*** waltminer has joined #automotive | 00:03 | |
*** waltminer has quit IRC | 00:30 | |
*** onion123 has joined #automotive | 01:18 | |
*** mdurnev has joined #automotive | 01:35 | |
*** onion123 has quit IRC | 02:26 | |
*** waltminer has joined #automotive | 02:31 | |
*** waltminer has quit IRC | 02:40 | |
*** dl9pf has quit IRC | 03:07 | |
*** waltminer has joined #automotive | 03:08 | |
*** mdurnev1 has joined #automotive | 03:26 | |
*** mdurnev has quit IRC | 03:27 | |
*** mdurnev1 is now known as mdurnev | 03:27 | |
*** waltminer has quit IRC | 03:41 | |
*** AlisonChaiken has joined #automotive | 05:28 | |
*** ToxicGumbo-work1 has joined #automotive | 05:53 | |
*** ToxicGumbo-work has quit IRC | 05:53 | |
*** ashwasimha_ has joined #automotive | 06:34 | |
*** gunnarx has quit IRC | 06:51 | |
*** rajm has joined #automotive | 07:18 | |
*** rdale has joined #automotive | 07:22 | |
*** CTtpollard has joined #automotive | 07:27 | |
*** gunnarx has joined #automotive | 07:38 | |
*** gunnarx has joined #automotive | 07:38 | |
CTtpollard | morning all | 07:38 |
---|---|---|
gunnarx | Morning and welcome back CTtpollard | 07:38 |
gunnarx | or maybe you were back last week... | 07:39 |
CTtpollard | back today :) | 07:39 |
gunnarx | Some people have prepared some pull requests for you to study. :) | 07:40 |
CTtpollard | to the rain of Manchester | 07:40 |
gunnarx | aaah, perfect work weather then | 07:40 |
*** RzR has quit IRC | 07:42 | |
*** kbroulik has joined #automotive | 07:45 | |
gunnarx | I have some Go.CD related discussion to have with you also, and pedroalvarez if he is interested :) | 07:46 |
CTtpollard | sure :) | 07:48 |
gunnarx | I've been doing some general cleanup. | 07:50 |
gunnarx | So the "with SDK" recipes are working again and they were previously testing the upload of sources from downloads folder | 07:50 |
*** reinar has joined #automotive | 07:50 | |
gunnarx | but I found uploading downloads/* does not work now that DL_DIR is outside of the pipeline directory, so we need to rethink that | 07:51 |
CTtpollard | that's odd | 07:54 |
gunnarx | it's a Go constraint - it will not allow uploading anything from outside of the pipeline | 07:54 |
gunnarx | uploading = define as artifact | 07:54 |
gunnarx | maybe a good thing | 07:54 |
CTtpollard | yeh I suppose | 07:56 |
gunnarx | The build itself can access the sources though, so it doesn't make too much sense from security perspective. I imagine a task could copy all sources into the pipeline before uploading them | 07:56 |
CTtpollard | can we not link the locations? | 07:57 |
gunnarx | we could try | 07:57 |
gunnarx | I used to have symlinks... actually I do have on my agent. /var/cache/yocto/downloads is symlinked to another dir | 07:57 |
gunnarx | but yes, I don't think there's a symlink in <pipeline>/build/downloads | 07:58 |
gunnarx | ok, so experiment one is to create that symlink in one of the first tasks | 07:58 |
gunnarx | Then, even more consolidation of templates maybe. I've added silk to a few that supported only porter, but some remain. | 08:00 |
*** dl9pf has joined #automotive | 08:01 | |
*** dl9pf has joined #automotive | 08:01 | |
pedroalvarez | oh, didn't know we were interested on uploading the downloads dir | 08:01 |
pedroalvarez | good morning btw :) | 08:01 |
pedroalvarez | now that the dir is shared.. wouldn't that be too much to upload? | 08:02 |
gunnarx | it was highly important when discussing how to publish binaries :) | 08:02 |
gunnarx | good morning... | 08:02 |
gunnarx | pedroalvarez, good point, it is a superset of all sources now. However, there's the archiver class which is likely better to use. | 08:02 |
