I’ve noticed jobs failing during the output data staging step that still get marked as “completed” rather than failed. Failures were most likely caused by SCP timeouts to a very slow host.
Looking at the code, outputDataStaging() [1] doesn’t call the checkFailures() [2] method like inputDataStaging and executeJobSubmission. It instead embeds the contents of checkFailures() directly in the method. The problem is that while checkFailures() throws an exception if there was a task failure, it looks like outputDataStaging does not throw an exception. Looks like a bug, but was this done for some reason? Marlon [1] https://github.com/apache/airavata/blob/master/modules/gfac/gfac-impl/src/main/java/org/apache/airavata/gfac/impl/GFacEngineImpl.java#L749 [2] https://github.com/apache/airavata/blob/master/modules/gfac/gfac-impl/src/main/java/org/apache/airavata/gfac/impl/GFacEngineImpl.java#L613
smime.p7s
Description: S/MIME cryptographic signature
