-
Notifications
You must be signed in to change notification settings - Fork 57
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add github workflow running jenkins job #255
base: master
Are you sure you want to change the base?
Conversation
57efde1
to
a9870c9
Compare
89801e4
to
fa8ad66
Compare
And do we need to run asymmetric-cluster test ? or it is better to have regular 3h longevity for sake of stability ? |
@roydahan Should the next steps be decided on SCT side e.g. which test to run and what should be the official location for those jenkins jobs? |
@sylwiaszunejko , please take a look at |
shouldn't it be named |
Done |
So it looks like we have an official location for a job, next thing would be to have a designated user for it, is there some generic one we can use or who could create such? |
7daedbc
to
45603e6
Compare
.github/workflows/extended-ci-longevity-large-partition-asymmetric-cluster-3h-test.yml
Outdated
Show resolved
Hide resolved
uses: sylwiaszunejko/jenkins_client@main | ||
with: | ||
jenkins_job_name: 'scylla-drivers/gocql/extended-ci/longevity-large-partition-asymmetric-cluster-3h-test' | ||
jenkins_job_parameters: '{"email_recipients": "[email protected]", "scylla_version": "6.1.1", "extra_environment_variables": "SCT_STRESS_IMAGE.scylla-bench=scylladb/gocql-extended-ci:scylla-bench-${{ github.event.pull_request.head.sha }}"}' |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm unsure if we should fix the scylla_version to something specific.
On one hand, it can make the CI more predictable (not sure if stable, but quite repeatable).
On the other hand, at some point we need to go back and update it and possibly face new issues.
BTW, there is an option to set here "scylla_version: 6.1" so it will take all scylla 6.1 patches.
Another option is to use an enterprise release like 2024.1 who is LTS and won't require us to update it very often (will automatically pick the latest patch release).
Regarding email_recipent, can we set it to the author of the PR (as long as it internal author?) - not a must.
Lastly, I noticed that the Jenkins job is set to master, which again brings back the question of stability vs being updated automatically.
@sylwiaszunejko / @dkropachev / @fruch - your thoughts?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm unsure if we should fix the scylla_version to something specific. On one hand, it can make the CI more predictable (not sure if stable, but quite repeatable). On the other hand, at some point we need to go back and update it and possibly face new issues.
having master
there is not an option.
We can do same thing we do for cpp-rust-driver
- get last release and run on it.
BTW, there is an option to set here "scylla_version: 6.1" so it will take all scylla 6.1 patches. Another option is to use an enterprise release like 2024.1 who is LTS and won't require us to update it very often (will automatically pick the latest patch release).
From one side using OSS will bring top feature to the test as fast as possible.
On other side using enterprise LTS will bring more stability.
It looks like we need both ))
Regarding email_recipent, can we set it to the author of the PR (as long as it internal author?) - not a must.
It is doable, but it will expose ip addresses, urls of our test infrastructure, if we are ok with it then sure, why not.
But we also should have [email protected] as CC.
Lastly, I noticed that the Jenkins job is set to master, which again brings back the question of stability vs being updated automatically.
Agree, it is better to target either branch-6.1
or branch-2023
or particular commit from master
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm unsure if we should fix the scylla_version to something specific. On one hand, it can make the CI more predictable (not sure if stable, but quite repeatable). On the other hand, at some point we need to go back and update it and possibly face new issues.
having
master
there is not an option.
We can do same thing we do forcpp-rust-driver
- get last release and run on it.BTW, there is an option to set here "scylla_version: 6.1" so it will take all scylla 6.1 patches. Another option is to use an enterprise release like 2024.1 who is LTS and won't require us to update it very often (will automatically pick the latest patch release).
From one side using OSS will bring top feature to the test as fast as possible.
On other side using enterprise LTS will bring more stability.
It looks like we need both ))Regarding email_recipent, can we set it to the author of the PR (as long as it internal author?) - not a must.
It is doable, but it will expose ip addresses, urls of our test infrastructure, if we are ok with it then sure, why not.
But we also should have [email protected] as CC.
Nothing secret in the emails
Also those should run only by trusted users (with some definition of trust), isn't it ?
Lastly, I noticed that the Jenkins job is set to master, which again brings back the question of stability vs being updated automatically.
Agree, it is better to target either
branch-6.1
orbranch-2023
or particular commit from master
release branch do get backports, on the other hand they might not have all of the fixes features for those test (not to mention we'll need to backport specific things unrelated to a release need)
A specific master commit, is also problematic, when one will update it ? what happens when it breaks and stops working ?
c021aaa
to
cdb6f0f
Compare
v2:
Things that still need to be addressed:
|
This workflow pushes scylla-bench docker image using gocql version from the PR and triggers jenkins job that uses this image of scylla-bench.
cdb6f0f
to
a3d9e25
Compare
https://github.com/scylladb/gocql/actions/runs/11157393428/job/31011588519?pr=263 I am not sure why the second workflow gets interrupted even though the timeout has not passed yet and the job was still running when it happened |
3f34405
to
e4da434
Compare
e4da434
to
af6c532
Compare
@dkropachev is https://github.com/scylladb-actions/jenkins-client an official location for this action? |
It is not, but it shouldn't matter since it is an GitHub action. |
I'm working on stabilizing another shorter test for CI (1h test that is based on this test). I'm not sure what you mean by "need to manually create new job". I also don't think this mechanism is necessary. IMO the CI should use 2024.1 as Scylla_version and the SCT branch is branch-2024.1. In addition, we can have another CI job if we think is necessary that points to "scylla_version: master:latest". |
This workflow pushes scylla-bench docker image using gocql version from the PR and triggers jenkins job that uses this image of scylla-bench.