fix: default to Recreate strategy and allow CNPG plugins

[kernelb00t]

Summary

• Upgrade of Paperless gets stuck due to Deployment strategy #79 – New paperless.strategy and redis.strategy values, defaulting to type: Recreate, so upgrades with ReadWriteOnce PVCs no longer get stuck on Multi-Attach errors. For Recreate, the chart renders rollingUpdate: null so existing releases (whose live Deployment still carries the default rollingUpdate block) upgrade without the API server rejecting the change. Set type: RollingUpdate to opt back in (RWX volumes).
• Allow plugin registration for provisioned CNPG cluster #77 – New cnpg.plugins list, rendered as spec.plugins on the CNPG Cluster, to register CNPG-I plugins such as barman-cloud.cloudnative-pg.io for backups / WAL archiving.
• Chart version 1.8.0 → 1.8.1, README parameter table updated.

Closes #79
Closes #77

Testing

• helm lint with ci/ct-values.yaml
• helm template checked: default (Recreate + rollingUpdate: null), RollingUpdate, strategy set to null (field omitted), and cnpg.plugins populated

🤖 Generated with Claude Code

https://claude.ai/code/session_01QD1kFPcmFFoZmmgw6t4XGw

---

Generated by Claude Code

---

Summary by cubic

Fixes upgrade stalls on ReadWriteOnce PVCs by defaulting to a Recreate update strategy, and adds support for registering CNPG-I plugins.

• New paperless.strategy and redis.strategy values default to type: Recreate, preventing Multi-Attach errors when pods move to a new node. For Recreate, the chart renders rollingUpdate: null so existing Deployments upgrade cleanly. Set type: RollingUpdate to opt back in (requires RWX volumes).
• New cnpg.plugins list is rendered as spec.plugins on the CNPG Cluster, enabling plugins like barman-cloud.cloudnative-pg.io for backups and WAL archiving. Requires CNPG >= 1.26 and the plugin deployed separately.
• Bumps chart version to 1.8.1 and updates the README parameter table.

Closes #79 and #77.

^{Written for commit 2ef4912. Summary will update on new commits.}

[cubic-dev-ai]

All reported issues were addressed across 7 files

<sub>Reply with feedback, questions, or to request a fix.

View guided diff | Turn on auto-fix | Re-trigger cubic</sub>

[kernelb00t]

Integration Test failed on 2ef4912 at the Smoke test (HTTP GET) step. The cause is the CI harness, not this PR:

• every pod was Running, kubectl rollout status succeeded, and Paperless logged Listening at: http://:::8000;
• 2ef4912 only changes README.md. The chart is byte-identical to d52878e, where Integration Test passed;
• the step output shows kubectl's attach message ("If you don't see a command prompt…") and no HTTP code. kubectl run --rm -i can lose a short-lived pod's output if the container exits before kubectl attaches, so grep saw nothing.

I re-ran the failed job once. If it keeps flaking, the fix is to make the smoke test robust, e.g. retry the curl in a loop, or use kubectl port-forward + curl from the runner. That would go in a separate PR.

---

Generated by Claude Code