Refactor deployments storage on redeploy and undeployment

[ariefrahmansyah]

What this PR does / why we need it:

The terminated endpoint means the user has triggered Undeployment and there's no deployed endpoint in the cluster. When we redeploy this terminated endpoint, the deployment history should show all previous successful deployment logs as Not Deployed and the new deployment as Pending.

However, when redeploying the terminated endpoint, the deployment history shows the previous successful deployment as the current and deployed one:

Before redeployment of terminated endpoint:

Actual view from redeployment of terminated endpoint:

Fixes introduced by this PR:

Before redeployment, the current label is given to the latest successful/terminated endpoint:

During redeployment:

After redeployment succeed:

*Note: Please ignore the time displayed as in 7 hours -- this is because I'm running from local and the time zone is not translated correctly.

---

Behind the scene, we refactor deployments table for given version endpoint for redeployment and unemployment:

1. Redeployment
   a. new deployment status = running/serving,
   b. old successful deployment status = terminated
2. Undeployment
   a. old successful deployment status = terminated

Which issue(s) this PR fixes:

• Fixes redeployment of terminated endpoint
• Fixes duplicate link icon on Mlflow run

Does this PR introduce a user-facing change?:

NONE

Checklist

• Added unit test, integration, and/or e2e tests
• Tested locally
• Updated documentation
• Update Swagger spec if the PR introduce API changes
• Regenerated Golang and Python client if the PR introduce API changes

[ghost]

👇 Click on the image for a new way to code review

Legend

[krithika369]

Hi @ariefrahmansyah, I will have another look at this PR today. In the meantime, I was thinking it may also be good to capture the data migration steps (the SQL script and any ad-hoc action that needs to be taken for deployments since Merlin 0.34.0) in this MR as a .md doc or something like that, so it's easier for everyone to review. Could you help update that too? Thanks.

[ariefrahmansyah]

Hi @ariefrahmansyah, I will have another look at this PR today. In the meantime, I was thinking it may also be good to capture the data migration steps (the SQL script and any ad-hoc action that needs to be taken for deployments since Merlin 0.34.0) in this MR as a .md doc or something like that, so it's easier for everyone to review. Could you help update that too? Thanks.

Hi Krithika, I'm still playing on the SQL on my local machine. Will update the MR with the SQL doc. Btw, I'll create the doc with this path: docs/dev-guide/deployments_table_migration.md

[krithika369]

@ariefrahmansyah left some comments. As discussed earlier, do capture the data migration script in this PR as well. I think it should be safe to add it as a normal SQL migration file as we don't anticipate any ad-hoc handling for the deployments status update. Thanks!

[krithika369]

@ariefrahmansyah Left some more comments. One the whole, LGTM. Will have another look when the open comments are resolved, as some of them may require some thought. Thanks!

[krithika369]

Left one comment. The rest LGTM. Thanks, @ariefrahmansyah !