I don't have a repro yet. But this is local handle state, not remote metadata state. AFAICS the remote metadata state handling is safe.
The core of the problem is that a second call to #close on a handle, will return successfully before the first one is even finished. If that thread then takes any action based on what it thinks the state of the ledger is, it risks being wrong, since the first call to #close could fail. [ Full content available at: https://github.com/apache/bookkeeper/issues/1712 ] This message was relayed via gitbox.apache.org for [email protected]
