Skip to content

Release Notes

Release notes record what changed in each Nexa platform version so you can plan an upgrade with no surprises. Read the entry for every version between your current one and your target before you upgrade — breaking changes and required manual steps are called out per release. Newest version is at the top.

Each entry follows the same structure: Breaking changes (with the exact action you must take) come first, followed by new features, improvements, bug fixes, known issues, and upgrade notes. Versions follow semantic versioning: MAJOR.MINOR.PATCH. A MAJOR bump can contain breaking changes; MINOR adds features backward-compatibly; PATCH is fixes only.

Adds Snowflake as a first-class data platform alongside Databricks and moves agent authoring into its own API. Contains one breaking change to the connection secret layout — read the required action before upgrading.

Change Who is affected Required action
Data-platform connection secrets moved from a single nexa-api-secrets Secret to per-service SecretProviderClass entries synced from your cloud secret store. All deployments. Before upgrading, add the new keys to your AWS Secrets Manager / Azure Key Vault secret and confirm each service’s SecretProviderClass references them. See Configuration Reference. The old flat Secret is ignored after this release.
Agent endpoints moved from nexa-backend to the new nexa-agents-api service. Deployments using agents/automations. Deploy the nexa-agents-api overlay and its namespace as part of this upgrade; update any external callers to the new base path.
  • Snowflake data platform support. Deploy nexa-snowflake-api to run pipelines and agents against Snowflake, selected per environment. Databricks deployments are unaffected.
  • Agentic automations promotion across environments (dev → test → prod) with change review.
  • Cost tracking dashboards for warehouse and compute spend under Governance.
  • Rolling updates now wait on stricter readiness probes, reducing the chance of the web tier reaching an API that is still starting.
  • Faster cold start for the API services (smaller runtime image layers).
  • Clearer ArgoCD sync health for multi-service releases.
  • Fixed run_id not being returned from the Databricks API, which broke workflow log retrieval.
  • Fixed PII tags being dropped when a source column was renamed during curation.
  • Fixed a race where an automation could deploy before its dependent projection finished.
  • On very large glossary imports, the first page load after upgrade can be slow while the cache warms. Workaround: reload after ~30 seconds.
  • Azure deployments: certificate renewal events are not yet surfaced in the Alerts view (tracked for a patch release).

Standard minor upgrade with one required secret-store action (above). Follow the Upgrade Guide — back up the application database first, upgrade non-production before production, and validate health endpoints after each service syncs.

Patch release. Drop-in; no required action.

None.

None.

  • Reduced ArgoCD sync noise for the dev and test environments.
  • Fixed intermittent 502s from the gateway during rolling updates of nexa-web.
  • Fixed a config-reference typo that set the wrong default log level in test.

None new since v1.3.0.

Patch upgrade — bump the image tags and sync. See the Upgrade Guide.