feat(ci): add update_server job for cluster maintenance in trunk workflow
#33770
Conversation
…rkflow ### Changes: - Introduced `update_server` job to initiate cluster shutdown with rolling restart. - Configured job to run after critical phases in the workflow. - Added `STAGING_TRUNK_API_TOKEN` for secured API request handling. ref: #33769
| SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }} | ||
|
|
||
| # Update Server job - triggers cluster shutdown with rolling restart | ||
| update_server: |
There was a problem hiding this comment.
This job should not really go at the end here. There are specific phases in these workflows that are common and finalize and report should always be at the end as they capture errors in any of the other jobs. This step should really be part of the "deployment" phase e.g. in ./.github/workflows/cicd_comp_deployment-phase.yml and go after the steps that do the push to dockerhub a flag can be set to allow it to be run or not passed in the deployment.with in this workflow but not passed from others. when moving it to ./.github/workflows/cicd_comp_deployment-phase.yml
It would be best to make this a little more general if putting into ./.github/workflows/cicd_comp_deployment-phase.yml e.g. by passing the domain we are calling as a parameter from this workflow. You may also need to double check how secrets are passed into these reusable workflows from other examples as they are not automatically visible.
Eventually we need to look in a more general way to indicate that a new image is available and preferably update the actual version that rather rely on the mutable tag, but that is for later and this is a good solution for now.
There was a problem hiding this comment.
If we put this at the end of the cicd_comp_deployment-phase.yml then if it fails everything else will still have completed, it could still fail the workflow. If we do not want a failure of calling the server from failing the workflow e.g. if the server is already down we can make the step always return success.
|
This PR is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 7 days. |
Changes:
update_serverjob to initiate cluster shutdown with rolling restart.STAGING_TRUNK_API_TOKENfor secured API request handling.ref: #33769