrymichal has quit IRC | 11:14 | |
*** bschanzel has joined #zuul | 11:25 | |
avass | tobiash: I think I found some more problems with the windows ssh things. it won't work unless the privatekey is added to an ssh-agent or stored in ~/.ssh. | 11:52 |
zbr | is there a way to combine tox_environment defined in parent and add new values in child job? | 11:52 |
zbr | i do have a base job that defines a set of variables, but I need to add one more in the child, without rewriting all of them. | 11:53 |
avass | we're still storing the private key in /var/ssh, so ssh gets stuck waiting for a password. | 11:53 |
zbr | and these are under "vars:".... no idea how to make it happen. | 11:53 |
tobiash | avass: we add the pubkey of the zuul master key to the windows image | 11:54 |
avass | tobiash: yeah it works if I use the private key. the problem is that the localhost shell command doesn't use it | 11:54 |
tobiash | we probably want to enable windows for the build-sskkey rokle as well to solve this problem | 11:54 |
tobiash | they should use it (zuul loads it to the job local ssh agent) | 11:55 |
avass | we need something like 'ssh -i {{ nodepool_private_key_file }} ...' for the add-build-sshkey role | 11:55 |
avass | tobiash: ah yeah that role adds the build key, but the original master key isn't | 11:56 |
tobiash | avass: ssh should use the agent though (which at first contains the master key and later the build key) | 11:58 |
avass | the official images doesn't seem to do that though | 11:58 |
tobiash | avass: just saw we have a windows wrapper for build-sshkey as well, I'll upload it later today | 11:59 |
avass | tobiash: I think you already did :) | 12:00 |
mordred | zbr: they should just combine naturally - we merge dicts | 12:00 |
avass | tobiash: anyway that's ^ what I mean. | 12:00 |
avass | uh, I guess I was too quick | 12:00 |
zbr | wow, that was unexpected. ansible does not do that. | 12:00 |
zbr | a pleasant surprise in that case, i will find out soon. | 12:01 |
zbr | i still have some doubts that this merging is nested | 12:01 |
zbr | if is nested it would be impossbile to override some vars from parent job that are declared as dicts. | 12:01 |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Use ansible_ssh_private_key_file when setting up windows https://review.opendev.org/740254 | 12:02 |
avass | tobiash: that ^ | 12:02 |
*** jcapitao_lunch is now known as jcapitao | 12:02 | |
*** swest has quit IRC | 12:02 | |
tobiash | avass: we should rely on the ssh agent instead, zuul sets up one for each job on purpose | 12:03 |
avass | tobiash: then I don't understand why it doesn't work | 12:03 |
*** bhagyashris is now known as bhagyashris|brb | 12:04 | |
tobiash | avass: I think we should try to find that out, can you try with ssh -vvv? | 12:04 |
avass | tobiash: it get's timed out so I'm not getting any logs. But I think it's because it's waiting for a password | 12:04 |
avass | but I have another idea | 12:05 |
zbr | i could also use some help generating a custom event with react, mainly for https://review.opendev.org/#/c/739482/ | 12:05 |
*** zbr_ has quit IRC | 12:05 | |
tobiash | avass: ok, we should enforce batch mode at least | 12:05 |
avass | yeah exactly | 12:06 |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Enforce BatchMode when setting up ssh for windows https://review.opendev.org/740254 | 12:08 |
avass | actually I missed one | 12:09 |
*** swest has joined #zuul | 12:09 | |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: Enforce BatchMode when setting up ssh for windows https://review.opendev.org/740254 | 12:10 |
avass | tobiash: how about that? | 12:10 |
tobiash | ++ | 12:12 |
avass | mordred: wanna +3 that? pretty please? :) | 12:12 |
*** wuchunyang has joined #zuul | 12:19 | |
*** rlandy has joined #zuul | 12:23 | |
*** wuchunyang has quit IRC | 12:23 | |
*** rlandy is now known as rlandy|ruck2 | 12:24 | |
avass | tobiash: I keep hitting pre-run failures because winrm isn't running yet so I'm thinking that it might be good idea for nodepool to verify that somehow. sort of how it doesn't hand over a node until it can get an ssh-key | 12:28 |
tobiash | avass: nodepool can do that (wait for specific port open) | 12:29 |
tobiash | let me dig up the config | 12:29 |
avass | hmm, so maybe just adding some user-data that doesn't open the winrm port in the firewall until winrm is running | 12:30 |
tobiash | add connection-port: 5986 and connection-type: winrm to the diskimage config | 12:31 |
tobiash | then it waits until the winrm port is open | 12:31 |
avass | well, that's already configured. so maybe it's something else going on | 12:31 |
*** bhavikdbavishi has quit IRC | 12:31 | |
tobiash | what's not possible though is to wait for winrm AND ssh ports open | 12:31 |
mordred | avass: done | 12:32 |
avass | because sometimes I'm getting a 503 respose in ansible. I'm mostly guessing that it's because winrm hasn't started completely | 12:32 |
avass | mordred: thanks! | 12:32 |
avass | tobiash: during "Running ansible setup" that is | 12:33 |
tobiash | avass: hrm, I'd assumed that winrm is up and running once the port is open | 12:33 |
tobiash | avass: can it be that in your setup winrm is restarted during boot? | 12:34 |
avass | tobiash: could be | 12:34 |
tobiash | that might explain that issue | 12:34 |
tobiash | in that case one would need to add a winrm login check into nodepool and add the winrm master key to nodepool (optionally) | 12:35 |
