| *** xjuan has quit IRC | 00:48 | |
| *** hergertme_ is now known as hergertme | 06:36 | |
| *** tristan has joined #buildstream | 06:47 | |
| *** mohan43u has quit IRC | 06:48 | |
| *** iker has joined #buildstream | 06:57 | |
| *** mohan43u has joined #buildstream | 07:02 | |
| paulsher1ood | +1 for persia's last comment | 07:20 |
|---|---|---|
| paulsher1ood | one has to choose which upstream to track... if you're going with a distro, that's your upstream, not the upstream they are tracking | 07:22 |
| *** finn has joined #buildstream | 07:39 | |
| *** finn has quit IRC | 07:40 | |
| *** finn has joined #buildstream | 07:41 | |
| *** bochecha has joined #buildstream | 07:50 | |
| bochecha | tristan: now that you removed blessings as a dependency, the Fedora maintainer decided to update it to a version that's recent enough for Buildstream 😏 | 07:54 |
| *** toscalix has joined #buildstream | 08:01 | |
| finn | Hi! So lastnight, I installed the bst-external plugins repo following the readme. I used: | 08:01 |
| finn | `pip show BuildStream-external` | 08:01 |
| finn | which tells me I've installed Version: 0.5. | 08:01 |
| finn | I then have a `test.bst` file which depends on `dock.bst`. | 08:02 |
| finn | Inside `dock.bst` I have tried: | 08:02 |
| finn | kind: import | 08:02 |
| finn | sources: | 08:02 |
| finn | - kind: docker | 08:02 |
| finn | I get this error: No Source type registered for kind 'docker' | 08:02 |
| finn | any idea on what's happening here? | 08:02 |
| Kinnison | Why not track the debian source via the Debian archive interfaces? | 08:03 |
| Kinnison | e.g. rmadison uses a CGI on qa.debian.org to learn about versions available in distributions, use that, then construct the appropriate filenames and download them from Debian infra | 08:03 |
| jennis | finn, have you declared the plugin in your project.conf? | 08:11 |
| jennis | https://buildstream.gitlab.io/buildstream/format_project.html#external-plugins | 08:12 |
| gitlab-br-bot | buildstream: merge request (Qinusty/skipped-rework-backport-1.2->bst-1.2: Backport skipped reword (!765)) #810 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/810 | 08:30 |
| qinusty | skipped backport should pass CI now tristan | 08:32 |
| gitlab-br-bot | buildstream: merge request (juerg/cas-batch->master: WIP: _artifactcache/cascache.py: Use BatchReadBlobs) #813 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/813 | 08:48 |
| juergbi | tristan: !813 implements the client side of batch download support. how shall we verify that this actually fixes #554? it requires a server update as well and the server side is not merged to bst-1.2 yet (it will still work with current server, but it can't use batch download in that case) | 08:51 |
| juergbi | I could backport the server part first, we update at least one server and then you could test with the branch from !813. or we merge it after review but without testing it across high latency connections, backport both to 1.2 and it can be verified afterwards | 08:53 |
| gitlab-br-bot | buildstream: merge request (willsalmon/outOfSourecBuild->master: Out of source builds) #776 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/776 | 09:19 |
| *** jonathanmaw has joined #buildstream | 09:22 | |
| *** lachlan has joined #buildstream | 09:23 | |
| *** lachlan_ has joined #buildstream | 09:23 | |
| gitlab-br-bot | buildstream: merge request (willsalmon/outOfSourecBuild->master: Out of source builds) #776 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/776 | 09:25 |
| *** ChanServ sets mode: +o tristan | 09:35 | |
| tristan | juergbi, Aha, this is great news ! | 09:35 |
| tristan | juergbi, Ok so... it won't make it into 1.2.1 anyway, but perhaps 1.2.1... | 09:37 |
| tristan | err 1.2.2 | 09:38 |
| tristan | juergbi, I am happy to merge both after releasing 1.2.1, and update one server and then compare performance | 09:39 |
| tpollard | Does buildstream have a public artifact server ooi? | 09:39 |
| lachlan | tristan: Did you manage to catch my question in the backlog yesterday? | 09:40 |
| tristan | juergbi, I think that even bochecha has been seeing very slow downloads, and he is in France, so it doesn't require *that* high latency | 09:40 |
| tristan | lachlan, just catching up now... no didn't catch that | 09:40 |
| lachlan | OK | 09:40 |
| tristan | bochecha, Anyway no worries, I've been meaning to remove blessings for a long time now | 09:40 |
| juergbi | ok. i'll just get this merged to master after review, then. and backport to 1.2 after 1.2.1 | 09:41 |
| gitlab-br-bot | buildstream: merge request (Qinusty/skipped-rework-backport-1.2->bst-1.2: Backport skipped reword (!765)) #810 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/810 | 09:41 |
