#So I m doing some funky stuff
1 messages Β· Page 1 of 1 (latest)
For reference: https://github.com/Azure/moby-packaging/blob/535c3ebc58e5cd62ba08ad7b7a4854e0b50645ef/package_test.go#L105-L135
go test -v --build-spec ./testdata/moby-runc.json --debug
got it π
I just ran this, will try yours π
go run make.go --build-spec=./moby-containerd.json
after writing that build-spec from the README.md
Yeah that just builds the package, go test will spin up the VM.
If I cancel my context and let everything run its course I wind up with: host alias: lookup E7F494QBNABDG on 10.87.0.1:53: no such host , which is the hostname of the vm container.
But I assume other things have been cancelled in the background and cleaned up (such as possibly the VM container).
If I don't specify the WithExposedPort(22, ...) then it errors out pretty quickly with the same sort of error.
That's weird.
filename too long π€
Definitely haven't seen that.
How'd you get that pretty output?
I'm running the new experimental Terminal User Interface π
If you've got the version v0.5.0 dagger CLI in path (can get from releases or brew etc), you can run it by prefacing your command with dagger run:
export _EXPERIMENTAL_DAGGER_TUI=true
dagger run go test -v --build-spec ./testdata/moby-runc.json --debug
I think my having the moby-containerd.json file around at the repo root was cause my particular outcomes. Removed and re-running
hmm...maybe not
Yeah I see in the TUI it looks like its having issues in different places, like fetching sources from github even.
Hit:1 http://ports.ubuntu.com/ubuntu-ports jammy InRelease
Hit:2 http://ports.ubuntu.com/ubuntu-ports jammy-updates InRelease
Hit:3 http://ports.ubuntu.com/ubuntu-ports jammy-backports InRelease
Hit:4 http://ports.ubuntu.com/ubuntu-ports jammy-security InRelease
Reading package lists...
Reading package lists...
Building dependency tree...
Reading state information...E: Unable to locate package linux-image-5.15.0-1031-kvm
E: Couldn't find any package by glob 'linux-image-5.15.0-1031-kvm'
E: Couldn't find any package by regex 'linux-image-5.15.0-1031-kvm'
E: Unable to locate package linux-modules-5.15.0-1031-kvm
E: Couldn't find any package by glob 'linux-modules-5.15.0-1031-kvm'
E: Couldn't find any package by regex 'linux-modules-5.15.0-1031-kvm'
Definitely does not fail for me.
That just sits and does nothing.
Sometimes fetching from github (using client.Git) just sits there forever.
I even added exit 1 at the beginning of the test_runner.sh script and still it just sits there.
@rain spade @mint merlin any ideas?
@hushed glade maybe we should get @silver grail onto the Dagger Cloud trial so he can better visualize whatβs going on π
taking a look at this now!
@silver grail hmm no hanging here, seeing a lot of logs from test_runner.sh:
ok 15 validate cli package version # skip no package version specified to compare against
timeout: failed to run command 'docker': No such file or directory
ok 16 validate containerd package version # skip no package version specified to compare against
timeout: failed to run command 'docker': No such file or directory
ok 17 validate buildx package version # skip no package version specified to compare against
timeout: failed to run command 'docker': No such file or directory
ok 18 validate compose package version # skip no package version specified to compare against
timeout: failed to run command 'docker': No such file or directory
not ok 19 compose plugin is installed in 0sec
timeout: failed to run command 'docker': No such file or directory
not ok 20 buildx plugin is installed in 0sec
/usr/local/bin/test_runner.sh: 21: let: not found
when it hangs for you, do the spinners stop animating?
They continue animating.
gotcha, seems like the UI isn't stuck at least. if you look in ps or htop do you see a running test_runner.sh process?
gonna try running with the latest stable dagger CLI instead of dev
No, only qemu is running.
oh interesting, it hangs in the same way if I use the latest stable version (engine v0.5.0)
maybe we fixed something in main? (running again for good measure)
yep, worked again
doing a bisect to track down what actually fixed it
apparently it's this commit?
https://github.com/dagger/dagger/commit/48fb31df3c383da743c9cba5723f4291e0e6f8df
Running this from master now.
At least it's showing that there's an error executing test_runner.sh but no details as to what it is.
Oh I still have my exit 1 at the top of the script 
So sweet, this seems to be working!
nice! yeah, not totally sure why that commit fixes it, but it does. the only thing I can think of is a seemingly unrelated fix to WithRootfs, which I see you use, but the rest of it is an additive feature π€·ββοΈ
Not even using that with the test case I'm running.
π€·ββοΈ fixing hidden π π