tobiash | corvus: something like that might possibly solve your hostkey issue as well without ignoring them ^ | 12:36 |
*** ysandeep|rover is now known as ysandeep|afk | 12:37 | |
*** bhagyashris|brb is now known as bhagyashris | 12:37 | |
avass | tobiash: yeah I was hoping there would be a way to just verify that it's running somehow. but I can't find anything that doesn't require you to actually login and set up a session | 12:38 |
tobiash | that's the only way I can imagine atm (assuming that an unconfigured winrm is running at first) | 12:39 |
*** tumble has quit IRC | 12:43 | |
avass | tobiash: oh actually I think it might not be winrm | 12:45 |
openstackgerrit | Merged zuul/zuul-jobs master: Enforce BatchMode when setting up ssh for windows https://review.opendev.org/740254 | 12:47 |
avass | tobiash: looks like some no_proxy settings were slightly wrong and we haven't noticed that before since ssh isn't affected by it. so you might be right about the port | 12:50 |
openstackgerrit | Benjamin Schanzel proposed zuul/zuul master: [TEST DNM] Log streamer terminate time https://review.opendev.org/740271 | 12:57 |
*** ysandeep|afk is now known as ysandeep | 12:58 | |
*** pabelanger has left #zuul | 13:01 | |
tobiash | avass: you're fighting with proxy as well all the time? | 13:01 |
openstackgerrit | Merged zuul/zuul-jobs master: emit-job-header: add inventory hostname https://review.opendev.org/738963 | 13:02 |
avass | tobiash: yeah | 13:03 |
tobiash | we poor guys | 13:03 |
fungi | eventually people will learn to stop breaking the internet with proxies and address translation ;) | 13:04 |
tobiash | in the last 8 years that proxy cost me probably several months of working time so far | 13:04 |
mordred | tobiash: and provided no value in return | 13:06 |
avass | mordred, tobiash: oh tell me about it. | 13:06 |
tobiash | it took me literally weeks to find out how to run an openshift behind a proxy without breaking essential use cases | 13:08 |
avass | yeah I hit similar issues trying to set up kubernetes behind a proxy | 13:09 |
mordred | tobiash: what I love is when I work places that want me to install a corporate CA on my machine for internal ssl sites | 13:09 |
mordred | which is always a hard "nope" | 13:09 |
mordred | now - if I could install one in my browser and tie it to a domain or a set of domains that I trust it to provide certs for - sure | 13:09 |
tobiash | mordred: we can chose between internal and external certs | 13:10 |
mordred | and they wonder why people just install stuff into public clouds without telling them ... | 13:11 |
avass | mordred: I would never :) | 13:11 |
mordred | tobiash: but do they ask you to install a custom CA onto your laptop so that the internal certs show as valid? that's the thing I refuse to do | 13:12 |
mordred | avass: "of course not" | 13:12 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update "fetching info ..." and refresh animation https://review.opendev.org/738010 | 13:13 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update buildset result page (new layout and styling) https://review.opendev.org/738011 | 13:13 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update build result page layout https://review.opendev.org/739972 | 13:13 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Add icons and empty state to artifacts section on build result page https://review.opendev.org/739973 | 13:13 |
tobiash | mordred: custom CAs are preinstalled on all of our boxes, but services that are meant to be accessed by suppliers (like our CI) use real certs | 13:14 |
tobiash | so no cert issue with 'my' services ;) | 13:14 |
mordred | tobiash: :) | 13:17 |
*** Goneri has joined #zuul | 13:20 | |
avass | tobiash: looks like it still gets stuck even though it's in batchmode | 13:24 |
tobiash | avass: that is surprising as it should either succeed or fail (except if there is a firewall issue or so) | 13:26 |
avass | tobiash: yeah and I can ssh to the node from that executor using the private key | 13:26 |
tobiash | avass: is the syntax for the batch mode right? | 13:27 |
tobiash | there it's slightly different: https://www.thegeekstuff.com/2009/10/how-to-execute-ssh-and-scp-in-batch-mode-only-when-passwordless-login-is-enabled/ | 13:27 |
avass | tobiash: yep, it fails if I give it a bad user | 13:28 |
avass | instead of prompting for a password if it's not using batchmode | 13:29 |
tobiash | avass: maybe you could try it with -vvv and combine it with the timeout util so you might get debug logs despite it being blocked | 13:30 |
avass | tobiash: oh! looks like it was able to get the host-key, so it's not the first command that fails | 13:31 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update buildset result page (new layout and styling) https://review.opendev.org/738011 | 13:32 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update build result page layout https://review.opendev.org/739972 | 13:32 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Add icons and empty state to artifacts section on build result page https://review.opendev.org/739973 | 13:32 |
avass | tobiash: oh and actually I think I see another problem. we're running an administrator user so we would need to add the build-key to %programdata%\ssh\administrators_authorized_keys | 13:33 |
avass | or just not use an administrator user I guess | 13:33 |