gunnarx | OK, thanks pedroalvarez - for sources, set up again a test using archiver class. | 08:03 |
gunnarx | that's action #2. | 08:03 |
gunnarx | Hmm, I'm feeling executive powers today, must be the coffee ;) | 08:04 |
CTtpollard | please share the caffeine around | 08:07 |
gunnarx | Uh wait, I guess action #2 replaces experiment #1 above. This is what happens when taking decisions too fast. | 08:07 |
gunnarx | Anyway, next thing. As I said "with SDK" recipes and more templating... What would you think about a task like: if $INCLUDE_SDK then bitbake gdp -c populate_sdk in the shared template | 08:10 |
gunnarx | Common template brings me to... I know I've asked before but do we need the 4 thread limit... I think some pipelines are not using it and they appear to be working. I guess it triggers some race condition, only some of the time? | 08:12 |
*** leon has joined #automotive | 08:13 | |
CTtpollard | We had to put it in place when there was a recurring race stemming from qt, I'd be happy to increase the cap and see how it goes | 08:13 |
*** leon is now known as Guest43901 | 08:13 | |
gunnarx | Have we upgraded Qt since? | 08:14 |
CTtpollard | Will have, it was quite a while a go from memory | 08:15 |
CTtpollard | I see no harm in running them for a bit without the hard limit on threads / make and see | 08:15 |
*** Guest43901 is now known as leon-anavi | 08:17 | |
leon-anavi | morning | 08:17 |
CTtpollard | hi leon-anavi | 08:17 |
gunnarx | OK, let's try removing it and see | 08:18 |
*** jonathanmaw has joined #automotive | 08:22 | |
gunnarx | What does CTtpollard and pedroalvarez think about having a common template that can build the SDK if the environment says it should (as suggested above) | 08:27 |
gunnarx | CTtpollard, a question. BSP submodules are not populated until you choose target with init right? | 08:27 |
CTtpollard | I can't see any reason against it if it means we can consolidate things further | 08:28 |
CTtpollard | gunnarx: correct | 08:28 |
gunnarx | I've seen some error where workingdir=renesas but it does not exist | 08:28 |
pedroalvarez | +1 to consolidation | 08:28 |
gunnarx | you know the conditional task we have, the working dir is still expected to exist... | 08:28 |
pedroalvarez | ooouch | 08:29 |
pedroalvarez | that's kind of stupid | 08:29 |
gunnarx | I think it works because we usually don't wipe everything nowadays | 08:30 |
gunnarx | anyway, instead of workingdir=meta-renesas and ./copy_gfx_binaries, let's just : workingdir=. and task = "cd meta-renesas && ./copy_gfx_binaries" | 08:30 |
gunnarx | because the actual task execution is conditional on target being RENESAS | 08:31 |
gunnarx | oops don't know why I'm shouting the name :) | 08:34 |
gunnarx | to be specific workingdir is meta-renesas/meta-rcar-gen2. So it is the inner directory I expect does not exist unless the submodule was populated, which it likely is not if the first built target is non-renesas | 08:35 |
radiofree | were you waving a fist in the air as you typed it? | 08:35 |
* rjek imagines Kirk shouting "RENESAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAS" | 08:37 | |
gunnarx | I don't recall, it's not totally unlikely :) | 08:37 |
radiofree | rjek: dito :) | 08:37 |
rjek | I have shouted that at $job-1 | 08:37 |
gunnarx | pedroalvarez, CTtpollard does what I wrote earlier make sense? | 08:37 |
pedroalvarez | it does | 08:39 |
CTtpollard | gunnarx: yep, and yes that subdir does not exist until meta-renesas has been init'd | 08:39 |
