Github user arunmahadevan commented on a diff in the pull request:
https://github.com/apache/storm/pull/1044#discussion_r55965890
--- Diff:
external/storm-hdfs/src/main/java/org/apache/storm/hdfs/bolt/AbstractHdfsBolt.java
---
@@ -198,22 +194,62 @@ public final void execute(Tuple tuple) {
}
}
- if(this.rotationPolicy.mark(tuple, this.offset)) {
- try {
- rotateOutputFile();
- this.rotationPolicy.reset();
- this.offset = 0;
- } catch (IOException e) {
- this.collector.reportError(e);
- LOG.warn("File could not be rotated");
- //At this point there is nothing to do. In all
likelihood any filesystem operations will fail.
- //The next tuple will almost certainly fail to write
and/or sync, which force a rotation. That
- //will give rotateAndReset() a chance to work which
includes creating a fresh file handle.
- }
+ if (writer != null && writer.needsRotation()) {
+ doRotationAndRemoveWriter(writerKey, writer);
}
}
}
+ private AbstractHDFSWriter getOrCreateWriter(String writerKey, Tuple
tuple) throws IOException {
+ AbstractHDFSWriter writer;
+
+ writer = writers.get(writerKey);
+ if (writer == null) {
+ if (writers.size() >= maxOpenFiles)
+ {
+ String keyToOldest = getKeyToOldestWriter();
+ AbstractHDFSWriter oldest = writers.get(keyToOldest);
+ rotateOutputFile(oldest);
+ writers.remove(keyToOldest);
--- End diff --
Is the oldest entry removed to maintain the max size of the writers
hashtable (max open files limit) ? nit: most of the above can be taken care of
if the writers map was a LinkedHashMap extension with a LRU logic.
---
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.
---
