| juergbi | nanonyme: the only possible downside I can think of is that single-threaded buildbox-fuse may be blocked during a large copy_file_range. I don't know whether that would be a practical issue or not | 05:41 |
|---|---|---|
| juergbi | oh and it wouldn't be compatible with the overlayfs support, I suppose | 05:43 |
| juergbi | however, we would also only benefit if there are actions that actually use it. iirc, ostree doesn't currently use reflinks for the operation we were looking at | 05:45 |
| juergbi | it would probably make more sense to first implement reflink support in buildbox-casd for file capture | 05:46 |
| abderrahim[m] | Indeed | 06:47 |
| nanonyme | juergbi: right. For checkout reflinks seems right approach but maybe overlayfs is the right solution for sandboxes indeed | 10:18 |
| nanonyme | Remind me, how new kernel do we need to test your branch? | 10:18 |
| juergbi | nanonyme: 5.11 | 10:19 |
| nanonyme | We hit a roadblock with backports kernel. There's in 6.0 a regression that such that disk resizing makes machine unbootable. | 10:20 |
| nanonyme | Let's see if we can get 5.11 or newer somehow though | 10:20 |
| nanonyme | juergbi: I noticed though in https://gitlab.com/BuildGrid/buildbox/buildbox-run-bubblewrap/-/commit/114a1c84382ca6eecd00296cb9f160a3504f9baa that the build failed. Isn't that a problem? | 13:53 |
| juergbi | nanonyme: it depends on the corresponding buildbox-common branch | 14:14 |
| nanonyme | I see | 14:16 |
| nanonyme | juergbi: so if we combine both, we should get Docker image sufficient for testing? | 14:16 |
| juergbi | nanonyme: yes | 14:17 |
| nanonyme | https://gitlab.com/freedesktop-sdk/infrastructure/freedesktop-sdk-docker-images/-/merge_requests/381 okay, let's bake some images | 15:24 |
| nanonyme | Excellent, looks like we now have an image to play with. Now all we need is newer kernel for x86_64 builders to test it out | 19:15 |
| nanonyme | juergbi: I'm now locally testing your branches and it's complaining I don't have sh | 20:21 |
| nanonyme | Either I'm doing something wrong or reads aren't properly propagating to FUSE | 20:23 |
| juergbi | nanonyme: maybe the overlayfs mount is not successful and it's missing an error path? I didn't test it extensively but it passed the buildstream test suite with just a couple of exceptions, iirc | 20:45 |
| nanonyme | Could be. Should I be seeing logs anywhere? | 20:45 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!