Remove redundant testing and circular dependency from docker acceptance testing #18181
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🤖 GitHub commentsExpand to view the GitHub comments
Just comment with:
|
|
This pull request does not have a backport label. Could you fix it @donoghuc? 🙏
|
donoghuc
added
the
backport-active-all
kaisecheng
reviewed
Sep 23, 2025
There was a problem hiding this comment.
I am wondering if the env var LOCAL_ARTIFACTS and here could help, so we don't need to do string substitution with sed and awk. It is a bit hard to maintain.
If I recall correctly, the env var is used to build docker image with local artifact, but this needs to be verified.
yaauie
reviewed
Sep 23, 2025
docker/Makefile
Outdated
| tar -zxf ../../logstash-$(VERSION_TAG)-docker-build-context.tar.gz && \ | ||
| sed 's/artifacts/snapshots/g' Dockerfile > Dockerfile.tmp && mv Dockerfile.tmp Dockerfile && \ | ||
| cp ../../logstash-$(VERSION_TAG)-linux-$(ARCHITECTURE).tar.gz . && \ | ||
| awk '/# Add Logstash itself and set permissions/{print; print "COPY logstash-$(VERSION_TAG)-linux-$(ARCHITECTURE).tar.gz /tmp/logstash.tar.gz"; next}1' Dockerfile > Dockerfile.tmp && \ |
There was a problem hiding this comment.
This creates a hidden dependency on the precise wording of a comment from the dockerfile template. Is there any way to make this less fragile?
One way would be to add an ERB comment syntax to the dockerfile template that makes it obvious that the line's exact wording is depended upon. As ERB comments, they will not make it through to the generated Dockerfile:
diff --git a/docker/templates/Dockerfile.erb b/docker/templates/Dockerfile.erb index 355a76cff..d92c8bf1a 100644 --- a/docker/templates/Dockerfile.erb +++ b/docker/templates/Dockerfile.erb @@ -64,7 +64,7 @@ RUN \ <% end -%> # Provide a non-root user to run the process -# Add Logstash itself and set permissions +# Add Logstash itself and set permissions<%# HOOK: do not reword %> <% if image_flavor == 'full' || image_flavor == 'oss' -%> RUN groupadd --gid 1000 logstash && \ adduser --uid 1000 --gid 1000 \ @@ -84,9 +84,9 @@ RUN addgroup -g 1000 logstash && \ <% if image_flavor == 'full' || image_flavor == 'oss' -%> arch="$(rpm --query --queryformat='%{ARCH}' rpm)" && \ <% end -%> - curl --fail --location --output logstash.tar.gz <%= url_root %>/<%= tarball %> && \ - tar -zxf logstash.tar.gz -C /usr/share && \ - rm logstash.tar.gz && \ + curl --fail --location --output logstash.tar.gz <%= url_root %>/<%= tarball %> && \<%# HOOK: do not reword %> + tar -zxf logstash.tar.gz -C /usr/share && \<%# HOOK: do not reword %> + rm logstash.tar.gz && \<%# HOOK: do not reword %> mv /usr/share/logstash-<%= elastic_version %> /usr/share/logstash && \ chown -R logstash:root /usr/share/logstash && \ chmod -R g=u /usr/share/logstash && \
Another way would be to pull this logic up to the dockerfile template, and to generate a dockerfile-local "correctly" (instead of mutating the dockerfile-full). This seems like the long-term approach, but wiring it all the way through to whatever depends on it may be complex.
|
Thanks @kaisecheng and @yaauie for the reviews. The reason I did not pursue the local_artifacts route was we hard code those to Line 135 in 2fab5f4 I will track down how to unify our use of the |
donoghuc
changed the title
|
I've updated my PR description #18181 (comment) with detailed reasoning. I am open to filing another ticket for explicitly testing public dockerfiles outsie of exhaustive test pipeline, happy to talk through what that might look like :) |
|
kaisecheng
approved these changes
Oct 1, 2025
There was a problem hiding this comment.
Great to see awk, sed, and the redundant make tasks gone! 👍 Highly appreciate your effort in improving CI.
I tested locally for DRA, and it works as expected.
I have small suggestion to clean up the leftover in artifacts.rake
- If
build_docker_from_dockerfiles()is not used anymore, can we remove it ? - rename
build_docker()tobuild_docker_from_local_artifacts()and remove passing the env"LOCAL_ARTIFACTS" => LOCAL_ARTIFACTS, as it always builds from local
Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline.
Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing.
This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker.
donoghuc
force-pushed
the
use-local-artifact
branch
from
fe862d5 to
a219f32
Compare
|
Thanks @kaisecheng ! I applied your suggestions (and rebased on head of main). I removed the |
💚 Build Succeeded
History
|
kaisecheng
approved these changes
Oct 1, 2025
|
@Mergifyio backport 8.18 8.19 9.0 9.1 9.2 |
✅ Backports have been created
|
mergify bot
pushed a commit
that referenced
this pull request
Oct 1, 2025
…ce testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake
mergify bot
pushed a commit
that referenced
this pull request
Oct 1, 2025
…ce testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake
mergify bot
pushed a commit
that referenced
this pull request
Oct 1, 2025
…ce testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake
This was referenced Oct 1, 2025
mergify bot
pushed a commit
that referenced
this pull request
Oct 1, 2025
…ce testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c)
mergify bot
pushed a commit
that referenced
this pull request
Oct 1, 2025
…ce testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake
donoghuc
added a commit
that referenced
this pull request
Oct 6, 2025
…ncy from docker acceptance testing (#18253) * Remove redundant testing and circular dependency from docker acceptance testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake * resolve merge conflicts --------- Co-authored-by: Cas Donoghue <[email protected]>
donoghuc
added a commit
that referenced
this pull request
Oct 6, 2025
…ency from docker acceptance testing (#18251) * Remove redundant testing and circular dependency from docker acceptance testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake * fix merge conflicts --------- Co-authored-by: Cas Donoghue <[email protected]>
donoghuc
added a commit
that referenced
this pull request
Oct 6, 2025
…ency from docker acceptance testing (#18252) * Remove redundant testing and circular dependency from docker acceptance testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake * fix merge conflicts * fix merge conflict and update rake task name --------- Co-authored-by: Cas Donoghue <[email protected]>
donoghuc
added a commit
that referenced
this pull request
Oct 6, 2025
…ncy from docker acceptance testing (#18254) * Remove redundant testing and circular dependency from docker acceptance testing (#18181) * Use locally build artifact to build container from public dockerfile Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from `https://snapshots.elastic.co/downloads/logstash'` then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image to `snapshots`. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release. This commit modifies the dockerfile to use the artifact prepared in the first step instead of curling the last published one. This solves both issues as the tests run against the code from the tip fo the branch being tested and there is no dependency on an artifact existing as a result of a unified release pipeline. * Remove redundant docker steps from workflows Previously for docker acceptance tests three steps were performed: 1. Build container images (based on local artifacts) 2. Build "public" dockerfiles 3. Build container based on (a modified) file from step 2. The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency. This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to `false` is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to `true` when testing. * Remove unused function and argument This commit removes the unused function for building from dockerfiles. It also removes an unused argument for the make task for build_docker. (cherry picked from commit a994c7c) # Conflicts: # docker/Makefile # rakelib/artifacts.rake * fix merge conflicts --------- Co-authored-by: Cas Donoghue <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Release notes
[rn:skip]
What does this PR do?
Previously we would build an image (which would not actually be used), build dockerfiles, modify dockerfiles to curl from
https://snapshots.elastic.co/downloads/logstash'then build the image used for testing based on the modified dockerfile. This resulted in testing the last published image tosnapshots. This presents two problems 1. The test is running against the last published image (not the tip of the branch being tested) 2. this carries a dependency on both a DRA and unified stack release having been run. Therefor acceptance tests will fail in between the time we bump logstash version and a successful run of unified release.For docker acceptance tests three steps were performed:
The ONLY difference between the dockerfile that ultimately is used to define an image between 1 and 2 is WHERE the logstash source is downloaded from. In acceptance testing we WANT to use the source at the current checkout of logstash (not a remote). Using remote causes a dependency issue when changes are not published. Publishing is tied to unified release and gated on tests so naturally this is a bad fit for that dependency.
This commit removes the redundancy by ONLY generating images for testing (step 1 from above). This also firms up our use of LOCAL_ARTIFACTS. Namely, the ONLY time we want that set to false is when we build a "public" dockerfile. We explicitly set that in the corresponding DRA script now. Similarly we explicitly set it to true when testing.
Why is it important/What is the impact to the user?
n/a
Checklist
How to test this PR locally
You can run the test locally with:
Related issues