tobiash | oh yeah, getting the authorized_keys right was the hardest thing getting to work | 13:34 |
tobiash | we use the administrator user btw | 13:34 |
tobiash | avass: we do this during the image build: http://paste.openstack.org/show/795706/ | 13:35 |
avass | tobiash: I couldn't get the Repair-AuthorizedKeyPermission to work so we do something else | 13:37 |
avass | tobiash: anyway, looks like scp <file> <user>@<host>:.ssh doesn't work | 13:37 |
avass | but that shouldn't lock it | 13:37 |
avass | tobiash: actually I was stupid and used ssh instead of scp | 13:38 |
avass | and yep that locks it | 13:38 |
avass | I guess it's because the .ssh directory doesn't exist | 13:39 |
avass | yup that's it, we need to make sure .ssh exists before adding the build-key, and maybe add a variable to configure which file to add it to | 13:40 |
openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: tests: use a more inclusive word for the default password https://review.opendev.org/740291 | 13:41 |
avass | tobiash: we do this instead: http://paste.openstack.org/show/795707/ :) | 13:43 |
*** wuchunyang has joined #zuul | 13:43 | |
tobiash | probably more lightweight | 13:43 |
*** felixedel has joined #zuul | 13:58 | |
felixedel | zuul-maint: I have a bunch of PF4 patches ready for review https://review.opendev.org/#/q/status:open+project:zuul/zuul+branch:master+topic:patternfly-4 | 13:59 |
felixedel | I hope you like the new design :) | 13:59 |
*** saneax has quit IRC | 14:00 | |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update buildset result page (new layout and styling) https://review.opendev.org/738011 | 14:01 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Update build result page layout https://review.opendev.org/739972 | 14:01 |
openstackgerrit | Felix Edel proposed zuul/zuul master: PF4: Add icons and empty state to artifacts section on build result page https://review.opendev.org/739973 | 14:01 |
mordred | corvus: oh - uhm. see email from han-wen about checks plugin and CI reboot on gerrit mailing list | 14:03 |
*** piotrowskim has joined #zuul | 14:04 | |
mordred | corvus: I've left a comment on the design doc | 14:07 |
*** felixedel has quit IRC | 14:16 | |
avass | felixedel: oh that's a pretty big difference | 14:17 |
*** harrymichal has joined #zuul | 14:20 | |
tristanC | fungi: for some reason, clicking on the site-preview for https://review.opendev.org/#/c/738010/7 builds result in `Index File Not Found` | 14:28 |
tristanC | fungi: oops, that was meant for felixedel | 14:28 |
tobiash | the npm folder is empty | 14:31 |
tobiash | on all three builds | 14:32 |
tobiash | oh no, not another regression of https://review.opendev.org/663108 :/ | 14:32 |
tristanC | oh noes, could it be the keep fix that caused that? | 14:33 |
*** dennis_effa has joined #zuul | 14:33 | |
tobiash | yes, it is :/ | 14:33 |
tobiash | let me see if I can fix forward with that | 14:33 |
tobiash | what's weird is that the job doesn't fail | 14:34 |
tobiash | tristanC, mordred: I think we need to create zuul/web/static and the link before building the page | 14:36 |
*** jamesmcarthur has joined #zuul | 14:39 | |
bschanzel | Hello there, may I have another review on https://review.opendev.org/#/c/736019/ ? | 14:40 |
openstackgerrit | Tobias Henkel proposed zuul/zuul master: Fix dashboard preview after keep file removal https://review.opendev.org/740313 | 14:41 |
tobiash | tristanC, mordred: I think that might do it ^ | 14:42 |
openstackgerrit | Tobias Henkel proposed zuul/zuul master: Fix dashboard preview after keep file removal https://review.opendev.org/740316 | 14:46 |
tobiash | and that might be an alternative ^ | 14:46 |
corvus | mordred: yeah, i feel like we're still struggling to tell them that we just need gerrit to store some really simple ci results for us | 14:47 |
mordred | yeah | 14:47 |
corvus | mordred: han-wen replied to your comment, and i don't think he understood. i mean you said "we can't guarantee we can even reach the ci system" and he said "gerrit doesn't have to, just the user's browser" | 14:48 |
mordred | yeah | 14:48 |
*** jamesmcarthur has quit IRC | 14:52 | |
*** bhagyashris is now known as bhagyashris|dinn | 14:52 | |
*** vishalmanchanda has quit IRC | 14:54 | |
mordred | corvus: I have responded to his response | 14:55 |
mordred | I hope I found the right balance | 14:55 |
corvus | mordred: i'm throwing in a comment about the verify-status plugin | 14:56 |
mordred | awesome | 14:56 |
*** mnasiadka has joined #zuul | 14:59 | |
corvus | mordred: it sounds like they want the 'parters' to implement an api of gerrit's specification | 14:59 |
mordred | yeah | 15:00 |
corvus | mordred: i don't get how asking them to implement a read api designed by gerrit is going to be more successful than implementing a write api designed by gerrit | 15:00 |
openstackgerrit | Merged zuul/zuul master: tests: use a more inclusive word for the default password https://review.opendev.org/740291 | 15:00 |
mordred | or - perhaps they assume all partners have an api that can be called and they expect to be able to write per-partner plugins in gerrit to call them? | 15:00 |
corvus | mordred: i'm trying to understand if the labels section describes something that's helpful for us? | 15:02 |
