Github user dineshjoshi commented on a diff in the pull request:
https://github.com/apache/cassandra/pull/239#discussion_r201463376
--- Diff:
src/java/org/apache/cassandra/net/async/ByteBufDataOutputStreamPlus.java ---
@@ -135,6 +140,53 @@ public ChannelFuture writeToChannel(ByteBuffer buffer)
throws IOException
return channelFuture;
}
+ /**
+ * Writes all data in file channel to stream BUFFER_SIZE at a time.
+ * Closes file channel when done
+ *
+ * @param f
+ * @return number of bytes transferred
+ * @throws IOException
+ */
+ public long writeToChannel(FileChannel f, StreamRateLimiter limiter)
throws IOException
+ {
+ final long length = f.size();
+ long bytesTransferred = 0;
+
+ try
+ {
+ while (bytesTransferred < length)
+ {
+ int toRead = (int) Math.min(bufferSize, length -
bytesTransferred);
+ NonClosingDefaultFileRegion fileRegion = new
NonClosingDefaultFileRegion(f, bytesTransferred, toRead);
+
+ if
(!Uninterruptibles.tryAcquireUninterruptibly(channelRateLimiter, toRead, 5,
TimeUnit.MINUTES))
+ throw new IOException(String.format("outbound channel
was not writable. Failed to acquire sufficient permits %d", toRead));
+
+ limiter.acquire(toRead);
+
+ bytesTransferred += toRead;
+ final boolean shouldClose = (bytesTransferred == length);
// this is the last buffer, can safely close channel
+
+ channel.writeAndFlush(fileRegion).addListener(future -> {
--- End diff --
Yes, AFAICT, netty guarantees that the whole region will be transferred or
an error generated. We don't need to check the number of bytes transferred as
we have handed off the file region to the channel.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
