| Previous | Next |
| ERROR_CLUSTER_PARTIAL_READ | ERROR_CLUSTER_CANT_DESERIALIZE_DATA |
ERROR_CLUSTER_PARTIAL_WRITE
A short write must not be mistaken for an atomic cluster update
ERROR_CLUSTER_PARTIAL_WRITE indicates that fewer bytes were written than the cluster operation required. The destination may be an inter-node stream, a serialized configuration blob, or another internal channel. Because the full record was not produced, readers cannot safely infer that the prefix is a valid object or that the update committed.
Record the requested and completed byte counts and retain the original error from the destination. Check disk or memory pressure, connection resets, quota or size limits, and concurrent teardown. If a transaction or batch surrounds the write, verify that failure prevents commit and that retry reconstructs the complete record. Appending only the missing suffix is unsafe unless the documented protocol explicitly preserves write position and framing.
Checks before retrying the write
- Destination type and lifetime.
- Requested, written, and remaining byte counts.
- Whether any transaction was committed after the short write.
- Resource pressure, connection, quota, or teardown events.
References
- Microsoft: cluster log collection
- Microsoft: atomic cluster registry batch commands
- Microsoft: error handling practices for resource DLLs
Looking for a different code? Search another status or error code.