tree aba323b2bf23affec7f49186ffe047076f6a1be9
parent 6b892a45f02a6d8d0b1d33235d4134d91e3ea111
author massakam <massakam@yahoo-corp.jp> 1562918686 +0900
committer Enrico Olivelli <eolivelli@gmail.com> 1562918686 +0200

Issue #2103: Avoid stop of entry log compaction

### Motivation

As mentioned in #2103, if an exception occurs during compaction of a specific entry log, `GarbageCollectorThread` does not perform compaction of other entry logs until the bookie server is restarted. As a result, the number of entry logs continues to increase and eventually it will run out of disk space.

### Changes

The cause of the compaction stop is that the `compacting` flag remains true if `compactor.compact(entryLogMeta)` throws some exception.
https://github.com/apache/bookkeeper/blob/b2e099bbc7b13f13825fe78ab009ca132cb3a9ba/bookkeeper-server/src/main/java/org/apache/bookkeeper/bookie/GarbageCollectorThread.java#L504-L519

Therefore, fixed `GarbageCollectorThread` so that it set the compaction flag to false even if compaction of a specific entry log fails.

Master Issue: #2103

Reviewers: Enrico Olivelli <eolivelli@gmail.com>, Sijie Guo <sijie@apache.org>

This closes #2121 from massakam/entry-log-compaction, closes #2103