| tristan | qinusty, merged \o/ | 09:41 |
| tristan | qinusty, I'll kick back and watch a nice build log now :) | 09:42 |
| tpollard | Also with the work that's gone into buildstream-integration, would it be work that becoming part of CI? | 09:43 |
| tristan | On a logging related topic, I want MessageType.LOG messages issued in cleanup and cachesize jobs; those log files are pointless, not for the lack of me opening them curiously and looking inside of them :) | 09:43 |
| juergbi | tristan: to be clear, you also don't want the server side backported for 1.2.1? it's in master for 10 days now | 09:43 |
| tristan | juergbi, I was going to ask about that actually | 09:44 |
| tristan | juergbi, That could be better, I wasn't sure of how mature that part was | 09:44 |
| tristan | juergbi, I say backport it, so we have servers updated earlier; and reduce fuss of testing the client side | 09:45 |
| juergbi | I've tested it with buildbox client and now also with buildstream client (MR) | 09:45 |
| juergbi | that was my thought as well | 09:45 |
| tristan | lachlan, I'm having a hard time finding your question in the logs... can you re-ask it please ? | 09:46 |
| jjardon | Hi, does buildstream have any checks as part of the CI to be sure the cache key doesn't change for stable branches? | 09:46 |
| juergbi | yes, there are | 09:46 |
| tristan | jjardon, tests/cachekey/ does that... however there is *some* parts not covered well | 09:46 |
| tristan | We don't have coverage for all the different build element types (but we have coverage for all different element types and source types) | 09:47 |
| tristan | jjardon, It's very easy to add this, if you care to throw up an MR, that would be much appreciated | 09:48 |
| jjardon | rigth, how is this configured in the CI? I guess it is allow to fail in master, but not in stable branches | 09:48 |
| tristan | Nope | 09:48 |
| tristan | jjardon, Whenever we intentionally break cache key algorithm, we have to also update the test | 09:49 |
| tristan | jjardon, Lemme show you one sec | 09:49 |
| jjardon | ah ok, so it will always fail the test, gotcha | 09:49 |
| tristan | https://gitlab.com/BuildStream/buildstream/tree/master/tests/cachekey/project/elements | 09:49 |
| gitlab-br-bot | buildstream: merge request (jmac/remote_exec_checkout_fix->master: Remote exec: Remove early warning and check directory is not None) #800 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/800 | 09:50 |
| tristan | jjardon, See there, all the different elements, we add an extra element .bst file to exercise any feature / configuration which can effect cache keys in any way | 09:50 |
| tristan | jjardon, And we have a .expected file for the expected cache key of each element | 09:50 |
| gitlab-br-bot | buildstream: merge request (juerg/cas-batch-1.2->bst-1.2: _artifactcache/casserver.py: Implement BatchReadBlobs) #814 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/814 | 09:51 |
| tristan | jjardon, When adding an element, or when the expected files need to be updated, it can be automated with the update.py script here: https://gitlab.com/BuildStream/buildstream/tree/master/tests/cachekey | 09:51 |
| tristan | jjardon, So it doesn't require manually updating the expected keys | 09:51 |
| tristan | There is a source/ and an elements/ directory... We could use just a bare naked extra test for each element kind | 09:52 |
| tristan | in the elements/ directory... those have to be depended on by target.bst to be included in the test: https://gitlab.com/BuildStream/buildstream/blob/master/tests/cachekey/project/target.bst | 09:52 |
| gitlab-br-bot | buildstream: merge request (juerg/cas-batch->master: _artifactcache/cascache.py: Use BatchReadBlobs) #813 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/813 | 09:53 |
| tristan | juergbi, That would give us the added assurance that we do not change any default element configurations in the element .yaml files | 09:53 |
| jjardon | mmm, would be a good idea to have another test that build a real thing and check all the keys are as expected? something like https://gitlab.com/baserock/ybd/blob/master/.gitlab-ci.yml#L30 | 09:53 |
| tristan | jjardon, I don't know what that would add | 09:53 |
| *** lachlan_ has quit IRC | 09:54 | |
| *** lachlan has quit IRC | 09:54 | |
| tristan | jjardon, Our test aims to capture every configuration that could possibly have any effect on cache keys, I think it's very thorough | 09:55 |
| tristan | jjardon, A build will only ever capture a subset of configurations, also; a build need not be run just to discover the cache keys | 09:55 |
| gitlab-br-bot | buildstream: issue #663 ("Missing cache key in log messages (workspace open)") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/663 | 09:57 |
| jjardon | tristan: note I said this as an additional test, not removing the current ones; you can always forget to add a check; this would be an extra. ( and you even said there is not coverage for all build types at the moment) | 09:59 |
| qinusty | Where does buildstream save information relating to the currently open workspaces? | 09:59 |
| qinusty | I can't find a workspaces.yml in my local directory | 09:59 |
| jjardon | tristan: is there an issue for that? maybe I have time over the weekend to add those | 10:00 |
| * qinusty finds .bst/workspaces.yml | 10:00 | |
| tristan | jjardon, technically; our test is doing the same thing as the ybd test you point out; with manufactured project data, designed specifically for exercising every possible option | 10:01 |
| tristan | jjardon, Also, CI is costly, we we prefer to have isolated tests which only test what they are testing for (i.e. not running huge builds) | 10:01 |
| tristan | jjardon, I will draw up an issue :) | 10:02 |
| bochecha | tristan: I don't think what I had were very slow downloads… I think they were completely hung; bst still reporting it was doing something, but no download was actually happening | 10:03 |
| tristan | bochecha, Ah right, but I recall you were saying that long standing (1hrs ?) downloads were rather frequent, no ? | 10:03 |
| tristan | bochecha, I do recall that specific case was a hang, indeed | 10:03 |
| bochecha | ah yes | 10:04 |
| bochecha | but then maybe they are just very big elements (e.g the flatpak images resulting from building fdo sdk) | 10:04 |
| tristan | On my part, I download for a long long long long time, and my system monitor shows like 3 or 4 kbps | 10:04 |
| jjardon | tristan: that CI job takes seconds; it's not actually building anything. It's a special ybd mode to only compute the keys I tihnk | 10:04 |
| tristan | jjardon, Sure, that's what we're doing, with `bst show`, using a project that is designed to exercise all the options thoroughly | 10:05 |
| bochecha | tristan: how do you monitor download speed? | 10:06 |
| tristan | bochecha, gnome-system-monitor, which means... the download is taking *at most* 3 or 4 kbps | 10:06 |
| tristan | i.e. overall system network activity monitor | 10:06 |
| bochecha | let me try this then :) | 10:07 |
| jjardon | tristan: coolio, thanks for clarifications | 10:07 |
| bochecha | oh, gnome-system-monitor isn't in flathub :( | 10:07 |
| tristan | jjardon, We already have an issue: https://gitlab.com/BuildStream/buildstream/issues/318 :) | 10:08 |
| qinusty | tristan, can you confirm that this is undesired behaviour ? https://gitlab.com/BuildStream/buildstream/issues/663 | 10:08 |
| tristan | qinusty, Confirmed | 10:10 |
| qinusty | Might have time to fix it this afternoon if it doesn't go too deep down the rabbit hole | 10:10 |
| tristan | qinusty, It is *also* a bug that /usr/bin/git show 72b5902157316e173de2eec5b3a2772283eec3c7:.gitmodules is being shown at all | 10:12 |
| qinusty | noted. | 10:13 |
| tristan | qinusty, The messages which pertain to resolving element state, stemming from `Source.get_consistency()` are not supposed to be shown | 10:13 |
| bochecha | tristan: download speed is not too bad here actually: https://i.imgur.com/SUDYN3e.png | 10:15 |
| bochecha | tristan: I had only IRC using the network at the same time | 10:15 |
| bochecha | (that's a `bst build flatpak-images/sdk.bst` on a revision which is in the remote cache, only download involved no actual build) | 10:16 |
| *** lachlan has joined #buildstream | 10:18 | |
| *** lachlan_ has joined #buildstream | 10:18 | |
| tristan | bochecha, performance will be improved with juergbi's work on batch downloads, though :) | 10:23 |
| tristan | juergbi, Curiously, does CAS use pretty much the same approach we had with OSTree for push ? i.e. we were doing Tar streams and upload in a single stream | 10:23 |
| juergbi | tristan: no, push is still non-batch, unfortunately. want to fix this as well | 10:24 |
| tristan | bochecha, Note that for me, it really varies, and I believe it depends mostly on the size of the files being transfered | 10:24 |
| tristan | juergbi, Alright, when that is fixed, let's bring that into 1.2 as well; however pull is going to make a bigger impact on users so, very happy to have that now :) | 10:25 |
| qinusty | WSalmon, after having time to investigate that area of the code a little. It appears that logic was only intended to run on a success anyway. | 10:29 |
| bochecha | tristan: possibly yes… could the server create one big file, so that the client only makes one big download at max speed, and then splits it back? | 10:30 |
