[OAK-4291] FileStore.flush prone to races leading to corruption

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: Segment Tar 0.0.2, 1.6.0
• Component/s: segment-tar
• Labels: resilience
• Epic Link: Improve SegmentMK resilience
• Assignee: Michael Dürig
• Reporter: Michael Dürig
• Created: 22/Apr/16 19:36
• Updated: 27/Jun/16 08:21
• Resolved: 02/Jun/16 14:04

Description

There is a small window in FileStore.flush that could lead to data corruption: if we crash right after setting the persisted head but before any delay-flushed SegmentBufferWriter instance flushes (see SegmentBufferWriterPool.returnWriter()) then that data is lost although it might already be referenced from the persisted head.

We need to come up with a test case for this.

A possible fix would be to return a future from SegmentWriter.flush and rely on a completion callback. Such a change would most likely also be useful for OAK-3690.

Issue Links

• is related to: OAK-3348 Cross gc sessions might introduce references to pre-compacted segments