mordred | corvus: yeah "The Gerrit team will work with partners to make their existing frontend plugins use the new ingestion API. " - I think what that means is "the partners inside google all have UI plugins to display their ci results. the gerrit team at google will work with the partners at google to port their ui plugins to the new api" | 15:03 |
*** harrymichal has quit IRC | 15:03 | |
corvus | ah yep i missed the "existing frontent plugins" part | 15:03 |
corvus | mordred: maybe the labels section it will help with the 'comment spam' stream, but probably only the 'pluggable ci details' part is going to make an overview table | 15:04 |
mordred | corvus: also struggling to understand that. I feel like it's pretty vague | 15:04 |
mordred | yeah | 15:04 |
mordred | I'm starting to wonder if porting verify-status to notedb might not be a better path forward for us :( | 15:05 |
corvus | yeah -- did you see the comment i left there? | 15:05 |
mordred | I did | 15:05 |
corvus | maybe we should send something to the list and see if anyone else wants to collaborate on it | 15:06 |
corvus | (cause i'm getting a very googly vibe from this effort) | 15:06 |
mordred | like - if they're abandoning checks, that means subchecks is never gonna get written - and we're certainly not going to pick up that torch | 15:06 |
corvus | nope | 15:06 |
mordred | yeah | 15:06 |
tobiash | zuul-maint: this dashboard fixing variant works: https://review.opendev.org/740316 the other doesn't so I abandoned it | 15:06 |
mordred | this seems very much like google decided to give up and just solve google's problems | 15:06 |
corvus | mordred: ^ leaving 316 for you to +3 | 15:07 |
mordred | paladox: do you think anyone at wikimedia would be interested in collaborating on updating verify-status to use notedb? | 15:08 |
mordred | paladox: (given the apparent abandonment of checks) | 15:09 |
paladox | I don't think so as we don't use verify-status :(. Though we are currently using a plugin that displays jenkins checks as a table | 15:09 |
paladox | see https://gerrit.wikimedia.org/r/c/operations/puppet/+/556270 | 15:09 |
mordred | darn | 15:10 |
paladox | https://github.com/wikimedia/puppet/blob/production/modules/gerrit/files/homedir/review_site/static/gerrit-theme.html#L163 | 15:10 |
*** jamesmcarthur has joined #zuul | 15:11 | |
paladox | mordred you should be able to use it with the mysql db. | 15:11 |
mordred | paladox: is that using the checks api on the backend/ | 15:12 |
paladox | the plugin? nope it's using comments | 15:12 |
mordred | ? | 15:12 |
mordred | yeah - but once we update to 3.x I'd really prefer to not have to run a db anymore | 15:12 |
paladox | ah i see, hmm | 15:13 |
mordred | nod. so it's a plugin version of the thing we do with comment scraping | 15:13 |
paladox | yeh | 15:13 |
corvus | maybe that combined with the new labels stuff in the google doc would be sufficient for us? | 15:13 |
mordred | I see: https://github.com/wikimedia/puppet/blob/production/modules/gerrit/files/homedir/review_site/static/gerrit-theme.html#L258 | 15:13 |
*** jamesmcarthur has quit IRC | 15:14 | |
paladox | yup | 15:14 |
mordred | maybe? | 15:14 |
mordred | that code seems a bit easier to read that our current thing - so I bet we could fork it and modify it for our purposes fairly easy-ish and then it might be slightly more maintainable | 15:15 |
*** jamesmcarthur has joined #zuul | 15:15 | |
mordred | corvus: do we really need the new labels stuff? | 15:17 |
mordred | (I mean - I don't know what the new labels stuff actually does) | 15:17 |
corvus | mordred: i don't either, but i read that section as saying that there would be some facility for removing comment spam. like hideci | 15:18 |
mordred | ah - nod | 15:18 |
mordred | yeah | 15:18 |
mordred | that would be handy | 15:18 |
paladox | yeh, admins can erase comments i believe | 15:18 |
*** jamesmcarthur has quit IRC | 15:19 | |
corvus | paladox: sorry i didn't mean that kind of spam; i meant something to deal with large numbers of huge comments from ci systems | 15:19 |
paladox | oh | 15:19 |
corvus | something to remove them from the normal stream of comments so humans can actually read comments from other humans | 15:19 |
paladox | ah i see | 15:19 |
*** jamesmcarthur has joined #zuul | 15:25 | |
mordred | corvus: well - if we're not going to be waiting on subchecks now - it makes me want to consider just doing 2.13->3.2 over a weekend | 15:27 |
corvus | mordred: #opendev? | 15:28 |
mordred | corvus: yeah | 15:29 |
corvus | mordred: i mean, can we move this to #opendev? :) | 15:29 |
*** ysandeep is now known as ysandeep|away | 15:30 | |
fungi | paladox: for context, in opendev we have dozens of external ci systems operated by community volunteers, usually testing things like compatibility of drivers with proprietary software or hard to obtain hardware, so a change pushed to a project relevant to those might get 30+ different ci systems reporting results for a single change, drowning out the human discussion going on | 15:33 |
paladox | ah | 15:35 |
fungi | so we've long had a js overlay on the "new change screen" to toggle those comments on/off and present a concise results table parsed out of the comment text for them | 15:38 |
