Checksum enabled large volume with multiple rebuilding
- Create a 50 Gi volume. write around 30 Gi data into it.
- Enable the setting
Snapshot Data Integrity
.
- Keep writing in the volume continuously using dd command like
while true; do dd if=/dev/urandom of=t1 bs=512 count=1000 conv=fsync status=progress && rm t1; done
.
- Create a recurring job of backup for every 15 min.
- Delete a replica and wait for the replica rebuilding.
- Compare the performance of replica rebuilding from previous Longhorn version without the setting
Snapshot Data Integrity
.
- Verify the Longhorn manager logs, no abnormal logs should be present.
- Repeat the steps of deletion of the replica and verify Longhorn doesn’t take more time than the first iteration.
[Edit]