honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Make sure successful operations are correct if second phase is fast 

In TransportSearchTypeAction we  need to increment successful ops
first before we increment and compare the exit condition otherwise if we
are fast we could concurrently update totalOps but then preempt one
of the threads which can cause the successor to read a wrong value from
successfulOps if second phase is very fast ie. searchType == count etc.
This can cause wrong success stats in the search response.
This commit is contained in:
Simon Willnauer 2014-04-07 22:05:47 +02:00
parent f2181d5cbf
commit befa833385
1 changed files with 5 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/main/java/org/elasticsearch/action/search/type/TransportSearchTypeAction.java
View File

@ -246,11 +246,13 @@ public abstract class TransportSearchTypeAction extends TransportAction<SearchRe
void onFirstPhaseResult(int shardIndex, ShardRouting shard, FirstResult result, ShardIterator shardIt) {
result.shardTarget(new SearchShardTarget(shard.currentNodeId(), shard.index(), shard.id()));
processFirstPhaseResult(shardIndex, shard, result);
// we need to increment successful ops first before we compare the exit condition otherwise if we
// are fast we could concurrently update totalOps but then preempt one of the threads which can
// cause the successor to read a wrong value from successfulOps if second phase is very fast ie. count etc.
successulOps.incrementAndGet();
// increment all the "future" shards to update the total ops since we some may work and some may not...
// and when that happens, we break on total ops, so we must maintain them
int xTotalOps = totalOps.addAndGet(shardIt.remaining() + 1);
successulOps.incrementAndGet();
final int xTotalOps = totalOps.addAndGet(shardIt.remaining() + 1);
if (xTotalOps == expectedTotalOps) {
try {
innerMoveToSecondPhase();