Introduction
Even though HeatWave is highly stable, DBAs occasionally encounter load failures, memory issues, or cluster instability. This article covers the most common HeatWave problems and how to fix them.
1. HeatWave Load Failures
Common Causes
-
Unsupported column types
-
Corrupted InnoDB data
-
Exceeded memory limits
-
Large JSON fields
Solutions
Run:
Then fix:
-
Convert JSON → Generated columns
-
Reduce precision on large DECIMAL(65,30)
-
Normalize string columns
-
Increase HeatWave cluster size
2. HeatWave Cluster Node Failures
Nodes may enter ERROR state.
Check reason
Fix
-
Cluster auto-recovers if possible
-
If not → Restart HeatWave cluster:
3. Memory Allocation Issues
Cause:
Data segments exceed node RAM.
Fixes:
-
Add more worker nodes
-
Drop unneeded columns from HeatWave load list
-
Use
sys.heatwave_advisorto optimize data types
4. Query Not Using HeatWave
Symptoms
EXPLAIN shows MySQL using InnoDB instead of HeatWave.
Check reason
Common causes
-
Unsupported functions
-
Joins missing primary key
-
ON clause mismatch
-
Complex subqueries not rewriteable
Solutions
-
Rewrite query
-
Add primary keys
-
Apply required CAST/CONVERT functions
5. Table Out of Sync with HeatWave
Cause:
DML operations happen frequently.
Fix:
Reload:
Conclusion
Most HeatWave issues are fixable using built-in advisor tools, proper monitoring, and query adjustments. Understanding internals dramatically reduces troubleshooting time.