Github user steveloughran commented on a diff in the pull request:
https://github.com/apache/spark/pull/14371#discussion_r72426791
--- Diff:
core/src/main/scala/org/apache/spark/rdd/ReliableCheckpointRDD.scala ---
@@ -166,17 +166,25 @@ private[spark] object ReliableCheckpointRDD extends
Logging {
val tempOutputPath =
new Path(outputDir,
s".$finalOutputName-attempt-${ctx.attemptNumber()}")
- if (fs.exists(tempOutputPath)) {
- throw new IOException(s"Checkpoint failed: temporary path
$tempOutputPath already exists")
- }
val bufferSize = env.conf.getInt("spark.buffer.size", 65536)
- val fileOutputStream = if (blockSize < 0) {
- fs.create(tempOutputPath, false, bufferSize)
- } else {
- // This is mainly for testing purpose
- fs.create(tempOutputPath, false, bufferSize,
- fs.getDefaultReplication(fs.getWorkingDirectory), blockSize)
+ val fileOutputStream = try {
+ if (blockSize < 0) {
+ fs.create(tempOutputPath, false, bufferSize)
+ } else {
+ // This is mainly for testing purpose
+ fs.create(tempOutputPath, false, bufferSize,
+ fs.getDefaultReplication(fs.getWorkingDirectory), blockSize)
+ }
+ } catch {
+ case e: IOException =>
+ // could be FileAlreadyExistsException, or a more fundamental
create failure
--- End diff --
There's some ambiguity on create(), at least with the special case of:
destination is a directory, and local filesystems don't throw what is a Hadoop
exception. [filesystem spec]
One thing to consider is: why wrap any IOE here at all? Is it the message
or the declaration of the path which matters?
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
