Hello, I'm gregorst, the owner of COSMOON validator.

During the night of 16 february, it occurred a critical bug on Kusama node which caused my validator to slash.

[Why it occurred only now?]

I always used rocksdb as db storage, recently I made the switch on paritydb.

[Exaplain the bug]

Normally, when we use RocksDb on kusama, when we start the polkadot process, it creates inside the db storage, (ex: .local/share/polkadot/chains/ksmcc3/db) a file called LOCK, which avoids other polkadot instances be running on the same machine.

Let's assume, if the process is not responding, there are situations where the process is not killed properly, thanks to the LOCK file, if the process restarts , when the old is still active, the LOCK file kills the execution.

A critical bug, is currently active in the all last releases of polkadot where, WHEN using paritydb, it doesn't create this really important LOCK file; so this is what happens:
The Polkadot process, not seeing the LOCK file, assumes the instance has been terminated successfully when this is not true, so it continues the starting phase of polkadot process, allowing two or more instances to run.

In my case, the second instance started due to the missing LOCK file, corrupting my paritydb, and sending a bad block, causing my slash.

I ask the remotion of this slash due to the critical bug affecting paritydb

Polkadot version : Polkdot 0.8.28.1

[Debug]

Debugged on a new node, using screen

Kusama running with ParityDb:

Kusama running with RockDb