szlta commented on a change in pull request #1951:
URL: https://github.com/apache/hive/pull/1951#discussion_r571975616
##########
File path:
llap-server/src/java/org/apache/hadoop/hive/llap/io/api/impl/LlapIoImpl.java
##########
@@ -229,11 +230,20 @@ public void debugDumpShort(StringBuilder sb) {
new LinkedBlockingQueue<Runnable>(),
new
ThreadFactoryBuilder().setNameFormat("IO-Elevator-Thread-%d").setDaemon(true).build());
FixedSizedObjectPool<IoTrace> tracePool = IoTrace.createTracePool(conf);
+ if (isEncodeEnabled) {
+ int encodeThreads = numThreads * 2;
Review comment:
Yeah quite arbitrary I'll give you that.
Text reading is actually started by the "regular" IO threads, and if
encoding is needed for cache insertion, than one of these threads can produce
more async encode tasks. This may end up being bursty, and according to what
I've seen a bigger thread pool could come handy.
Anyway I made this configurable now.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