*** jamesmcarthur has quit IRC | 15:39 | |
*** jamesmcarthur has joined #zuul | 15:40 | |
*** jamesmcarthur has quit IRC | 15:41 | |
*** jamesmcarthur has joined #zuul | 15:42 | |
*** sgw1 has quit IRC | 15:44 | |
*** sgw1 has joined #zuul | 15:44 | |
*** jamesmcarthur has quit IRC | 15:48 | |
*** bhagyashris|dinn is now known as bhagyashris | 15:49 | |
*** marios has quit IRC | 16:00 | |
*** bhavikdbavishi has joined #zuul | 16:00 | |
*** jamesmcarthur has joined #zuul | 16:02 | |
openstackgerrit | Benoît Bayszczak proposed zuul/nodepool master: [provider][aws] now performs a multi-describe at once https://review.opendev.org/740328 | 16:03 |
openstackgerrit | Benoît Bayszczak proposed zuul/nodepool master: [DNM][provider][aws] now performs a multi-describe at once https://review.opendev.org/740328 | 16:07 |
openstackgerrit | Benoît Bayszczak proposed zuul/nodepool master: [WIP][provider][aws] now performs a multi-describe at once https://review.opendev.org/740328 | 16:07 |
openstackgerrit | Merged zuul/zuul master: Fix dashboard preview after keep file removal https://review.opendev.org/740316 | 16:13 |
*** hamalq has joined #zuul | 16:18 | |
*** hamalq_ has joined #zuul | 16:19 | |
*** jcapitao has quit IRC | 16:20 | |
*** bhavikdbavishi1 has joined #zuul | 16:20 | |
*** bhavikdbavishi has quit IRC | 16:22 | |
*** bhavikdbavishi1 is now known as bhavikdbavishi | 16:22 | |
*** hamalq has quit IRC | 16:22 | |
*** jamesmcarthur has quit IRC | 16:26 | |
*** jamesmcarthur has joined #zuul | 16:42 | |
*** andreaf has quit IRC | 16:48 | |
*** jbryce_ has joined #zuul | 16:48 | |
*** mnasiadka_ has joined #zuul | 16:48 | |
*** mwhahaha_ has joined #zuul | 16:48 | |
*** webknjaz has quit IRC | 16:55 | |
*** mwhahaha has quit IRC | 16:55 | |
*** mnasiadka has quit IRC | 16:55 | |
*** ChrisShort has quit IRC | 16:55 | |
*** jbryce has quit IRC | 16:55 | |
*** mordred has quit IRC | 16:55 | |
*** jbryce_ is now known as jbryce | 16:55 | |
*** mwhahaha_ is now known as mwhahaha | 16:55 | |
*** mnasiadka_ is now known as mnasiadka | 16:55 | |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: scheduler: Fix event process abide hasUnparsedBranchCache argument https://review.opendev.org/739042 | 16:55 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix branch name and project name for ref-updated create/delete https://review.opendev.org/738320 | 16:56 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: FakeGerritChange: Add Change-Id in commit message https://review.opendev.org/739197 | 16:56 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: WIP: Scheduler: Reconfiguration ref-updated create/delete https://review.opendev.org/739198 | 16:56 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: WIP: Scheduler: Reconfiguration ref-updated oldrev+newrev https://review.opendev.org/739078 | 16:56 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix github include/exclude unprotected cache https://review.opendev.org/740339 | 16:56 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix github branch protection while already unprotected https://review.opendev.org/740340 | 16:56 |
*** masterpe has quit IRC | 16:56 | |
*** webknjaz has joined #zuul | 16:58 | |
*** ChrisShort has joined #zuul | 16:58 | |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: scheduler: Fix event process abide hasUnparsedBranchCache argument https://review.opendev.org/739042 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix github include/exclude unprotected cache https://review.opendev.org/740339 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix github branch protection while already unprotected https://review.opendev.org/740340 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: Fix branch name and project name for ref-updated create/delete https://review.opendev.org/738320 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: FakeGerritChange: Add Change-Id in commit message https://review.opendev.org/739197 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: WIP: Scheduler: Reconfiguration ref-updated create/delete https://review.opendev.org/739198 | 17:00 |
openstackgerrit | Guillaume Chauvel proposed zuul/zuul master: WIP: Scheduler: Reconfiguration ref-updated oldrev+newrev https://review.opendev.org/739078 | 17:01 |
*** masterpe has joined #zuul | 17:02 | |
*** bhavikdbavishi has quit IRC | 17:02 | |
*** dennis_effa has quit IRC | 17:03 | |
*** bhavikdbavishi has joined #zuul | 17:04 | |
*** sshnaidm is now known as sshnaidm|afk | 17:12 | |
*** nils has quit IRC | 17:25 | |
*** mordred has joined #zuul | 17:25 | |
*** jamesmcarthur has quit IRC | 17:29 | |
*** jamesmcarthur has joined #zuul | 17:31 | |
*** hashar has joined #zuul | 17:35 | |
guillaumec | mordred, I was reading the document “Gerrit CI Experience Reboot" and inadvertently closed your comment, sorry | 17:36 |
guillaumec | reopened it ! | 17:37 |
*** jamesmcarthur has quit IRC | 17:41 | |
*** jamesmcarthur has joined #zuul | 17:43 | |
*** jamesmcarthur has quit IRC | 17:54 | |
*** jamesmcarthur has joined #zuul | 18:00 | |
*** jamesmcarthur has quit IRC | 18:03 | |
*** jamesmcarthur has joined #zuul | 18:10 | |
*** jamesmcarthur has quit IRC | 18:16 | |
corvus | mordred: it looks like https://review.opendev.org/726263 may have caught a fish; i'll take a look | 18:17 |
corvus | mordred: um, i set the wrong job name because i forgot we were in check not gate | 18:18 |
*** jamesmcarthur has joined #zuul | 18:20 | |
openstackgerrit | James E. Blair proposed zuul/zuul master: Add an icon next to result buttons on the console log https://review.opendev.org/740345 | 18:33 |
