Motivation

The RangeCache class contains several race conditions which cause instability.

When one thread removes the entry and another one uses it, that will become a problem.

The cacheEvictionIntervalMs setting is 10 ms by default. This results in the
RangeCache.evictLEntriesBeforeTimestamp method getting called about 100 times per second.
The default expiration is managedLedgerCacheEvictionTimeThresholdMillis which is 1000 ms by default.

It's also possible that 2 threads remove the entry at the same time. ManagedLedgerImpl.invalidateEntriesUpToSlowestReaderPosition will result in calls to RangeCache.removeRange method. These calls happen independently of the RangeCache.evictLEntriesBeforeTimestamp calls so there's a chance for race conditions.

Modifications

• Add checks to validate that the value hasn't already been released before it is removed.
  • add extra .retain() and .release() calls
  • add a way to validate that the value matches the key
    • this is to prevent a possible race condition where the entry has already been recycled and contains another valid entry which is for a different key
• Make all removals using the remove(key, value) method so that removals remove the correct value exactly once
  • Add a wrapper object for the value to ensure this so that exact identity match is used for the removal.
  • The wrapper instance is recycled to reduce the overhead.

Documentation

• doc
• doc-required
• doc-not-needed
• doc-complete