Skip to content

fix(acedatacloud): allow hosted MCP fleet to use OAuth state - #964

Merged
Germey merged 2 commits into
mainfrom
codex/mcp-oauth-fleet-redis-policy-20261008
Oct 8, 2026
Merged

Germey merged 2 commits into
mainfrom
codex/mcp-oauth-fleet-redis-policy-20261008

Conversation

@acedatacloud-dev

Copy link
Copy Markdown
Member

Summary

Allow the 25 other active hosted MCP Deployments to reach the existing encrypted OAuth Redis state store. The NetworkPolicy enumerates their current app labels and keeps ingress limited to those MCP Pods. Also extend the account MCP rollout wait from 300 to 600 seconds: its first shared-state deployment reached 2/2 current-image Pods, but the GitHub deploy job timed out while an old Pod was terminating.

Validation

  • Kubernetes server-side dry run of the Redis StatefulSet, Service, and NetworkPolicy.
  • Confirmed the allowlist exactly matches all 26 currently active OAuth MCP Deployments, including the account MCP.
  • sh -n acedatacloud/deploy/run.sh and git diff --check.

Rollout

Deploy this PR before the fleet OAuth code/configuration PR. Redis, its persistent volume, and the stable account MCP Secret already exist from #963. This change only widens ingress to the listed MCP Pods and adjusts the rollout wait.

@Germey
Germey merged commit 287df39 into main Oct 8, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants