feat: support atlas migration, issueref #56

[shreyasHpandya]

No description provided.

[sunnyyip]

Thanks for the PR, @shreyasHpandya, @semmet95 @ANIRUDH-333!

May I ask you to help me documenting the new features?

1. We use readme-generator to generate docs for the inputs. Can we add the docs for "atlas" in values.yaml please?
   This is how I run readme-generator at helm-charts/charts/guac: readme-generator-for-helm/bin/readme-generator -v ./values.yaml -r ./README.md -s schema.json
2. Can you please share your learning on running atlas migration? Perhaps a little "Atlas Migration" section under "Install Chart" to call out the importance?
   E.g.

• do you keep atlas.enabled on or only during version upgrade?
• Is it ok to run multiple replicas of graphql-server while atlas.enabled is on?
• Order of operations for migration...etc.

[ANIRUDH-333]

Hi @sunnyyip . Thank you for sharing the instructions on how to update the PR.
We have followed it and updated the PR to include all the params in Readme file. Please review these changes and let us know if there is anything else that needs to be added/changed to the documentation.
We are still working on the second point you mentioned.

[semmet95]

Fixes: #56

[semmet95]

Hi @sunnyyip
We have added an Atlas Migration section as well.
For reference, Atlas was introduced in the guac repo here

• Since the atlas migrations are supposed to be idempotent, if atlas.enabled is set to true the init container runs each time the graphql-server is updated/restarted.
• We observed that even when running multiple replicas of graphql-server atlas migration operation doesn't cause any issues. We also tested this while upgrading Guac version from v0.11.2 to v0.12.3
• When using atlas we set guac.backend.ent.db-migrate to false, so atlas has to take care of schema migrations before the graphql-server container starts. This is why we are running the atlas image as an init container.

[sunnyyip]

Thanks @semmet95! Let's rebase to pick up the blobstore addr update to get pass the tests?
Also, please see my comment above about atlas migration's image.

[semmet95]

Hey @sunnyyip
Regarding the blobstore addr update, that change is already in our branch here
I was working on both these PRs in parallel, so I had the fix added to them both.

Regarding atlas migration's image, I'm not sure which specific point you are referring to. Please let us know if we missed adding some doc or explanation.

[sunnyyip]

@semmet95 please have a look at my comment at https://github.com/kusaridev/helm-charts/pull/57/files/62579f2f11f8d213307e6b9d0ade4c3adf5c475a

We should only be using the atlas image because the guac image (default/fallback right now) won't work for running atlas. This applies to both digest/image tag.

[semmet95]

Hey @sunnyyip
I can't see the comment you've shared. The link takes me to the PR diff page.
Do you need to submit that comment or is it some visibility issue?

[sunnyyip]

Hey @sunnyyip I can't see the comment you've shared. The link takes me to the PR diff page. Do you need to submit that comment or is it some visibility issue?

sorry about the link...I mentioned you in the comment. Let me know if you see it. Thanks

[semmet95]

Hi @sunnyyip
I still can't see the comment :(
Can you please tag @ANIRUDH-333 there as well?

[pxp928]

It could be @sunnyyip and @semmet95 that the comment is not published yet. @sunnyyip you can see it but he cannot because you have not submitted to review.

[sunnyyip]

Should the image always be .Values.atlas.image.repository and not have a default?
The guac image won't have the bins for running migration.

[sunnyyip]

@semmet95 this is what I was referring to about image repo for atlas

[sunnyyip]

It could be @sunnyyip and @semmet95 that the comment is not published yet. @sunnyyip you can see it but he cannot because you have not submitted to review.

Oops...just submitted the comments for review. It should be visible now. Thanks.