| qinusty | It isn't caching, just updating the workspace for the latest successful ref | 10:31 |
| WSalmon | i wonder why its ok for the non workspace case... | 10:31 |
| WSalmon | cool sounds good | 10:31 |
| qinusty | The issue is purely related to workspacing, it only runs when a workspace is open. | 10:36 |
| qinusty | Are you able to check the patch against the issue you were seeing incase I'm fixing a separate issue? | 10:37 |
| bochecha | tristan: fwiw it finished downloading in 23 minutes, downloaded 1700MB, which is roughly 1.2MB/s average… if juergi makes it faster, that's going to be really good! :) | 10:41 |
| tristan | bochecha, :) | 10:42 |
| tristan | bochecha, The batch downloads should help, maybe not that much on your connection, not sure, but they should help a lot I think | 10:43 |
| tristan | bochecha, round trips are expensive | 10:43 |
| bochecha | what will the batch download actually do? | 10:43 |
| tristan | bochecha, FWIW, the upload with OSTree did not need to "create a huge file", we used python Tar, which allows "streaming", so on the client we add files to the stream and on the server we unpack the incomming files from the tar stream | 10:43 |
| tristan | bochecha, It will send multiple files over the same connection in batches, reducing round trips; from what I understand | 10:44 |
| bochecha | nice | 10:44 |
| gitlab-br-bot | buildstream: issue #664 ("Current buildstream master is failing to build freedesktop-sdk") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/664 | 10:44 |
| bochecha | streaming it all as one tar file could be nice as well | 10:44 |
| Kinnison | It's worth noting that we don't really want to be adding more BuildStream specific RPCs | 10:45 |
| Kinnison | Until/unless we cannot achieve reasonable performance any other way | 10:46 |
| *** lachlan_ has quit IRC | 11:02 | |
| *** lachlan has quit IRC | 11:02 | |
| toscalix | valentind: I wonder what is left to be done here https://gitlab.com/BuildStream/website/merge_requests/88 | 11:24 |
| valentind | toscalix, just approval. | 11:25 |
| toscalix | assign to tiagogomes_ then, please | 11:26 |
| toscalix | done | 11:26 |
| *** mohan43u has quit IRC | 11:29 | |
| toscalix | WSalmon: now the agenda section for the BuildSTream Conclave has been created, would you mind adding there your topic suggestion? | 11:37 |
| toscalix | WSalmon: https://wiki.gnome.org/Projects/BuildStream/Events#Proposed_topics | 11:38 |
| *** lachlan has joined #buildstream | 11:40 | |
| persia | Kinnison: Indeed. Using madison.cgi for tracking debian-format upstream archives is far superior to watch files or anything else for derivation purposes. | 11:52 |
| *** mohan43u has joined #buildstream | 11:54 | |
| tristan | qinusty, Is https://gitlab.com/BuildStream/buildstream/issues/515 fixed ? | 12:14 |
| gitlab-br-bot | buildstream: issue #515 ("`bst push` will report success even if the local artifacts don't exist (and therefore aren't pushed)") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/515 | 12:15 |
| qinusty | Indeed :) | 12:15 |
| tristan | qinusty, and https://gitlab.com/BuildStream/buildstream/issues/614 also ? | 12:15 |
| gitlab-br-bot | buildstream: issue #614 ("Follow-up from "Bschubert/backport log missed cache"") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/614 | 12:16 |
| qinusty | This is now much more intuitive in my opinion. | 12:16 |
| qinusty | Yup | 12:16 |
| qinusty | Apologies. | 12:16 |
| tristan | qinusty, Thanks, please close them :D | 12:16 |
| tristan | https://github.com/pallets/click/pull/782#issuecomment-421404152 | 12:19 |
| tristan | Looks like they are now finally considering my tab completion enhancements upstream \o/ | 12:19 |
| tristan | With some luck, and a couple days of work... maybe we can drop all those hacks we have in our completion re-implementation | 12:19 |
| *** alatiera_ has joined #buildstream | 12:20 | |
| tristan | juergbi, Are you ready with the backport of server side batching support ? | 12:21 |
| tristan | juergbi, I could delay and release tomorrow, am writing up NEWS presently | 12:21 |
| *** alatiera__ has joined #buildstream | 12:22 | |
| gitlab-br-bot | buildstream: merge request (Qinusty/634-workspace-failed-builds->master: WIP: Do not save workspace on failed build) #812 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/812 | 12:24 |
| *** alatiera_ has quit IRC | 12:24 | |
| gitlab-br-bot | buildstream: merge request (Qinusty/634-workspace-failed-builds->master: Do not save workspace on failed build) #812 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/812 | 12:26 |
| *** alatiera_ has joined #buildstream | 12:41 | |
| *** alatiera__ has quit IRC | 12:42 | |