*** EmilienM is now known as EmilienM|PTO | 18:38 | |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: add-build-sshkey: Ensure .ssh exists, enable admin authorized_keys https://review.opendev.org/740350 | 19:02 |
openstackgerrit | Albin Vass proposed zuul/zuul-jobs master: add-build-sshkey: Ensure .ssh exists, enable admin authorized_keys https://review.opendev.org/740350 | 19:05 |
*** jamesmcarthur has quit IRC | 19:07 | |
*** jamesmcarthur has joined #zuul | 19:14 | |
*** jamesmcarthur has quit IRC | 19:14 | |
*** jamesmcarthur has joined #zuul | 19:14 | |
*** rlandy|ruck2 is now known as rlandy|brb | 19:15 | |
*** y2kenny has joined #zuul | 19:26 | |
*** tosky has quit IRC | 19:26 | |
*** rlandy|brb is now known as rlandy | 19:36 | |
*** bhavikdbavishi has quit IRC | 19:42 | |
corvus | mordred: okay, caught 2 this time | 19:47 |
*** jamesmcarthur has quit IRC | 19:47 | |
*** jamesmcarthur has joined #zuul | 19:47 | |
*** y2kenny has quit IRC | 19:48 | |
tobiash | bug fishing? | 19:48 |
corvus | yeah, autoholding on https://review.opendev.org/726263 to try to figure out why we end up with docker images from the wrong architecture | 19:49 |
tobiash | interesting, I never did multi arch docker | 19:49 |
*** armstrongs has joined #zuul | 19:57 | |
corvus | mordred: inside the buildkit container, "buildctl du -v" shows the contents of the cache. all of the layers from python-builder say they're pulled from the amd64 manifest, all the layers for python-base say they're pulled from the arm64 manifest. | 20:04 |
corvus | that makes me think that perhaps the buildkit cache is not able to hold both arches at the same time | 20:05 |
*** armstrongs has quit IRC | 20:05 | |
corvus | or, at least it doesn't know to pull a second copy of something it already has (in the case of needing a previous layer) | 20:05 |
corvus | i don't see any arch information in the layer cache | 20:06 |
*** smyers has quit IRC | 20:11 | |
tobiash | hrm, so separate jobs for amd64 and arm64? | 20:13 |
tobiash | or cleaning the caches after each arch | 20:14 |
tobiash | or I just don't understand this multiarch stuff ;) | 20:14 |
corvus | tobiash: that's an option, but probably only for builds with different tags. we're aiming for a single tag; buildkit is supposed to produce a single manifest list at the end of the build which contains the image manifests for all arches. it may not be able to do that if we clear too much out | 20:16 |
tobiash | ah ok, I'll shut up :) | 20:16 |
corvus | please don't :) | 20:16 |
tobiash | looks like I have no idea how that multiarch docker stuff works :) | 20:17 |
corvus | it's pretty new to all of us | 20:18 |
corvus | i also see layers in the cache for things we've actually build; and in those cases, it does look like we may have one from each arch | 20:20 |
corvus | so maybe the fault is in the 'pull' aspect of this | 20:20 |
corvus | i wonder if we used multiple workers inside of a builder if that would behave differently? | 20:21 |
corvus | i also wonder what would happen if we did a single build+push with --no-cache | 20:24 |
tobiash | corvus: could this be related? https://github.com/moby/buildkit/issues/1057 | 20:25 |
tobiash | corvus: the last sentence of the first answer sounds interesting: "Currently, it should always work correctly if you do docker build --pull ." | 20:26 |
corvus | tobiash: maybe? | 20:26 |
corvus | and yeah | 20:26 |
corvus | i'm going to try some reproduction | 20:27 |
corvus | interestingly, i just re-ran the command in the build and it succeeded; it also downloaded the other 2 missing images | 20:29 |
*** smcginnis has joined #zuul | 20:30 | |
smcginnis | Some conversation with fungi in #opendev on this, but I am hoping to get some feedback on the approach in https://review.opendev.org/#/c/739272/2 | 20:30 |
smcginnis | This is to address an issue we're hitting with the nightly openstack/requirements update job that tries to use each version of python to generate updated requirements. | 20:31 |
corvus | i thought there was some work for the roles to be able to use a bunch of python installations? | 20:31 |
avass | smcginnis: does it need to be a loop? | 20:32 |
smcginnis | avass: Yes, it needs to install the package for each ensure_pip_from_upstream_interpreters specified. | 20:32 |
avass | smcginnis: I mean, you could probably add '-venv' to each item in the list and template the list instead of using a loop :) | 20:33 |
smcginnis | So for example here: https://packages.ubuntu.com/search?suite=bionic-updates&searchon=names&keywords=venv | 20:33 |
smcginnis | avass: We may be running up against my ansible knowledge here. :) | 20:33 |
*** jamesmcarthur has quit IRC | 20:34 | |
*** jamesmcarthur has joined #zuul | 20:35 | |
smcginnis | corvus: Do you mean this? https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/ensure-pip/defaults/main.yaml#L5 | 20:35 |
smcginnis | I thought that was all in place. | 20:35 |
corvus | smcginnis, avass: i thought there was talk about using some gnu util that does things with tarballs to have a bunch of python versions installed; but maybe that's an optimization that isn't relevant here yet? | 20:36 |
*** jamesmcarthur has quit IRC | 20:37 | |
avass | corvus: do you mean pyenv? | 20:37 |
*** jamesmcarthur has joined #zuul | 20:37 | |
corvus | nope, wasn't python specific | 20:37 |
corvus | stow | 20:37 |
corvus | gnu stow | 20:38 |
corvus | anyway, i think that was intended as an optimization; it seems like this is addressing functionality we expect to work without requiring that | 20:38 |
