Concurrent scenarios should wait for the state file to release the lock, not directly fail.

[banjiuqingshan]

After f46d0bd, concurrent logrotate tasks will fail if status file is not specified.Exclusive access to the state file is indeed necessary.However, in the default scenario, waiting for the last logrotate task to complete is better than returning a failure message.In addition, if different state files are configured with the same log, the current locking mechanism does not seem to be able to avoid concurrent dumps of logs.

[kdudka]

There is already the --wait-for-state-lock option for this and I think it is obvious from the man page why it is not the default behavior. See #457 for the context.