ableegoldman commented on a change in pull request #9695:
URL: https://github.com/apache/kafka/pull/9695#discussion_r553007907
##########
File path: streams/src/main/java/org/apache/kafka/streams/KafkaStreams.java
##########
@@ -947,24 +948,65 @@ private StreamThread createStreamThread(final long
cacheSizePerThread, final int
* @return name of the added stream thread or empty if a new stream thread
could not be added
*/
public Optional<String> addStreamThread() {
- synchronized (changeThreadCount) {
- if (isRunningOrRebalancing()) {
- final int threadIdx = getNextThreadIndex();
- final long cacheSizePerThread =
getCacheSizePerThread(threads.size() + 1);
+ if (isRunningOrRebalancing()) {
+ final int threadIdx;
+ final long cacheSizePerThread;
+ synchronized (changeThreadCount) {
+ threadIdx = getNextThreadIndex();
+ cacheSizePerThread = getCacheSizePerThread(threads.size() + 1);
resizeThreadCache(cacheSizePerThread);
- final StreamThread streamThread =
createStreamThread(cacheSizePerThread, threadIdx);
- synchronized (stateLock) {
- if (isRunningOrRebalancing()) {
- streamThread.start();
- return Optional.of(streamThread.getName());
- } else {
+ }
+ final StreamThread streamThread =
createStreamThread(cacheSizePerThread, threadIdx);
+
+ synchronized (stateLock) {
+ if (isRunningOrRebalancing()) {
+ streamThread.start();
+ return Optional.of(streamThread.getName());
+ } else {
+ streamThread.shutdown();
+ threads.remove(streamThread);
+ resizeThreadCache(getCacheSizePerThread(threads.size()));
+ log.warn("Cannot add a stream thread in state " + state());
+ return Optional.empty();
+ }
+ }
+ }
+ log.warn("Cannot add a stream thread in state " + state());
+ return Optional.empty();
+ }
+
+ /**
+ * Removes one stream thread out of the running stream threads from this
Kafka Streams client.
+ * <p>
+ * The removed stream thread is gracefully shut down. This method does not
specify which stream
+ * thread is shut down.
+ * <p>
+ * Since the number of stream threads decreases, the sizes of the caches
in the remaining stream
+ * threads are adapted so that the sum of the cache sizes over all stream
threads equals the total
+ * cache size specified in configuration {@link
StreamsConfig#CACHE_MAX_BYTES_BUFFERING_CONFIG}.
+ *
+ * @return name of the removed stream thread or empty if a stream thread
could not be removed because
+ * no stream threads are alive
+ */
+ public Optional<String> removeStreamThread() {
+ if (isRunningOrRebalancing()) {
+ synchronized (changeThreadCount) {
+ for (final StreamThread streamThread : threads) {
+ if (streamThread.isAlive()) {
streamThread.shutdown();
+ if
(!streamThread.getName().equals(Thread.currentThread().getName())) {
Review comment:
Sorry, my suggestion was vaguely worded. I meant that we should only let
a StreamThread remove itself, if we detect that `removeThread` has been called
by a StreamThread (which should be possible). If it's called by something else,
then yes we should just pick a random thread to remove. I think that addresses
your concern in the first sentence of your reply, but let me know if I'm
misinterpreting it.
>I do not like the idea of not letting a thread remove itself because, if we
have one stream thread left not letting that thread remove it self (maybe
leaving a global thread) this seems incomplete.
That's a good point. In that case I would advocate for _only_ letting a
StreamThread remove itself. Do you see any problems with that?
----------------------------------------------------------------
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:
us...@infra.apache.org