| *** alatiera__ has joined #buildstream | 12:43 | |
| *** alatiera_ has quit IRC | 12:44 | |
| WSalmon | toscalix, i dont seem to be able to edit the page do i need to be added to something? or am i looking in the wrong place | 12:51 |
| toscalix | you need an account in the gnome wiki | 12:52 |
| WSalmon | just have | 12:52 |
| toscalix | jjardon: might help you there | 12:52 |
| toscalix | then you should be able | 12:52 |
| juergbi | tristan: I already opened !814 for the backport | 12:53 |
| WSalmon | were is the edit button? | 12:53 |
| WSalmon | toscalix, ^ | 12:54 |
| gitlab-br-bot | buildstream: merge request (Qinusty/634-workspace-failed-builds->master: Do not save workspace on failed build) #812 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/812 | 12:54 |
| toscalix | there is on the upper right side | 12:54 |
| qinusty | Can I get a review on https://gitlab.com/BuildStream/buildstream/merge_requests/799? | 12:54 |
| toscalix | if you do not have it.... something else needs to be done | 12:54 |
| toscalix | I am a simple user.... jjardon knows what to do | 12:55 |
| WSalmon | all i can see is a little padlok that says "Immutable Page" | 12:55 |
| tristan | tlater[m], Care to comment on this https://gitlab.com/BuildStream/buildstream/merge_requests/812#note_103112195 ? | 12:55 |
| tristan | tlater[m], Sorry I keep getting confused around there | 12:55 |
| jjardon | WSalmon: what is your user in the gnome wiki? I will add you to the white list | 12:56 |
| WSalmon | i am WilliamSalmon | 12:56 |
| tlater[m] | tristan: Sure | 12:56 |
| WSalmon | i didnt realise that was gona be my username | 12:56 |
| tlater[m] | Ah, that comment again... Let's hope I still understand that. | 12:57 |
| qinusty | tristan, I'm not familiar enough with the usage of workspaces to fully answer your queries on the desired behaviour there. Regarding https://gitlab.com/BuildStream/buildstream/merge_requests/812 | 12:57 |
| jjardon | WSalmon: added you to https://wiki.gnome.org/TrustedEditorGroup; can you try now, please? | 12:57 |
| gitlab-br-bot | buildstream: merge request (juerg/cas-batch-1.2->bst-1.2: _artifactcache/casserver.py: Implement BatchReadBlobs) #814 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/814 | 12:57 |
| tristan | juergbi, Ah sorry didn't notice this, thanks ! | 12:58 |
| tristan | qinusty, That's why I asked tlater[m], I have a feeling your patch is totally correct, but this is always confusing | 12:58 |
| qinusty | The patch fixed a bug and corrected the behaviour to what was stated in the surrounding comments. I guess you're suggesting there is a larger problem at hand following the caching of failed builds | 12:58 |
| WSalmon | I can edit, but need to disapear to a meeting, will update the page when i am back | 12:58 |
| tristan | qinusty, It might be good to digest what tlater[m] has to say, and update the comment... not sure the rationale behind how this works is completely explained anywhere in the code | 12:59 |
| qinusty | agreed | 12:59 |
| *** finn_ has joined #buildstream | 12:59 | |
| *** phildawson has quit IRC | 13:00 | |
| *** phildawson has joined #buildstream | 13:00 | |
| *** finn has quit IRC | 13:01 | |
| *** jonathanmaw_ has joined #buildstream | 13:01 | |
| *** jonathanmaw has quit IRC | 13:02 | |
| tristan | whoa | 13:07 |
| * tristan just reproduced an old issue | 13:07 | |
| tristan | maybe this is new | 13:11 |
| tristan | Anyone recognize this crash when fetching a tarball https://bpaste.net/show/40b77c5cbefc ? | 13:11 |
| tlater[m] | qinusty, tristan: I'm going through all the edge cases we covered here, I'll add a write-up to that MR. It's an annoying one we keep forgetting. | 13:12 |
| * tlater[m] finds it tough to think through, frankly | 13:12 | |
| *** alatiera_ has joined #buildstream | 13:14 | |
| *** alatiera__ has quit IRC | 13:15 | |
| tlater[m] | We do have tests that confirm this is working though, so the patch itself looks to be fine. | 13:16 |
| tristan | tlater[m], Thanks, it is indeed. | 13:16 |
| tristan | I will have to fix this crash, for some reason, the source has fetch() being called, without a ref | 13:17 |
| tristan | mirrors breakage | 13:18 |
| tristan | jonathanmaw_, https://gitlab.gnome.org/GNOME/gnome-build-meta/blob/master/project.conf#L64 That looks alright to me right ? | 13:19 |
| tristan | jonathanmaw_, Maybe we are missing tests for declared partial mirrors, where not all aliases are redefined in a mirror ? | 13:19 |
| tristan | nice, it's perfectly reproducible = fixable | 13:21 |
| *** jonathanmaw_ is now known as jonathanmaw | 13:23 | |
| jonathanmaw | tristan: yep, that looks like the right format for mirrors. And yeah, I think the tests might be missing the case of mirrors not covering all aliases | 13:23 |