gunnarx | ok I just modified the generic recipe so either it all goes down in flames, or it's time for another fist pump, the celebration kind | 08:39 |
gunnarx | By the way, the reason for discussing it was so someone else can go and do it... :) but then I do it myself :) I have "some" other things that need doing. | 08:42 |
gunnarx | I recently removed unnecessary template meta-ivi-build.2, FYI | 08:42 |
*** RzR has joined #automotive | 08:53 | |
CTtpollard | hmm, sent a PR but no tests present in github gui yet | 09:50 |
CTtpollard | ah just lag | 09:57 |
gunnarx | trust the process! :) | 09:58 |
gunnarx | seriously it feels pretty stable now | 09:58 |
CTtpollard | looks like a wildcard in path is failing | 09:59 |
CTtpollard | 'The rule [gdp-src-build/tmp/deploy/licenses/genivi-dev-platform*/license.manifest] cannot match any resource under [pipelines/GDP-Yocto-RPI2-GITHUB-PR]' | 09:59 |
CTtpollard | same with porter | 09:59 |
CTtpollard | brb | 10:00 |
gunnarx | let me check | 10:00 |
gunnarx | CTtpollard, I think the build failed before that. | 10:23 |
CTtpollard | hmm a do unpack on audiomanager | 10:30 |
gunnarx | I found that logs.tar does not include these logs that are in work/temp so I adjusted it | 10:35 |
CTtpollard | not seeing anything glaring in meta-ivi changes to the audiomanager recipe that I'd think would cause the unpacking of a previous build to fail | 10:35 |
gunnarx | I can give you the logfile if it helps | 10:36 |
CTtpollard | there's a switch to fetch over git http that we applied to mgd recipes, that's about it | 10:36 |
CTtpollard | that might prove helpful, seems to be the same error on the 4 pipelines | 10:38 |
gunnarx | one sec | 10:40 |
gunnarx | http://paste.baserock.org/nuyipafuje | 10:40 |
gunnarx | Next time it should be in logs.tar :-P | 10:41 |
pedroalvarez | wow, Gunnar using paste.baserock.org. this has to mean something :P | 10:41 |
gunnarx | This was painful -- ssh to machine, attach to the docker container, find the files, cat them together, export them from docker container, scp them to my machine, clipboard the contents, paste to web... ugh | 10:41 |
gunnarx | someone should automate this... :) | 10:42 |
gunnarx | And I'd be surprised if you get anything out of it CTtpollard :) | 10:43 |
CTtpollard | the final patch it searches for is a recent addition to meta-ivi | 10:47 |
CTtpollard | i.e new since the cached audiomanager it's trying to unpack | 10:48 |
gunnarx | OK, so it fails because it should fail basically? | 10:48 |
CTtpollard | That's what I'm grasping out for now | 10:51 |
CTtpollard | trying to unpack a file that does not exist | 10:51 |
CTtpollard | it finds it in the local checkout of meta-ivi, but not in the cache | 10:52 |
gunnarx | that's no good then | 10:52 |
gunnarx | I goofed a bit on redefining logs.tar but fixed now | 10:53 |
CTtpollard | suppose the first course of action as per usual is to tell it to clean/wipe the shared cache for audiomanager | 10:55 |
CTtpollard | -c cleansstate audiomanager | 10:56 |
CTtpollard | if this is the issue though, it's not pleasant as it is quite a common scenario | 10:57 |
gunnarx | hope not | 10:59 |
gunnarx | recently I feel the PRs have built correctly, the errors have been real errors | 10:59 |
CTtpollard | I'll do the clean on the codethink agtent | 11:02 |
CTtpollard | as the porter was building on there | 11:02 |
gunnarx | Something funky going on... there's no DL_DIR defined in local.conf on the RPi one | 11:02 |
gunnarx | ok, so RPI now started on CT agent, and porter on 2131 whatever. porter dir is completely wiped. So I'd check the results of that... | 11:04 |
