Bump buildx and docker vendoring#1012
Conversation
Apple Silicon you mean? |
|
|
||
| go 1.15 | ||
|
|
||
| // the distribution version from ecs plugin is quite old and it breaks containerd |
There was a problem hiding this comment.
You got me, I thought all were removed, but you just moved them below! :D
There was a problem hiding this comment.
Yeah, I tried to make it cleaner :)
There was a problem hiding this comment.
I bumped docker to v20.10.0-rc2 so this should now be a bit cleaner.
|
This is in draft because a lot was added to |
chris-crone
force-pushed
the
bump-buildx
branch
2 times, most recently
from
2997bad to
60bba5f
Compare
chris-crone
changed the title
chris-crone
force-pushed
the
bump-buildx
branch
from
60bba5f to
3c883d4
Compare
chris-crone
force-pushed
the
bump-buildx
branch
2 times, most recently
from
d091183 to
171403e
Compare
gtardif
left a comment
gtardif
left a comment
There was a problem hiding this comment.
Testing the removed warning, I don't see any output when docker build:
$ ./bin/docker compose build --workdir tests/compose-e2e/volume-test/
The new 'docker compose' command is currently experimental. To provide feedback or request new features please open issues at https://github.com/docker/compose-cli
[+] Building 0.0s (0/2)
$
Same with compose up, all the build output is masked (after rmi the images)
|
As a matter of fact, the build actually fails test added in #1039 |
|
Hmm.. I see a truncated output which is weird. I didn't change anything related to the stdout streaming so not sure what's causing this. Running a build multiple times results in a different output. This leads me to think it's either a race or we're not flushing before the process dies. |
chris-crone
force-pushed
the
bump-buildx
branch
from
171403e to
15b316c
Compare
gtardif
left a comment
gtardif
left a comment
There was a problem hiding this comment.
Don't see issues with log being truncated, seems to something else with your local env @chris-crone (apple M1 ??)
|
I see this on Intel as well so have moved to draft :( Worse it seems that the images aren't saved to the daemon: $ ./bin/docker compose up --workdir tests/compose-e2e/fixtures/volume-test/
The new 'docker compose' command is currently experimental. To provide feedback or request new features please open issues at https://github.com/docker/compose-cli
[+] Building 1.2s (6/6) FINISHED
=> [volume-test_nginx internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 32B 0.0s
=> [volume-test_nginx internal] load .dockerignore 0.0s
=> => transferring context: 2B 0.0s
=> [volume-test_nginx2 internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 32B 0.0s
=> [volume-test_nginx2 internal] load .dockerignore 0.0s
=> => transferring context: 2B 0.0s
=> [volume-test_nginx2 internal] load metadata for docker.io/library/nginx:latest 1.1s
=> CACHED [volume-test_nginx 1/1] FROM docker.io/library/nginx@sha256:6b1daa9462046581ac15be20277a7c75476283f969cb3a61c8725ec38d3b01c3 0.0s
[+] Running 0/2
⠋ volume-test_nginx_1 Creating 0.0s
⠋ volume-test_nginx2_1 Creating 0.0s
Error response from daemon: No such image: volume-test_nginx2:latest |
Signed-off-by: Chris Crone <christopher.crone@docker.com>
Signed-off-by: Chris Crone <christopher.crone@docker.com>
|
Found the issue, I have |
If a user has BUILDX_NO_DEFAULT_LOAD=1 set, then the images will not be stored in the image store unless we explicitly say that they must be. Signed-off-by: Chris Crone <christopher.crone@docker.com>
chris-crone
force-pushed
the
bump-buildx
branch
from
15b316c to
87153b9
Compare
|
@chris-crone did that explain the logs you didn't see as well ? |
For some of the logs yes. The rest can be explained by the missing export. |
3 participants
What I did
Bumped buildx in go.mod so that we can use a newer containerd. This removes the annoying error message on macOS arm64: