| *** cs-shadow has quit IRC | 01:00 | |
| *** tristan has quit IRC | 04:26 | |
| gitlab-br-bot | juergbi opened MR !1606 (juerg/artifactshare->master: tests/testutils/artifactshare.py: Add SIGTERM handler to subprocess) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1606 | 04:55 |
|---|---|---|
| *** narispo has quit IRC | 06:36 | |
| *** narispo has joined #buildstream | 06:36 | |
| *** tristan has joined #buildstream | 06:49 | |
| *** narispo has quit IRC | 07:48 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1606 (juerg/artifactshare->master: tests/testutils/artifactshare.py: Add SIGTERM handler to subprocess) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1606 | 07:48 |
| *** narispo has joined #buildstream | 07:50 | |
| benschubert | juergbi: I've tried building llvm (with clang + libc++) using BuildStream, with a remote artifact server on the same machine. It failes at pusghin ls | 07:56 |
| benschubert | *it fials at pushing the artifact because it goes OOM | 07:56 |
| benschubert | I have 32G of RAM and 32G of swap. The artifact cache local after the build is roughly 10Go | 07:57 |
| benschubert | is that expected? That we would require so much memory | 07:57 |
| benschubert | (and I can see the memory usage increase until it gets killed) | 07:58 |
| juergbi | oh, no, certainly not expected | 08:00 |
| juergbi | benschubert: is the excessive memory usage in BuildStream or buildbox-casd? | 08:00 |
| benschubert | haven't looked more in details, it happens when doing a 'push' (the build went on fine) | 08:01 |
| benschubert | I can have a look tonight maybe | 08:01 |
| juergbi | wondering whether the issue is something like a single huge blob or just the total byte size of an artifact | 08:02 |
| juergbi | benschubert: did you see which process got OOM'd? i.e., python/bst or casd? that should be in the system log | 08:02 |
| gitlab-br-bot | marge-bot123 merged MR !1606 (juerg/artifactshare->master: tests/testutils/artifactshare.py: Add SIGTERM handler to subprocess) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1606 | 08:03 |
| benschubert | juergbi: I'll have to look at it tonight, it's on my laptop at home :) | 08:04 |
| juergbi | ah, I see, ta | 08:05 |
| juergbi | there might still be a lot of optimization potential in casd | 08:06 |
| benschubert | Ok, I'll be reporting those problems then :) | 08:14 |
| *** ahmed89 has joined #buildstream | 08:19 | |
| coldtom | can someone take a look at https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/34 please? | 08:21 |
| *** ahmed89 has quit IRC | 08:24 | |
| *** ahmed89_ has joined #buildstream | 08:25 | |
| *** ahmed89_ is now known as ahmed89 | 08:25 | |
| benschubert | coldtom: I'll have a look at it today | 08:26 |
| coldtom | ty benschubert | 08:26 |
| *** rdale has joined #buildstream | 08:36 | |
| benschubert | coldtom: just two things ot double check and then LGTM! | 08:38 |
| *** ahmed89 has quit IRC | 08:53 | |
| *** ahmed89 has joined #buildstream | 08:54 | |
| *** jonathanmaw has joined #buildstream | 08:59 | |
| *** ahmed89 has quit IRC | 09:00 | |
| *** traveltissues has joined #buildstream | 09:14 | |
| benschubert | coldtom: lgtm then, thanks :) | 09:23 |
| coldtom | thanks benschubert :) | 09:24 |
| coldtom | could someone with maintainer superpowers hit merge then please :) | 09:24 |
| benschubert | done | 09:29 |
| coldtom | ty | 09:36 |
| *** bochecha has joined #buildstream | 09:44 | |
| *** ahmed89 has joined #buildstream | 09:48 | |
| *** ahmed89 has quit IRC | 10:08 | |
| *** ahmed89 has joined #buildstream | 10:09 | |
| benschubert | Question about the sandbox: I get a '/dev/shm' not found. Does someone sees a reason we wouldn't be mounting a /dev/shm as shared memory? | 10:14 |
| benschubert | *as tmpfs | 10:14 |
| juergbi | benschubert: I don't see a real issue pointing /dev/shm to a new tmpfs (could consider limiting its size, though). we should not pass through host /dev/shm, though | 10:17 |
