This release adds recovery paths for failed deployments, an audit trail for who changed which deployment revision, a self-healing storage layer for stuck PersistentVolumeClaims, and finer-grained container patching during upgrades.Documentation Index
Fetch the complete documentation index at: https://docs.chainstack.com/llms.txt
Use this file to discover all available pages before exploring further.
Recover from failed deployments
Failed deployments are no longer terminal. Submitting a new revision against a deployment in thefailed state transitions it back to pending and re-runs the deployment with the updated configuration. Operators can adjust the node spec and retry without manual database surgery or recreating the deployment from scratch.
Revision audit trail
Each deployment revision now records who initiated it. The revision payload and history view in the Control Panel expose the initiating user, giving teams a clear audit trail of who changed which node configuration and when.Self-healing stuck volumes
cp-bolt now ships a reaper that detects PersistentVolumeClaims stuck on finalizers after pod deletion and clears them automatically. Stuck PVCs were a common cause of node restarts that never completed on clusters where the CSI driver crashed or was reinstalled mid-operation. The reaper closes that failure mode and exposes metrics so operators can see when it has had to intervene.Container-level upgrade patching
cpctl upgrade gains an option to patch individual containers as part of the upgrade pipeline, instead of treating every release as an all-or-nothing chart rollout. This shortens the change footprint for targeted image bumps and reduces the blast radius of routine maintenance upgrades.
Component versions
| Component | Version |
|---|---|
| cp-ui | v3.0.0 |
| cp-deployments-api | v0.34.0 |
| cp-workflows | v1.14.0 |
| cp-auth | v0.4.2 |
| cp-bolt | v1.9.0 |
| cp-tracer | v0.1.1 |