S3 Writes Hang Due to Transient Aggregate Lock Conflicts During Dynamic Grow
Applies to
- ONTAP S3
- ONTAP 9
Issue
- Users unable to write data to S3 buckets on the SVM.
- Symptoms and log outputs observed were:
- S3 write operations hang or timeout; reads succeed.
- Application error:
Upload operation failed! InvalidArgument: x-amz-content-sha256 must be UNSIGNED-PAYLOAD, STREAMING-AWS4-HMAC-SHA256-PAYLOAD or a valid sha256 value.
- EMS logs:
[Nodename:config_thread:raid.aggr.lock.conflict:info]: Exclusive access on the aggregate by operation a:w:dynamic_grow failed because of a conflict with current operation a:w:dynamic_grow. Retry after a few minutes.[Nodename:wafl_exempt17:wafl.cp.toolong:error]: Aggregate experienced a long CP.[kern_mgwd:info] rdb::TM: Apparent starvation! Time elapsed since beginning of coord_commit (31812ms) is greater than LOCAL_COMMIT_BUDGET (17000ms)[kern_mgwd:info] unstableQuorum: Declare unstable Quorum in epoch: 373 at 140543s (transaction manager).
- Aggregate intermittently reported offline by mgwd/rdb.