gunnarx | build is pretty quick now with caches and stuff. | 11:05 |
gunnarx | ^^ CTtpollard give it a few minutes | 11:06 |
CTtpollard | yup tacking | 11:06 |
CTtpollard | *tracking | 11:08 |
gunnarx | I found a revnant of a downloads catalog in R-Pi directory, and also for some reason DL_DIR not set. It might just be due to moving stuff around a bit. I hacked on the *generic script this morning after all... | 11:13 |
gunnarx | oops, remnant | 11:14 |
gunnarx | No revenant here. DiCaprio is nowhere to be seen... :) | 11:14 |
gunnarx | The two failing seem to work now CTtpollard, I triggered the other two. | 11:34 |
CTtpollard | thanks gunnarx, was on lunch sorry | 11:37 |
CTtpollard | so all the has been applied is a clean to am? | 11:38 |
gunnarx | no parse error | 11:49 |
gunnarx | waiting for the last two. If they succeed we'll mark it up as a glitch due to recent hacking around in the recipes :) | 11:50 |
CTtpollard | s/the/that | 11:53 |
gunnarx | weirdness ensues | 11:58 |
gunnarx | or continues in fact | 11:59 |
gunnarx | I'll run Minnowboard with fresh sstate, see what happens | 12:32 |
gunnarx | brb | 12:33 |
*** ToxicGumbo-work1 has quit IRC | 13:08 | |
*** ToxicGumbo-work has joined #automotive | 13:08 | |
*** chbae has joined #automotive | 13:31 | |
chbae | Hello | 13:38 |
CTtpollard | hi chbae | 13:38 |
chbae | Hi CTtpollard: Welcome to coming back :) | 13:39 |
*** toscalix has joined #automotive | 13:39 | |
CTtpollard | chbae: thanks, it was a nice break :) | 13:40 |
chbae | that’s good. | 13:40 |
*** oan has quit IRC | 13:40 | |
*** oan has joined #automotive | 13:42 | |
chbae | CTtpollard: Do you get the error in audiomanager for your meta-ivi upgrade PR? | 13:43 |
CTtpollard | chbae: I'm waiting for the pipelines to report back | 13:44 |
chbae | that’s fetch error because you need to change meta-genivi-dev’s audiomanager_7.0.bbappend | 13:44 |
chbae | I’m upgrading to GDP-11 and you can refer https://github.com/chbae/meta-genivi-dev/commit/0dfeade3c996949344375359cfa6ab77510ec5f4. | 13:45 |
CTtpollard | chbae: I'm doing the same, although I've switched to using meta-ivi11 src uri for audiomanager | 13:46 |
CTtpollard | as they now support the plugin repo recipe | 13:46 |
CTtpollard | chbae: but your point on meta-ivi10.0.1 PR, I fear issues like that may not be caught by the CI | 13:47 |
chbae | Yes. the error cause “SRC_URI_remove” in audiomanager_7.0.bbappend | 13:47 |
CTtpollard | as 2 of the pipelines have passed the PR test | 13:47 |
chbae | CTtpollard: Yes. that’s problem. I thought that it should get the error. | 13:48 |
*** mdurnev has quit IRC | 13:48 | |
CTtpollard | So far I've only found audiomanager & curl to be an issue using ivi11 | 13:48 |
*** nisha has joined #automotive | 13:48 | |
CTtpollard | chbae: I will test a clean build locally from scratch for that PR overnight, and report back | 13:49 |
chbae | audiomanager_7.0.bb in meta-ivi: SRC_URI = git://git.projects.genivi.org/AudioManager.git;branch=master;protocol=http | 13:49 |
chbae | audiomanager_7.0.bbappend in meta-genivi-dev: SRC_URI_remove = “git://git.projects.genivi.org/AudioManager.git;branch=master" | 13:49 |
chbae | The “protocol=http” is missed. | 13:50 |
* rjek blinks; what does protocol= set if not the protocol in the URL? | 13:51 | |
CTtpollard | it defaults to the url protocol | 13:51 |
CTtpollard | so git | 13:52 |
CTtpollard | chbae: yes I see that, RzR patch to support http in included in ivi10.0.1 | 13:52 |
CTtpollard | in this PR, I will report back | 13:52 |
chbae | Ok. | 13:52 |