| tristan | jonathanmaw, interestingly `bst fetch sdk/brotli.bst` reproduces this crash from gnome-build-meta perfectly for me | 13:24 |
| tristan | jonathanmaw, I have to run now... would you care to take a look at this crasher ? | 13:24 |
| * tristan needs to delay the release for this | 13:24 | |
| tristan | What I have noticed, is in Source._fetch(), after instantiating the new source | 13:25 |
| tristan | I've added assert source._update_state() and assert source._get_consistency() != Consistency.INCONSISTENT, the assertion raises | 13:25 |
| *** alatiera__ has joined #buildstream | 13:25 | |
| tristan | So, the newly instantiated source does not have it's ref set | 13:25 |
| tristan | Aha, and there lies the problem ! | 13:26 |
| tristan | jonathanmaw, I see... the source in fact does not have the ref set | 13:26 |
| tristan | jonathanmaw, And bst fetch is crashing instead of noticing that it must track | 13:26 |
| tristan | Even in a build, it's not noticing | 13:27 |
| *** alatiera_ has quit IRC | 13:27 | |
| tristan | Weird, ok I have something to go on if it's not magically fixed by tomorrow | 13:27 |
| tristan | gnome-build-meta uses project.refs, but my old project.refs has core-deps/brolti.bst there, so it should have required a track (it seems to have been renamed to sdk/brolti.bst) | 13:29 |
| tristan | jonathanmaw, I no longer suspect that it is a source mirroring issue, by the way :) | 13:29 |
| tristan | Of course, I'd be delighted if I found it fixed behind my back :) | 13:30 |
| *** alatiera_ has joined #buildstream | 13:33 | |
| *** tristan has quit IRC | 13:33 | |
| *** alatiera__ has quit IRC | 13:34 | |
| paulsher1ood | please can we change this "conclave" name? it's a clear mistake | 13:41 |
| qinusty | tlater[m] do you know what a sources cache key is supposed to be in the log? | 13:41 |
| qinusty | Is it the task id (it's elements id)? | 13:41 |
| qinusty | +1 paulsher1ood | 13:41 |
| *** finn_ has quit IRC | 13:43 | |
| *** phildawson has quit IRC | 13:43 | |
| *** phildawson has joined #buildstream | 13:46 | |
| tlater[m] | qinusty: Sorry, not off the top of my head :| | 13:47 |
| tlater[m] | I did finish my little write-up: https://gitlab.com/BuildStream/buildstream/merge_requests/812#note_103125727 | 13:47 |
| * tlater[m] hopes that explains it well enough, and goes back to double-checking references in a report. | 13:47 | |
| *** finn_ has joined #buildstream | 13:48 | |
| jjardon | tlater[m]: did you have a chance to look to the CAS not emptying the cache? we got the server filled up again today :( https://gitlab.gnome.org/GNOME/gnome-build-meta/issues/48 | 13:51 |
| *** finn_ is now known as finn | 13:53 | |
| toscalix | paulsher1ood: changed https://wiki.gnome.org/Projects/BuildStream/Events#BuildStream_Gathering_2018 | 13:53 |
| paulsher1ood | toscalix: i think conclave in spanish is a "false friend" | 13:53 |
| tlater[m] | jjardon: Yes, I've not managed to fill a server up and break it yet. I have suspicions, but I can't really fix anything without testing it - I hope to get a bit more done this weekend though. | 13:54 |
| paulsher1ood | toscalix: tvm | 13:54 |
| jjardon | tlater[m]: I can give you access to the GNOME server if that helps | 13:55 |
| toscalix | it is a latin word so it has a common meaning based on the root and then in each language might have additional meanings. But I do not think is worth arguing about the name. I added gathering | 13:55 |
| *** finn_ has joined #buildstream | 13:55 | |
| paulsher1ood | :-) | 13:55 |
| *** finn has quit IRC | 13:56 | |
| * jjardon now wants to play magic the gathering | 13:56 | |
| tlater[m] | jjardon: Appreciated, but I don't think it will help much - I'll want to run out of disk space every time I test, waiting to fill up a big server will take too long ;p | 13:56 |
| jjardon | tlater[m]: we fill up it in one day yesterday. But sure I understand :) | 13:57 |
| *** SotK has left #buildstream | 13:58 | |
| *** SotK has joined #buildstream | 13:58 | |
| jjardon | tlater[m]: btw, if you want to create a big file quickly: fallocate -l 10G 10Gigfile | 13:58 |
| tlater[m] | jjardon: Yeah, the problem is that that sort of "filling up" doesn't actually cause the exception | 14:00 |
| tlater[m] | I suspect it's lack of thinking about concurrency, so instead I'm trying mulitple pushes at the same time | 14:00 |
| * tlater[m] does wonder if implementing something that allows just pushing files manually might help | 14:01 | |
| *** alatiera__ has joined #buildstream | 14:01 | |
| persia | tlater[m]: Consider creating a small-ish file, and using that as a loopback mount for the cache directory. | 14:01 |
