You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
While making the changes for #42 I spotted that make test does not pass. I don't know what the build pipeline for this repo looks like these days, so it's unclear how long this has not been working.
There were two1 general classes of error that I saw.
Incorrect SHA in Docker image
not ok 6 the image has a vcs-ref label set to the current head commit in github
# (in test file common/php.bats, line 40)
# `[ "$output" = `git rev-parse --short HEAD` ]' failed
# status: 0
# output: c9e3ee6
The original intent behind this test was to ensure that all images had been rebuilt on each push to github. Looking at the output for the PHP 5.6 image, the SHA in it is c9e3ee6 which looks to be the last commit before Travis stopped working. If all images are no longer being built, it would be worth removing this test.
Errors on Apple Silicon machines
Running make test on an Apple Silicon machine highlights a number of different errors, but all with a root cause being that there aren't ARM versions of all the images in dockerhub. I'm assuming this is because they aren't being automatically built any more as Travis is gone. Only some of the PHP versions have ARM images available. I guess these were manually built and pushed?
not ok 1 the image does not return any warnings
# (in test file common/php-arm64.bats, line 7)
# `[ "$status" -eq 0 ]' failed
# status: 125
# output: Unable to find image 'graze/php-alpine:5.6' locally 5.6: Pulling from graze/php-alpine ec18bdb80f36: Pulling fs layer 8cae0e1ac61c: Pulling fs layer 558bb2464c90: Pulling fs layer 8e8cc45adb89: Pulling fs layer 558bb2464c90: Already exists 8e8cc45adb89: Already exists ec18bdb80f36: Already exists 8cae0e1ac61c: Already exists Digest: sha256:a7a78eae570af5defe9d8d846e2d30bdcc0411366daad1fd293a06562b210910 Status: Downloaded newer image for graze/php-alpine:5.6 docker: Error response from daemon: No such image: graze/php-alpine:5.6. See 'docker run --help'.
This test case expects no warnings but of course there are some because there is no ARM image available.
not ok 12 php should have a memory limit of 1024M
# (in test file common/php.bats, line 89)
# `[ "$output" = "memory_limit => 1024M => 1024M" ]' failed
# status: 0
# output: WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requested
# memory_limit => 1024M => 1024M
There are a lot of test failures like this one where the expected output is there (eg. 1024M memory limit) but the test fails because of the extra warning text.
Footnotes
Well, there was a third but I fixed that as part of the above PR. ↩
The text was updated successfully, but these errors were encountered:
Hello,
While making the changes for #42 I spotted that
make test
does not pass. I don't know what the build pipeline for this repo looks like these days, so it's unclear how long this has not been working.There were two1 general classes of error that I saw.
Incorrect SHA in Docker image
The original intent behind this test was to ensure that all images had been rebuilt on each push to github. Looking at the output for the PHP 5.6 image, the SHA in it is c9e3ee6 which looks to be the last commit before Travis stopped working. If all images are no longer being built, it would be worth removing this test.
Errors on Apple Silicon machines
Running
make test
on an Apple Silicon machine highlights a number of different errors, but all with a root cause being that there aren't ARM versions of all the images in dockerhub. I'm assuming this is because they aren't being automatically built any more as Travis is gone. Only some of the PHP versions have ARM images available. I guess these were manually built and pushed?This test case expects no warnings but of course there are some because there is no ARM image available.
There are a lot of test failures like this one where the expected output is there (eg. 1024M memory limit) but the test fails because of the extra warning text.
Footnotes
Well, there was a third but I fixed that as part of the above PR. ↩
The text was updated successfully, but these errors were encountered: