Incorrect database status with multiple databases in one cluster
Details
| Detail name | Value |
|---|---|
| Changelog Number | 14293 |
| Type | Bug |
| Status | Resolved |
| Affected Versions | |
| Fix Versions | Exasol 8.34.0 |
| Resolution Date | 2025-04-16 |
Background
Multiple databases can run in one cluster. These databases may share nodes it runs on or each database may run on a separate group of nodes in the cluster. For example, database 1 may use n11..n13, and database 2 uses n14..16.
Description
In the following scenario, some databases may incorrectly display as being offline in EXAoperation:
- A cluster contains multiple databases
- The databases share one or more nodes
- One of the shared nodes fails, causing both databases to restart using a reserve node at the same time.
Additionally, the EXAoperation log contains the error:
System failed suddenly after startup
Trying to restart the database will also fail because the database is already running and returns the error message:
Unknown error: Could not start database: System does not have enough active nodes or DWAd was not able to create startup parameters for System
Note
If there is a future crash or node failure, the database will not be able to start properly.
Workaround
The database is usable as normal, so no immediate action is required. To ensure that the status is recorded correctly and automatic restarts happen correctly for future node failures or database crashes, the whole cluster needs to be restarted in the usual way, as documented in "Stop and Start Nodes"
Fix
The database status reflects the actual state of the database in this scenario.