| benschubert | ah wait, we already seem to have one in the base sandbox | 10:18 |
| benschubert | yeah, the host would defeat slightly our purpose :) | 10:18 |
| benschubert | juergbi: oh, in the sandbox, /dev/shm, is 755, outside it is 777. I guess bubblewrap is doing something wrong here? | 10:28 |
| juergbi | benschubert: I can confirm it's 755, however, it's owned by the uid the sandbox command is running as | 10:31 |
| juergbi | so I wouldn't expect any permission issues | 10:31 |
| benschubert | I'm trying to run bazel inside BuildStream, and it tries to run its own sandbox inside, remounting /dev/shm rw, since it doesn't see it like that x') | 10:32 |
| benschubert | guess I'll try disabling the sandbox of bazel | 10:32 |
| juergbi | I don't recall the details of bazel's sandbox | 10:33 |
| juergbi | you could probably also simply chmod a+rw /dev/shm inside the sandbox before running bazel, at least as a test | 10:34 |
| benschubert | it is bind-mounting a bunch of stuff | 10:34 |
| benschubert | right, I'll try this | 10:34 |
| *** narispo has quit IRC | 11:05 | |
| *** narispo has joined #buildstream | 11:06 | |
| *** tristan has quit IRC | 12:06 | |
| *** narispo has quit IRC | 12:25 | |
| *** ikerperez is now known as Iker | 13:23 | |
| tpollard | I'm having trouble running the sourcetests standalone | 13:54 |
| tpollard | Running the example command in tox.ini https://hastebin.com/ecagamerit.md | 13:55 |
| tpollard | they'll run when doing a full run of all tests though | 13:58 |
| coldtom | related to this i think tpollard https://gitlab.com/BuildStream/buildstream/issues/1121 | 13:59 |
| tpollard | cheers coldtom | 14:00 |
| tpollard | I'll add a comment | 14:00 |
| coldtom | at least, marking as usedevelop has fixed that error whenever that has happened to me | 14:01 |
| jjardon | benschubert: hey, not sure you noticed but you need to update the format-version on the fdsdk branch you are using for the nightly tests: https://gitlab.com/BuildStream/buildstream/-/jobs/296562613 | 15:13 |
| benschubert | jjardon: I saw that and my branch is fixed but not pushed :) thanks! | 15:14 |
| jjardon | benschubert: ah cool :) | 15:26 |
| *** bochecha has quit IRC | 15:30 | |
| gitlab-br-bot | traveltissues opened (was WIP) MR !1563 (traveltissues/985->master: workspaces via sourcecache) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1563 | 15:52 |
| traveltissues | !1563 isn't quite ready but it's at the stage where I'd like to open it to review | 15:57 |
| traveltissues | btw benschubert, i sent you a fix on bst-benchmarks | 15:59 |
| benschubert | traveltissues: let me have a look | 16:08 |
| benschubert | traveltissues: merged | 16:12 |
| traveltissues | ty | 16:12 |
| benschubert | I'll also aim to review your PR tomorrow | 16:12 |
| benschubert | I'm really looking forward to it :D | 16:12 |
| traveltissues | i've decided to split it into 2 since it was becoming sufficiently complicated | 16:15 |
| juergbi | traveltissues: based on a very brief look it seems the commit history is not cleaned up yet, i.e., some later commits fix up issues in earlier commits | 16:16 |
| juergbi | this won't allow proper per-commit review. or is it anyway necessary to squash all commits as this can't be done in incremental steps without breaking the test suite? | 16:17 |
| traveltissues | i will do some additional history cleanup juergbi | 16:19 |
| juergbi | ta | 16:19 |
| tlater[m] | Blergh, sorry for the double mail | 16:39 |
| tlater[m] | Internet hickup when I sent it | 16:40 |
| *** jonathanmaw has quit IRC | 17:10 | |
| *** narispo has joined #buildstream | 17:12 | |
| *** traveltissues has quit IRC | 17:19 | |
| *** narispo has quit IRC | 17:44 | |
| *** narispo has joined #buildstream | 17:44 | |
| *** rdale has quit IRC | 18:26 | |
| *** cs-shadow has joined #buildstream | 19:05 | |
| *** tristan has joined #buildstream | 22:35 | |
| *** narispo has quit IRC | 23:05 | |
| *** narispo has joined #buildstream | 23:05 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!