self-hosted: update CI test to use new aws modular terraform structure #32665
Conversation
| @@ -12,32 +12,32 @@ provider "aws" { | |||
| } | |||
|
|
|||
There was a problem hiding this comment.
The CI is currently failing with:
2025-06-05T19:08:53.636032Z ERROR k8s_controller::controller: Materialize reconciliation error. err=reconciler for object Materialize.v1alpha1.materialize.cloud/12345678-1234-1234-1234-123456789012.materialize-environment failed source=failed to apply object: ApiError: "404 page not found\n": Failed to parse error data (ErrorResponse { status: "404 Not Found", message: "\"404 page not found\\n\"", reason: "Failed to parse error data", code: 404 }) source.sources=[ApiError: "404 page not found\n": Failed to parse error data (ErrorResponse { status: "404 Not Found", message: "\"404 page not found\\n\"", reason: "Failed to parse error data", code: 404 }), ApiError: "404 page not found\n": Failed to parse error data (ErrorResponse { status: "404 Not Found", message: "\"404 page not found\\n\"", reason: "Failed to parse error data", code: 404 }), "404 page not found\n": Failed to parse error data] controller="orchestratord.materialize.cloud/materialize"
Seems more like an issue with main rather than the TF module changes.
There was a problem hiding this comment.
Temporary test is green: https://buildkite.com/materialize/nightly/builds/12236
Upgrade test is green: https://buildkite.com/materialize/nightly/builds/12237
bobbyiliev
force-pushed
the
self-hosted-update-refactored-aws-module-tests
branch
from
9af232c to
c6fea29
Compare
| resource "random_password" "db_password" { | ||
| length = 32 | ||
| special = false | ||
| provider "kubernetes" { |
There was a problem hiding this comment.
@def- the new structure comes with many breaking changes, so we might should probably theardown the old aws persistent infra for now and then spin it up again using the new module structure.
Also the aws-persistent and the aws-temporary projects are identical except the name_prefix so should probably get rid of one of those and in the mzcompose.py file we could just pass the name_prefix as a -var. But we could do it in a follow up PR!
| git commit -m "Bump to operator $CI_HELM_CHART_VERSION, materialize $CI_MZ_VERSION" | ||
| git diff HEAD~ | ||
| run_if_not_dry git push origin main |
There was a problem hiding this comment.
This is just a copy of how we currently bump the terraform-helm-materialize mdouel, open to suggestions if we should do this the same way for the new refactored AWS module.
We could leave the helm chart version as null in the Terraform module so it would default to the latest helm version here, but we would still need the bump for the envd image version.
There was a problem hiding this comment.
We probably need a broader discussion about how we manage versions of these modules. That should probably take place verbally or in slack, rather than on this PR.
| git commit -m "Bump to operator $CI_HELM_CHART_VERSION, materialize $CI_MZ_VERSION" | ||
| git diff HEAD~ | ||
| run_if_not_dry git push origin main |
There was a problem hiding this comment.
We probably need a broader discussion about how we manage versions of these modules. That should probably take place verbally or in slack, rather than on this PR.
| @@ -346,6 +346,7 @@ def kubectl_setup( | |||
| "requests": {"cpu": "100m", "memory": "256Mi"}, | |||
| }, | |||
| "backendSecretName": "materialize-backend", | |||
| "authenticatorKind": "None", | |||
There was a problem hiding this comment.
Thank you for adding this. The password support code got reverted, and I'm including this in my next try.
Motivation
Depends on MaterializeInc/terraform-aws-materialize#69
Ready for a review but we should not merge it yet until MaterializeInc/terraform-aws-materialize#69 gets merged.