Github user arunmahadevan commented on a diff in the pull request:
https://github.com/apache/storm/pull/767#discussion_r42546136
--- Diff:
external/storm-hdfs/src/main/java/org/apache/storm/hdfs/bolt/AbstractHdfsBolt.java
---
@@ -127,9 +145,112 @@ public void run() {
}
@Override
+ public final void execute(Tuple tuple) {
+
+ synchronized (this.writeLock) {
+ boolean forceSync = false;
+ if (TupleUtils.isTick(tuple)) {
+ LOG.debug("TICK! forcing a file system flush");
+ forceSync = true;
+ } else {
+ try {
+ writeTuple(tuple);
+ tupleBatch.add(tuple);
+ } catch (IOException e) {
+ //If the write failed, try to sync anything already
written
+ LOG.info("Tuple failed to write, forcing a flush of
existing data.");
+ this.collector.reportError(e);
+ forceSync = true;
+ this.collector.fail(tuple);
+ }
+ }
+
+ if (this.syncPolicy.mark(tuple, this.offset) || (forceSync &&
tupleBatch.size() > 0)) {
+ int attempts = 0;
+ boolean success = false;
+ IOException lastException = null;
+ // Make every attempt to sync the data we have. If it
can't be done then kill the bolt with
+ // a runtime exception. The filesystem is presumably in a
very bad state.
+ while (success == false && attempts < fileRetryCount) {
+ attempts += 1;
+ try {
+ syncTuples();
+ LOG.debug("Data synced to filesystem. Ack'ing [{}]
tuples", tupleBatch.size());
+ for (Tuple t : tupleBatch) {
+ this.collector.ack(t);
+ }
+ tupleBatch.clear();
+ syncPolicy.reset();
+ success = true;
+ } catch (IOException e) {
+ LOG.warn("Data could not be synced to filesystem
on attempt [" + attempts + "]");
--- End diff --
@dossett as @revans2 brought up, I think its better to use the curly brace
{} anchor for logging throughout to be consistent.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
