Add agent.base_image setting in system configuration file #788
Conversation
| description: > | ||
| "Elastic Agent image to be used for testing. Setting default will be used the | ||
| same Elastic Agent image as the stack." | ||
| type: string | ||
| enum: | ||
| - default | ||
| - complete | ||
| default: "default" |
There was a problem hiding this comment.
Complete would mean to use elastic-agent-complete image for docker.
Default the same image as the Elastic Agent from the stack.
| enum: | ||
| - docker | ||
| default: docker | ||
| type: |
There was a problem hiding this comment.
Nit. Maybe we can be more explicit on the name of this parameter. But I don't have a better idea for this, maybe base_image, or base_image_type?
| type: | |
| base_image: |
We have to keep in mind that this should be runtime-agnostic, but I think "image" is agnostic enough.
There was a problem hiding this comment.
I was thinking in that and I had the same doubts.
I was wondering what would mean if at some point it is supported as runtime VMs.
Would it be used docker image inside that VM too ?
There was a problem hiding this comment.
I was wondering what would mean if at some point it is supported as runtime VMs.
Would it be used docker image inside that VM too ?
I don't think so. Ideally we would install the agent directly in the VMs, or use an image that has the agent installed as a package or from tar/zip.
For the case of "complete" maybe we need to provide a different base image that includes the dependencies.
In that sense base_image would be fine and agnostic enough. It would not refer specifically to a docker image, but to a image in the general sense.
Another approach we could follow is to define this as dependencies or dependency sets:
depends: - synthetic_testing This indicates the runtime that the test depends on the agent running on a system with capability to run synthetic tests. This is resolved in the docker runtime by using the complete image, but could be resolved in other systems in different ways.
There was a problem hiding this comment.
In that sense base_image would be fine and agnostic enough. It would not refer specifically to a docker image, but to a image in the general sense.
Using base_image looks good for me.
It would be like defining which kind of elastic-agent artifact/tar/zip it is required for that system test in a given package.
Another approach we could follow is to define this as dependencies or dependency sets
This approach would allow developers to be more detailed to specify the requirements of each test. However, not sure if that could be too fine grained and therefore complicate the logic to select the required artifact/zip/docker.
There was a problem hiding this comment.
Another approach we could follow is to define this as dependencies or dependency sets
This approach would allow developers to be more detailed to specify the requirements of each test. However, not sure if that could be too fine grained and therefore complicate the logic to select the required artifact/zip/docker.
Agree, so far we have only had two kinds of images and I don't think we are going to have many more cases.
Co-authored-by: Jaime Soriano Pastor <jaime.soriano@elastic.co>
mrodm changed the title 
💚 Build Succeeded
History
cc @mrodm |
3 participants
What does this PR do?
Add a new setting into the system test configuration file to indicate the Elastic Agent base image to be used for testing.
Why is it important?
This new setting allows developers to set the required Elastic Agent base image for their packages.
Checklist
test/packagesthat prove my change is effective.spec/changelog.yml.Related issues