Subject
Repeated Bitcoin Knots LevelDB Chainstate Corruption on Start9 Pure (Internal NVMe)
Device
- Hardware: Start9 Pure
- Power: On UPS (no power interruptions)
- Storage: Internal NVMe (Pure default)
Issue Summary
Bitcoin Knots repeatedly crashes due to LevelDB chainstate corruption, even after multiple clean, uninterrupted reindexes. The issue persists across reinstalls and reindexes and appears unrelated to power loss or user interruption.
Observed Behavior
- Bitcoin Knots syncs successfully to ~99.9%
- Runs normally for a period
- Then crashes with a fatal LevelDB error
- Corruption occurs in different
.ldbfiles each time - Electrs subsequently fails due to loss of RPC
- Reindexing temporarily resolves the issue, but corruption returns
Key Error (repeats across attempts)
LevelDB read failure: Corruption: block checksum mismatch:
/root/.bitcoin/chainstate/*.ldb
Fatal LevelDB error
Error reading from database, shutting down.
PROCESS TERMINATED
Example from most recent crash:
Corruption: block checksum mismatch:
/root/.bitcoin/chainstate/038987.ldb
Troubleshooting Already Performed
- Multiple full and chainstate reindexes
- Reindexes were never interrupted
- System rebuild performed
- Services left untouched during sync
- Electrs and Sparrow troubleshooting ruled out (downstream symptoms)
- UPS in place to prevent power loss
- Issue persists after each rebuild
Conclusion
Given:
- Repeated corruption after clean reindexes
- Different
.ldbfiles affected - Internal NVMe storage (Pure)
- UPS-protected power
- No user interruptions
This strongly suggests a storage-level hardware fault or NVMe/controller issue on the Pure.
Request
Please advise on:
- Any additional diagnostics you would like me to run, or
- Next steps for hardware inspection / repair / replacement (RMA)
I’m happy to provide additional logs or follow guided steps as needed.
Thank you.