This topic describes critical bug fixes for StarRocks versions, including impact levels and recommended actions. Use the problem descriptions to identify issues that may affect your clusters and take preventive measures.
Background
Bug and fix rules
-
Bug persistence: A bug found in a specific version is generally also present in all earlier versions. For example, a bug in version 3.3.8 usually exists in earlier versions such as 3.3.2 and 3.2.0. Any exceptions are clearly noted.
-
Fix inclusion: A bug fixed in a specific version remains fixed in all subsequent versions. For example, a bug fixed in version 3.3.8 is also fixed in 3.3.13 and all later versions.
Bug levels
| Level | Action | Impact |
|---|---|---|
| P0 | Upgrade immediately | Affects online services when triggered; can impact query accuracy and write success rates |
| P1 | Upgrade recommended | Prevents potential issues before they occur |
| P2 | Upgrade optional | Issues occur occasionally; can usually be resolved by rewriting a method or restarting the service |
A version affected by a P0 bug immediately enters the approaching End of Service and Support (EOS) state and reaches EOS three months later. After a version reaches EOS, technical support and bug fixes are no longer provided. Upgrade as soon as possible.
August 2025
| Bug description | Bug level | Cause | Instance type | Affected versions | Recommended action |
|---|---|---|---|---|---|
| FE node fails to start | P0 | The FE replay thread incorrectly deletes StarOS metadata, which can prevent the FE node from starting. | Storage-compute separation | 3.3.0 to 3.3.8 | Upgrade to the latest 3.3.x version with compute-storage separation. The current latest version is 3.3.13-1.1.1. |
| BE node goes down | P0 | Batch Publish can cause version discontinuities, which may cause a BE node to go down. | Storage-compute separation | 3.3.2 to 3.3.13-1.0.0; 3.2.x | Upgrade to the latest 3.3.x version for shared-data clusters. The current latest version is 3.3.13-1.1.1. |