gunnarx | rjek, I was informed about this point another time. By CTtpollard even I think ? :) | 14:05 |
rjek | Madness | 14:05 |
gunnarx | So if you wrote http:// then that means you want to wget something like a tarball | 14:05 |
gunnarx | therfore, it must be git:// protocol=http | 14:05 |
gunnarx | to do a git clone over http | 14:05 |
*** toscalix has quit IRC | 14:09 | |
RzR | CTtpollard, hi do I need to submit anything again ? | 14:13 |
CTtpollard | RzR: hmm, depends what you want to submit :P | 14:14 |
*** toscalix has joined #automotive | 14:14 | |
CTtpollard | if you're referring to when I pinged you, then no, I was just talking about a commit you have in meta-ivi | 14:14 |
RzR | ok | 14:24 |
*** dabukalam has quit IRC | 14:45 | |
*** chbae has quit IRC | 14:45 | |
*** dabukalam has joined #automotive | 14:46 | |
*** tjamison has joined #automotive | 14:55 | |
*** AlisonChaiken has quit IRC | 15:06 | |
*** mvick has joined #automotive | 15:10 | |
*** AlisonChaiken has joined #automotive | 15:43 | |
gunnarx | CTtpollard: Had a clean build running today and it failed same way - I used the minnowboard configuration. | 15:45 |
gunnarx | The PR builds are probably doing their job - we should trust them :) | 15:45 |
*** rajm has quit IRC | 15:47 | |
CTtpollard | I believe the PR will fail, but not due to the issue it is reporting | 15:50 |
CTtpollard | it should fail to parse the bbappend in m-g-d correctly for audiomanager | 15:53 |
*** kbroulik has quit IRC | 15:53 | |
CTtpollard | but maybe it would spot that, once it gets past the unpack error | 15:55 |
gunnarx | maybe | 15:55 |
CTtpollard | gunnarx: was your clean build still sharing cache? | 15:56 |
gunnarx | No, not unless I goofed again :) | 15:57 |
gunnarx | No, it was not. No SSTATE_DIR set in conf file, and the sstate dir is local | 15:58 |
gunnarx | Not even sharing download actually | 15:58 |
CTtpollard | and it had an unpack error? :S | 15:58 |
gunnarx | yes it's weird | 15:59 |
gunnarx | it's not finished I noticed, it's fetching kernel source :-/ | 15:59 |
gunnarx | So unpack problem comes quite early I guess | 15:59 |
*** jonathanmaw has quit IRC | 16:05 | |
*** ashwasimha_ has quit IRC | 16:11 | |
*** radiofree has quit IRC | 16:19 | |
*** radiofree has joined #automotive | 16:21 | |
*** toscalix has quit IRC | 16:22 | |
*** Saint_Isidore has joined #automotive | 16:26 | |
*** gunnarx has quit IRC | 16:30 | |
*** ashwasimha_ has joined #automotive | 16:30 | |
*** chbae has joined #automotive | 16:41 | |
*** leon-anavi has quit IRC | 16:57 | |
*** ToxicGumbo-work has quit IRC | 17:01 | |
*** ToxicGumbo-work has joined #automotive | 17:01 | |
*** chbae has quit IRC | 17:12 | |
*** rdale has quit IRC | 18:03 | |
*** jlrmagnus has joined #automotive | 18:27 | |
*** ashwasimha_ has quit IRC | 18:41 | |
*** Saint_Isidore has quit IRC | 18:56 | |
*** praneeth has quit IRC | 19:29 | |
*** praneeth has joined #automotive | 19:30 | |
*** Saint_Isidore has joined #automotive | 19:41 | |
*** jlrmagnus has quit IRC | 20:00 | |
*** jlrmagnus has joined #automotive | 20:15 | |
*** waltminer has joined #automotive | 20:50 | |
*** waltminer has left #automotive | 20:51 | |
*** jlrmagnus has quit IRC | 21:01 | |
*** waltminer has joined #automotive | 21:25 | |
*** jlrmagnus has joined #automotive | 21:31 | |
*** tlab has quit IRC | 21:34 | |
*** waltminer has quit IRC | 22:08 | |
*** jlrmagnus has quit IRC | 22:22 | |
*** nisha has quit IRC | 22:59 | |
*** tjamison has quit IRC | 23:28 | |
*** jlrmagnus has joined #automotive | 23:41 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!