mynameborat commented on a change in pull request #912: SEP-19 : Refactoring
sideInputs from SamzaContainer to ContainerStorageManager
URL: https://github.com/apache/samza/pull/912#discussion_r260973315
##########
File path:
samza-core/src/main/scala/org/apache/samza/storage/ContainerStorageManager.java
##########
@@ -364,14 +616,144 @@ public void start() throws SamzaException {
executorService.shutdown();
- // Stop consumers
- this.systemConsumers.values().forEach(systemConsumer ->
systemConsumer.stop());
+ // Stop store consumers
+ this.storeConsumers.values().forEach(systemConsumer ->
systemConsumer.stop());
// Now re-create persistent stores in read-write mode, leave
non-persistent stores as-is
recreatePersistentTaskStoresInReadWriteMode(this.containerModel,
jobContext, containerContext,
- storageEngineFactories, changelogSystemStreams, serdes,
taskInstanceMetrics, taskInstanceCollectors);
+ storageEngineFactories, serdes, taskInstanceMetrics,
taskInstanceCollectors);
- LOG.info("Restore complete");
+ LOG.info("Store Restore complete");
+ }
+
+ // Read sideInputs until all sideInputStreams are caughtup, so start() can
return
+ private void startSideInputs() {
+
+ LOG.info("SideInput Restore started");
+
+ // initialize the sideInputStorageManagers
+ getSideInputStorageManagers().forEach(sideInputStorageManager ->
sideInputStorageManager.init());
+
+ // start the checkpointing thread at the commit-ms frequency
+ sideInputsFlushFuture = sideInputsFlushExecutor.scheduleWithFixedDelay(new
Runnable() {
+ @Override
+ public void run() {
+ try {
+ getSideInputStorageManagers().forEach(sideInputStorageManager ->
sideInputStorageManager.flush());
+ } catch (Exception e) {
+ LOG.error("Exception during flushing side inputs", e);
+ sideInputException = Optional.of(e);
+ }
+ }
+ }, 0, new TaskConfig(config).getCommitMs(), TimeUnit.MILLISECONDS);
+
+ // set the latch to the number of sideInput SSPs
+ this.sideInputsCaughtUp = new
CountDownLatch(this.sideInputStorageManagers.keySet().size());
+
+ // register all side input SSPs with the consumers
+ for (SystemStreamPartition ssp : sideInputStorageManagers.keySet()) {
+ String startingOffset =
sideInputStorageManagers.get(ssp).getStartingOffset(ssp);
+
+ if (startingOffset == null) {
+ throw new SamzaException("No offset defined for SideInput
SystemStreamPartition : " + ssp);
+ }
+
+ // register startingOffset with the sysConsumer and register a metric
for it
+ sideInputSystemConsumers.register(ssp, startingOffset, null);
+
taskInstanceMetrics.get(sideInputStorageManagers.get(ssp).getTaskName()).addOffsetGauge(
+ ssp, ScalaJavaUtil.toScalaFunction(() ->
sideInputStorageManagers.get(ssp).getLastProcessedOffset(ssp)));
+
+ SystemStreamMetadata systemStreamMetadata =
streamMetadataCache.getSystemStreamMetadata(ssp.getSystemStream(), false);
+ SystemStreamMetadata.SystemStreamPartitionMetadata sspMetadata =
+ (systemStreamMetadata == null) ? null :
systemStreamMetadata.getSystemStreamPartitionMetadata().get(ssp.getPartition());
+
+ // record a copy of the sspMetadata, to later check if its caught up
+ initialSideInputSSPMetadata.put(ssp, sspMetadata);
+
+ // check if the ssp is caught to upcoming, even at start
+ checkSideInputCaughtUp(ssp, startingOffset,
SystemStreamMetadata.OffsetType.UPCOMING, false);
+ }
+
+ // start the systemConsumers for consuming input
+ this.sideInputSystemConsumers.start();
+
+ // create a thread for sideInput reads
+ Thread readSideInputs = new Thread(() -> {
Review comment:
Can we set this to be a daemon thread? The java default is non-daemon thread.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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
With regards,
Apache Git Services
