Docker release job: Intermittent failure

[danotorrey]

When attempting to release the Docker images for Graylog 4.3.5 (https://github.com/Graylog2/graylog-docker/commit/b326f09c00d9b3f246acddffe56e73f73e8fe189), the following error occurred for the 4.3.5-1 tag release job:

12:29:38  #45 0.523 standard_init_linux.go:228: exec user process caused: exec format error
12:29:38  #45 ERROR: process "/bin/sh -c install -d -o \"${GRAYLOG_UID}\" -g \"${GRAYLOG_GID}\" -m 0755 ${GRAYLOG_HOME}/plugin" did not complete successfully: exit code: 1
12:29:38  ------
12:29:38   > [linux/arm64 graylog-downloader 15/16] RUN install -d -o "1100" -g "1100" -m 0755 /usr/share/graylog/plugin:
12:29:38  #45 0.523 standard_init_linux.go:228: exec user process caused: exec format error

To retry, I bumped the image release version to 4.3.5-2 in https://github.com/Graylog2/graylog-docker/commit/adb69e039358288af5a017b779c3404855fa4156 and re-ran the release job. This time, it ran successfully.

From doing some research, it seems that the exec format error might be related to a platform/target mismatch, but that seems to not be the case here, since the retry worked.

@bernd Capturing this as an issue, to see if we can find the cause and avoid a recurrence.

[mpfz0r]

This happened on the arm64 part of the build. Could be that the cross compilation is having an issue. @bernd could we update the docker version on the jenkins node?

What is particularly weird is that the other install (#44) command worked just fine:

19:29:37  #44 [linux/arm64 graylog-downloader 14/16] RUN install -d -o "1100" -g "1100" -m 0755 /usr/share/graylog/plugins-merged
19:29:37  #44 DONE 0.7s
19:29:37  
19:29:37  #45 [linux/arm64 graylog-downloader 15/16] RUN install -d -o "1100" -g "1100" -m 0755 /usr/share/graylog/plugin
19:29:38  #45 0.523 standard_init_linux.go:228: exec user process caused: exec format error
19:29:38  #45 ERROR: process "/bin/sh -c install -d -o \"${GRAYLOG_UID}\" -g \"${GRAYLOG_GID}\" -m 0755 ${GRAYLOG_HOME}/plugin" did not complete successfully: exit code: 1
19:29:38  ------
19:29:38   > [linux/arm64 graylog-downloader 15/16] RUN install -d -o "1100" -g "1100" -m 0755 /usr/share/graylog/plugin:
19:29:38  #45 0.523 standard_init_linux.go:228: exec user process caused: exec format error
19:29:38  ------
19:29:38  Dockerfile:70
19:29:38  --------------------
19:29:38    68 |     RUN mv ${GRAYLOG_HOME}/plugin ${GRAYLOG_HOME}/plugins-default
19:29:38    69 |     RUN install -d -o "${GRAYLOG_UID}" -g "${GRAYLOG_GID}" -m 0755 ${GRAYLOG_HOME}/plugins-merged
19:29:38    70 | >>> RUN install -d -o "${GRAYLOG_UID}" -g "${GRAYLOG_GID}" -m 0755 ${GRAYLOG_HOME}/plugin
19:29:38    71 |     
19:29:38    72 |     COPY config ${GRAYLOG_HOME}/data/config

[bernd]

@mpfz0r The build workers run Docker version 20.10.17 which is the latest version, AFAIK.

[danotorrey]

From a chat with @bernd, since the error is intermittent, we can ignore it for now. I am OK with closing this if we don't plan to take any action on this now. We can always open it again if the issue happens again.