Open edsantiago opened 1 year ago
Error: COPY --from=alpine: no stage or image found with that name
...and this is the Dockerfile:
Maybe that should be
from=base
, or maybe line 4 should beFROM alpine AS somethingelse
? I'm not wise enough to know. @nalind this is your code from June 2021, I'll leave it in your hands.
Nope, that's testing that COPY --from=... can refer to an image by its name, and "alpine" is the name of an image that we definitely know we have at that point, since it's the base image for that stage. If that step is failing, then we broke something.
And, I wonder if we should install qemu in our test VMs?
It shouldn't break anything if we do.
Nope, that's testing that COPY --from=... can refer to an image by its name, and "alpine" is the name of an image that we definitely know we have at that point, since it's the base image for that stage. If that step is failing, then we broke something.
This should be reproduce-able outside of multiple platform test as well, let me try reproducing it on a simpler test.
huh i was wrong this bug does not persists on regular multi-stage build
FROM alpine as one
RUN touch hey
RUN echo hello
FROM alpine as two
COPY --from=one hey .
COPY --from=alpine /tmp .
./buildah build -t test .
[1/2] STEP 1/3: FROM alpine AS one
[1/2] STEP 2/3: RUN touch hey
[1/2] STEP 3/3: RUN echo hello
hello
[2/2] STEP 1/3: FROM alpine AS two
[2/2] STEP 2/3: COPY --from=one hey .
[2/2] STEP 3/3: COPY --from=alpine /tmp .
[2/2] COMMIT test
Getting image source signatures
Copying blob 994393dc58e7 skipped: already exists
Copying blob 3cdd3495f0b5 done
Copying config 2134847aca done
Writing manifest to image destination
Storing signatures
--> 2134847acaa
Successfully tagged localhost:5000/test:latest
2134847acaa4235fbba828f2785b6285699819113fe5657dd16ce00523ec5dba
Is this the one that you and I (mostly you) fixed today?
@rhatdan I think this is different unrelated to that.
A friendly reminder that this issue had no activity for 30 days.
@flouthoc Any chance you can look at this?
A friendly reminder that this issue had no activity for 30 days.
Had no success reproducing this last time on regular mult-stage builds will try again for multiple architectures with emulation enabled.
A friendly reminder that this issue had no activity for 30 days.
I don't understand the failure to reproduce. It still reproduces 100% on buildah @ main:
$ env BUILDAH_BINARY=$(pwd)/bin/buildah bats --filter "bud-multiple-platform-values" tests/bud.bats
....
Error: creating build container: locating pulled image "docker.io/library/alpine:latest" name in containers storage: docker.io/library/alpine:latest: image not known
This is on my f37 laptop, and obviously with qemu installed because the test is skipped otherwise.
@edsantiago That was bad explanation on my part here: https://github.com/containers/buildah/issues/4396#issuecomment-1302240116, i assumed that this should be reproduceable even with regular multi-stage build but this only persists on a multi-stage builds when built for multiple platforms.
Filed #4596 to hard-skip this test, because it is triggering in Fedora gating tests. If/when anyone fixes this, please remove the skip
.
A friendly reminder that this issue had no activity for 30 days.
A friendly reminder that this issue had no activity for 30 days.
A friendly reminder that this issue had no activity for 30 days.
Two tests are not run in CI because the CI VMs don't have emulation tools:
Both are broken. I can fix the "-failure" one (it's just a wrong expected exit code, 125 when it should be 1) but I don't know the intention of the "-values" one and can't fix it. This is the failure:
...and this is the Dockerfile: https://github.com/containers/buildah/blob/3cfa17a69564a7a3968a5b4671f6c203c9ec1eda/tests/bud/multiarch/Dockerfile#L8
Maybe that should be
from=base
, or maybe line 4 should beFROM alpine AS somethingelse
? I'm not wise enough to know. @nalind this is your code from June 2021, I'll leave it in your hands.And, I wonder if we should install qemu in our test VMs?