Node gets wrong upgrade_log entries after restart and runs the wrong upgrade

Overwiew:
Node gets wrong upgrade_log entries after restart and run the wrong upgrade.

Build Info:
indy-node 1.1.171

Steps to Reproduce:
1. Install a pool.
2. Schedule an upgrade txn.
3. Cancel the txn from Step 2.
4. Shut down the primary and start it after some time (~1 minute).

Actual Results:
The primary has 3 entries in upgrade_log: scheduled, cancelled, scheduled. All other nodes have 2 entries.

Expected Results:
All nodes should have 2 entries in upgrade_log in this case.