avass | oh, haven't heard of that. taking a look | 20:38 |
corvus | avass: basically the idea was to build images with a bunch of different pythons pre-installed from upstream (not distro) with stow, and be able to access all of them. i think. my memory is fuzzy on this. :) | 20:39 |
smcginnis | Just for context, the issue I'm trying to address is a nightly job that runs. | 20:39 |
avass | corvus: is this for things like we did with tox to symlink it to /usr/bin to appear in PATH? | 20:39 |
smcginnis | We use ensure-pip: https://zuul.opendev.org/t/openstack/build/50cfc14cc11f4ded8877438f78b8585a/log/job-output.txt#306-342 | 20:39 |
corvus | avass: yeah | 20:39 |
avass | ok I think I got it then | 20:39 |
corvus | smcginnis: understood; just trying to bring up all relevant context :) | 20:39 |
smcginnis | But then end up erroring out here: https://zuul.opendev.org/t/openstack/build/50cfc14cc11f4ded8877438f78b8585a/log/job-output.txt#850 | 20:39 |
smcginnis | corvus: Yeah, same here. Don't want to be one of those people that pop in with no context and say hey, merge my code. | 20:40 |
smcginnis | :) | 20:40 |
smcginnis | I'm stumbling through this, so if I'm not on the right track, I'm very open to getting some direction. | 20:41 |
avass | smcginnis: I think it looks good but I think we can avoid the loop | 20:42 |
smcginnis | avass: Do you have an example of what that looks like? | 20:42 |
smcginnis | Happy to update it, but I don't think I've used that syntax yet. | 20:43 |
avass | smcginnis: I'm trying to figure out how to do it :) | 20:44 |
corvus | smcginnis: i also agree that this seems reasonable; i left one minor ansible thing in there; i agree with avass that avoiding the loop should be more efficient, so worth doing | 20:44 |
openstackgerrit | Sorin Sbarnea (zbr) proposed zuul/zuul master: Add user preferences dialog https://review.opendev.org/739482 | 20:44 |
corvus | smcginnis: (a package command inside of a loop is basically a series of "apt-get update; apt-get install" commands in sequence, whereas doing a list is "apt-get update; apt-get install foo bar" | 20:44 |
smcginnis | Just for my education because I'm curious, do loops in ansible have a lot of overhead? | 20:45 |
smcginnis | Ahh. | 20:45 |
smcginnis | So if we get rid of the loop, only one apt update, multiple apt installs. | 20:46 |
avass | corvus: actually for the 'package' module on debian it's just a apt-get install | 20:46 |
corvus | smcginnis: a little, but generally not enough to worry; this is a special case worth avoiding :) | 20:46 |
avass | it never runs apt-get update | 20:46 |
corvus | avass: well, still install in series | 20:46 |
avass | yeah | 20:46 |
smcginnis | So the idea is to change the list of [python3.6, python3.7] to one "apt install python3.6-venv python3.7-venv". Got it. | 20:46 |
smcginnis | Hmm, noticing one other thing. | 20:48 |
smcginnis | avass, corvus: Should we have another _from_packages version of ensure_pip_from_upstream_interpreters? | 20:48 |
avass | corvus: templating in name is possible | 20:48 |
*** bolg has quit IRC | 20:48 | |
smcginnis | Will it be too confusing to have an upstream list used for installing packages? | 20:48 |
corvus | avass: never works for me :) | 20:49 |
avass | corvus: huh | 20:49 |
corvus | avass: at least in loops | 20:49 |
corvus | i think the task itself sits outside the loop | 20:49 |
corvus | anyway, it's a moot point if we drop the loop :) | 20:50 |
avass | corvus: yeah, can't template the the loop_var in name | 20:51 |
avass | in my opinion it's better to not template the name so it's easier to find the task :) | 20:51 |
corvus | true | 20:51 |
corvus | i have increased the accuracy of my comment for posterity :) | 20:53 |
smcginnis | Does a jinja file need to be added to do the templating? Or is there a way to specify that inline? | 20:53 |
avass | corvus: :) | 20:54 |
corvus | smcginnis: can all be done inline with something vaguely resembling a list comprehension | 20:54 |
corvus | (a filter chain) | 20:54 |
corvus | avass maybe working on that? if not, i can | 20:55 |
avass | okay I think I got it | 20:55 |
corvus | cool | 20:55 |
avass | but there must be an easier way to do this | 20:55 |
corvus | some days i think embedded postscript would be easier than jinja | 20:56 |
avass | I commented | 20:56 |
corvus | avass: i like that, and i think there's probably not an easier way | 20:57 |
*** jamesmcarthur has quit IRC | 20:57 | |
openstackgerrit | Sean McGinnis proposed zuul/zuul-jobs master: Install venv for all platforms in ensure-pip https://review.opendev.org/739272 | 20:57 |
smcginnis | I think I got that. ^ | 20:57 |
corvus | smcginnis: drop the '-' so it's just "name: {{ ... }}" | 20:58 |
avass | corvus: I just wanted a regular list comprehension instead of creating a list of lists then joining the inner lists together :( | 20:58 |
smcginnis | Still the earlier question I threw in the middle though - should a _for_upstream variable be used in a _for_packages task. | 20:58 |
*** jamesmcarthur has joined #zuul | 20:58 | |
corvus | smcginnis: (the result of the jinja filter chain will itself be a list) | 20:58 |
fungi | trying to follow along while cooking dinner, i think the gnu stow work was going on in the ensure-python role | 20:58 |
