Github user jtstorck commented on a diff in the pull request:
https://github.com/apache/nifi/pull/2971#discussion_r214136451
--- Diff:
nifi-nar-bundles/nifi-hadoop-bundle/nifi-hdfs-processors/src/main/java/org/apache/nifi/processors/hadoop/PutHDFS.java
---
@@ -266,6 +268,13 @@ public Object run() {
throw new
IOException(configuredRootDirPath.toString() + " could not be created");
}
changeOwner(context, hdfs, configuredRootDirPath,
flowFile);
+ } catch (IOException e) {
+ if (!Strings.isNullOrEmpty(e.getMessage()) &&
e.getMessage().contains(String.format("Couldn't setup connection for %s",
ugi.getUserName()))) {
--- End diff --
@ekovacs I think we should be more selective in this check. I don't think
there's a better way to detect this error scenario than string matching at this
point, but the exception stack should be inspected to see if you can find the
GSSException as the root cause:
`Caused by: org.ietf.jgss.GSSException: No valid credentials provided
(Mechanism level: Failed to find any Kerberos tgt) `
If you iterate through the causes when PutHDFS encounters an IOException,
and see that GSSException, we can do a penalize with a session rollback.
Otherwise, we'd want to pass the flowfile to the failure relationship.
---