| tlater[m] | persia: Yep, I'm using an ext4-formatted file as a docker volume | 14:02 |
| tlater[m] | I've also tested with randomly-generated files in case file holes cause issues, etc. | 14:02 |
| persia | tlater[m]: does that permit the file to grow, or is the "server" limited to only 10MiB or so? | 14:02 |
| *** alatiera_ has quit IRC | 14:03 | |
| tlater[m] | persia: "server" is limited to 3G (because we have a minimum size of 2G) | 14:03 |
| persia | Hrm, yeah, filling 3G ends up taking a while, assuming you want real content. | 14:04 |
| WSalmon | toscalix, i have had a clash with some one when editing the wiki, i think i have fixed it but could you check you are happy, apologies if i have missed/messedup something | 14:07 |
| tlater[m] | persia: It will probably be helpful to modify the server a little to make testing easier, have some additional debugging output, etc., but I've only had ~10 hours of time on this so far spread accross a few days... So progress is slow. | 14:08 |
| tlater[m] | I also think I might need multiple BuildStream instances pushing concurrently, which I haven't tested yet. | 14:09 |
| tlater[m] | But anyway, I don't really have time to do much more today - I promise to shout out if I can fix it :) | 14:10 |
| qinusty | tlater[m], before I forget. I'd also like to see the configurable warnings logic moved out of the plugin messaging API and into context | 14:11 |
| qinusty | Which will allow us to produce fatal warnings from outside of plugin logic | 14:11 |
| tlater[m] | qinusty: Is that commented on the relevant MR? | 14:12 |
| * qinusty checks | 14:12 | |
| tlater[m] | I'll forget if it's just here ;p | 14:12 |
| qinusty | Added a comment :D | 14:14 |
| persia | tlater[m]: Makes sense. Good luck. | 14:15 |
| qinusty | Apologies for dumping this onto you tlater[m], I carried the flame as far as I could | 14:15 |
| tlater[m] | qinusty: No problem at all, I mean, you sort of took the flame from me and ran with it. | 14:17 |
| tlater[m] | Hehe | 14:17 |
| qinusty | I guess :D | 14:17 |
| qinusty | I'm not sure what you want to do about those commits in the log. There is two commits from you which I didn't spend a load of time getting to work under the massive rebase. They're probably best being flattened on merge | 14:18 |
| *** alatiera_ has joined #buildstream | 14:19 | |
| *** lachlan has quit IRC | 14:20 | |
| *** alatiera__ has quit IRC | 14:20 | |
| *** mohan43u has quit IRC | 14:30 | |
| *** alatiera__ has joined #buildstream | 14:32 | |
| *** alatiera_ has quit IRC | 14:34 | |
| *** lachlan has joined #buildstream | 14:36 | |
| *** mohan43u has joined #buildstream | 14:44 | |
| *** abderrahim has quit IRC | 14:47 | |
| *** abderrahim has joined #buildstream | 14:48 | |
| toscalix | WSalmon: I erased the topics from the specific agenda because we ned to know first the room availability and people's availability. Your suggestions remain in the proposed topics section. Thanks for adding them | 14:56 |
| *** mohan43u has quit IRC | 14:57 | |
| toscalix | I might use the BuildSTream calendar instead of the wiki for consolidating the schedule | 14:57 |
| toscalix | jjardon: is still master broken? | 14:59 |
| toscalix | so you cannot build freedesktop-sdk? | 14:59 |
| gitlab-br-bot | buildstream: merge request (willsalmon/outOfSourecBuild->master: Out of source builds) #776 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/776 | 15:01 |
| *** mohan43u has joined #buildstream | 15:01 | |
| jjardon | toscalix: yes | 15:01 |
| jjardon | https://gitlab.com/BuildStream/buildstream/issues/664 (note that bst-1.2 works fine) | 15:03 |
| toscalix | I label the bug as critical and assigned it to tristan. We need to fix it right away, if we can | 15:03 |
| toscalix | valentind: around? able to help? | 15:03 |
| toscalix | we should put as test to build freedesktop-sdk | 15:04 |
| toscalix | being our main source if users reports, we should take it as validation project | 15:04 |
| toscalix | until we have a better one | 15:04 |
| toscalix | it seems buildgrid might be another option in the near future | 15:04 |
| toscalix | if you commit breaks freedesktop-sdk.... it is no worth reviewing | 15:05 |
| toscalix | s/validation/pre-integration | 15:06 |
| toscalix | tiagogomes_: around? able to help? | 15:12 |
| toscalix | The idea should be to revert the change that caused the problem, de-escalate the severity and then tomorrow work out what happened | 15:12 |
| toscalix | but leave today freedesktop-sdk working before we go home | 15:13 |
| toscalix | jjardon: are you able to identify the commit that has caused the issue? | 15:14 |
| *** alatiera_ has joined #buildstream | 15:29 | |
| *** alatiera__ has quit IRC | 15:31 | |