openstackgerrit | Sean McGinnis proposed zuul/zuul-jobs master: Install venv for all platforms in ensure-pip https://review.opendev.org/739272 | 20:58 |
avass | smcginnis: I think you need to have the string on the same line as 'name: ' | 21:01 |
corvus | smcginnis: maybe a new var makes sense? i'd probably defer to others on that | 21:01 |
corvus | and should we ask ianw to review this? | 21:01 |
avass | or is that valid jinja? | 21:01 |
avass | uh, *yaml | 21:01 |
* smcginnis hovers over :wq | 21:01 | |
corvus | i think that's valid yaml, but not 100%. i've never done it. :) | 21:02 |
smcginnis | Safer if I change it? | 21:02 |
corvus | more consistent if that helps :) | 21:02 |
openstackgerrit | Sean McGinnis proposed zuul/zuul-jobs master: Install venv for all platforms in ensure-pip https://review.opendev.org/739272 | 21:02 |
avass | lgtm | 21:03 |
smcginnis | I guess it's probably good keeping with ensure_pip_from_upstream_interpreters rather than adding another pretty much identical variable. Or going through deprecation of the one for a common variable. | 21:04 |
smcginnis | I would expect this isn't something that is going to be needed often. | 21:04 |
fungi | seems likely to be needed by any job which uses more than one python interpreter version over the course of the build | 21:05 |
*** jamesmcarthur has quit IRC | 21:06 | |
smcginnis | At least there's a way to do it, even if the choice of variable name might be slightly confusing. | 21:07 |
*** jamesmcarthur has joined #zuul | 21:07 | |
*** jamesmcarthur has quit IRC | 21:11 | |
*** jamesmcarthur has joined #zuul | 21:11 | |
fungi | we might ought to consider making that variable an alias for a better name and deprecating the old name | 21:13 |
avass | smcginnis: deprecating ensure_pip_from_upstream_interpreters should be too hard. you could add something like ensure_pip_from_python_interpreters that defaults to ensure_pip_from_upstream_interpreters if it's set otherwise ansible_python.executable | 21:14 |
avass | shouldn't be too hard* | 21:15 |
*** yolanda has quit IRC | 21:16 | |
*** yolanda has joined #zuul | 21:17 | |
smcginnis | avass: Do you think I should do that in this patch? | 21:17 |
avass | smcginnis: I'd rather do it sooner than later, but I don't think it's necessary | 21:21 |
smcginnis | Yeah, doesn't look too bad. I'll just add it here. | 21:22 |
avass | but if it's just changing the name of a variable it shouldn't be hard. unless someone else can tell me that I'm wrong :) | 21:22 |
smcginnis | avass: I guess the tricky part could be that I can default them to be the same, but then someone may be using one or the other. So I think I would have to check both vars? | 21:23 |
avass | ensure_pip_from_python_interpreters: "{{ ensure_pip_from_upstream_interpreters | default(ansible_python.executable) }}" | 21:24 |
avass | smcginnis: I think that should be all, and remove the default value of ensure_pip_from_upstream_interpreters :) | 21:24 |
avass | smcginnis: in roles/ensure-pip/defaults/main.yaml if that wasn't clear | 21:25 |
smcginnis | Ah, I was thinking the defaults would get loaded first, then overwritten by what someone sets in their own variables. | 21:26 |
avass | smcginnis: that's the idea :) | 21:26 |
smcginnis | avass: Wait, I want to make sure I have that right. Are you saying what I was thinking is actually the right order? | 21:27 |
avass | yeah | 21:28 |
avass | the idea is that if the old variable is set it will be used, if it's not set it will use the default variable. but the new variable is prioritized | 21:29 |
smcginnis | OK, so then my concern was valid. So initially _from_python and _from_upstream are set the same, and all is good. Then someone locally sets _from_upstream to the list of interpreters they want. Now they are no longer the same. And in the task we use the newer _from_python variable, which is still defaulted to the initial value and not the new value they have specified. | 21:29 |
avass | smcginnis: oh yeah you're right, that could be a problem if both are set. | 21:32 |
smcginnis | I guess if we default ensure_pip_from_upstream_interpreters to be empty, then we could do that conditional check? | 21:32 |
avass | I think it might actually be harder than I thought :) | 21:34 |
smcginnis | Things always look too easy at first. ;) | 21:35 |
smcginnis | I'll leave my patch as is then. If there's anything I can change to make it better, I'll watch for comments. Maybe someone has a good idea how to handle that. | 21:35 |
smcginnis | But at least I think it's just a little non-obvious with the variable name. It's just a cosmetic issue. | 21:36 |
corvus | oh, the other shas i got running it the second time aren't the other arches, they are the shas from dockerhub since the buildset registry is gone | 21:39 |
corvus | this may be hard to reproduce without also holding the registry | 21:40 |
*** avass has quit IRC | 21:53 | |
corvus | running buildx one platform at a time, it still seems to download only one sha; there must be more going on that we can't see | 21:58 |
fungi | pay no attention to the whale behind the curtain | 22:02 |
*** hashar has quit IRC | 22:13 | |
*** smyers has joined #zuul | 22:34 | |
*** piotrowskim has quit IRC | 22:54 | |
*** rfolco has joined #zuul | 23:03 | |
*** rfolco has quit IRC | 23:07 | |
*** harrymichal has joined #zuul | 23:35 | |
*** dmellado has quit IRC | 23:39 | |
*** rlandy has quit IRC | 23:58 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!