Behaviour on setting TrailingLogs to 0

[fahadullah]

https://gist.github.com/fahadullah/e49a4578094f4efd7746

If TrailingLogs is set to 0 and raft node is restarted right after taking a snapshot, it tries to look up the last index but all entries were already compacted. TrailingLogs being 0 should be a valid config though a little inefficient under certain cases.

[Lz-Gustavo]

Is this still an observed issue on the current version?

Running the same test routine pointed by @fahadullah with some small modifications to adequate to the new API, I've reached the same "log not found" state. At first, I thought there could be a bug on InmemStore DeleteRange(), which is the implementation of LogStore interface used on unit tests. After some time, I've came back to the test definition and I think it's equivocated.

Please correct me if I'm wrong, but the expected log index after a successfully snapshot installation, and entire log truncation, shouldn't be zero? Not 102 as was currently defined.