| gitlab-br-bot | buildstream: merge request (Qinusty/235-manifest->master: WIP: Implement generated build manifests) #596 changed state ("closed"): https://gitlab.com/BuildStream/buildstream/merge_requests/596 | 15:40 |
| tpollard | jjardon: do you know if gnome-build-meta is broken against bst master? | 15:49 |
| gitlab-br-bot | buildstream: merge request (willsalmon/outOfSourecBuild->master: Out of source builds) #776 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/776 | 15:49 |
| *** alatiera__ has joined #buildstream | 15:54 | |
| gitlab-br-bot | buildstream: merge request (jmac/remote_exec_checkout_fix->master: Remote exec: Remove early warning and check directory is not None) #800 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/800 | 15:54 |
| tpollard | ugghh | 15:55 |
| *** alatiera_ has quit IRC | 15:56 | |
| *** johanneshilling has quit IRC | 16:01 | |
| *** qinusty has quit IRC | 16:01 | |
| *** jennis has quit IRC | 16:01 | |
| *** coldtom has quit IRC | 16:01 | |
| *** adds68 has quit IRC | 16:01 | |
| *** tiagogomes_ has quit IRC | 16:01 | |
| *** qinusty has joined #buildstream | 16:02 | |
| *** iker has quit IRC | 16:02 | |
| *** johanneshilling has joined #buildstream | 16:05 | |
| *** jennis has joined #buildstream | 16:07 | |
| *** adds68 has joined #buildstream | 16:13 | |
| *** lachlan has quit IRC | 16:13 | |
| *** lachlan has joined #buildstream | 16:19 | |
| *** lachlan has quit IRC | 16:19 | |
| *** lachlan has joined #buildstream | 16:20 | |
| jjardon | tpollard: no idea | 16:23 |
| jjardon | tpollard: well probably yes, as gnome-build-meta uses freedesktop-sdk | 16:23 |
| jjardon | toscalix: as I said, freedesktop-sdk is working fine; we use a stable release of buildstream to build. It's only the daily job to build with bst master the one that is failing | 16:24 |
| *** alatiera_ has joined #buildstream | 16:46 | |
| *** alatiera__ has quit IRC | 16:48 | |
| *** alatiera__ has joined #buildstream | 16:48 | |
| *** alatiera_ has quit IRC | 16:50 | |
| *** alatiera_ has joined #buildstream | 16:50 | |
| *** lachlan has quit IRC | 16:51 | |
| *** alatiera__ has quit IRC | 16:52 | |
| *** finn_ has quit IRC | 16:57 | |
| juergbi | jjardon: do you have a last known working pipeline for a buildstream master build? it seems it was already broken yesterday (d8450166) but I can't easily find older pipelines for buildstream master | 16:58 |
| *** xjuan has joined #buildstream | 17:03 | |
| jjardon | juergbi: no , sorry; we setup the job to build freedesktop-sdk with bst master yesterday | 17:03 |
| juergbi | ah ok | 17:04 |
| *** lachlan has joined #buildstream | 17:07 | |
| *** tristan has joined #buildstream | 17:08 | |
| gitlab-br-bot | buildstream: issue #665 ("Buildstream tries to use message without a message handler if linux platform and no user namespace support") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/665 | 17:15 |
| *** lachlan has quit IRC | 17:16 | |
| *** lachlan has joined #buildstream | 17:23 | |
| *** jonathanmaw has quit IRC | 17:26 | |
| *** toscalix has quit IRC | 17:26 | |
| *** tristan has quit IRC | 17:30 | |
| *** alatiera__ has joined #buildstream | 17:31 | |
| *** alatiera_ has quit IRC | 17:33 | |
| *** alatiera_ has joined #buildstream | 17:34 | |
| *** lachlan has quit IRC | 17:35 | |
| *** alatiera__ has quit IRC | 17:35 | |
| *** alatiera__ has joined #buildstream | 17:36 | |
| *** alatiera_ has quit IRC | 17:37 | |
| gitlab-br-bot | buildstream: merge request (jmac/vdir_import_test->master: Import test for virtual directories) #815 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/815 | 17:39 |
| *** alatiera_ has joined #buildstream | 17:39 | |
| *** alatiera__ has quit IRC | 17:39 | |
| *** tristan has joined #buildstream | 17:44 | |
| *** finn has joined #buildstream | 18:14 | |
| *** tristan has quit IRC | 18:18 | |
| *** tristan has joined #buildstream | 19:23 | |
| *** coldtom has joined #buildstream | 20:20 | |
| valentind | jjardon, seems like the flatpak_image plugin does not work with buildstream master. Is that the problem? | 20:43 |
| *** tristan has quit IRC | 21:12 | |
| *** cs-shadow has quit IRC | 21:32 | |
| *** mohan43u has quit IRC | 22:29 | |
| *** mohan43u has joined #buildstream | 22:30 | |
| *** mohan43u has quit IRC | 22:33 | |
| *** mohan43u has joined #buildstream | 22:36 | |
| *** mohan43u has quit IRC | 22:43 | |
| *** mohan43u has joined #buildstream | 23:01 | |
| *** mohan43u has quit IRC | 23:19 | |
| *** mohan43u has joined #buildstream | 23:25 | |
| *** hergertme_ has joined #buildstream | 23:33 | |
| *** hergertme has quit IRC | 23:35 | |
| *** hergertme_ is now known as hergertme | 23:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
