Bugzilla – Bug 3199
NPE on recovery of delete request
Last modified: 2005-04-20 13:44:37
You need to log in before you can comment on or make changes to this bug.
I'm getting the following exception on recovery of at least some of my jobs which were in the FileCleanUp state when the container crashed: java.lang.NullPointerException at org.globus.util.GlobusURL.<init>(GlobusURL.java:46) at org.globus.transfer.reliable.service.TransferWork.getTransferClient(TransferWork.java:320) at org.globus.transfer.reliable.service.TransferWork.run(TransferWork.java:671) at org.globus.wsrf.impl.work.WorkManagerImpl$WorkWrapper.run(WorkManagerImpl.java:342) at java.lang.Thread.run(Thread.java:534) I will attach a multi-container log and the IDs of the jobs that didn't recover properly.
Created an attachment (id=582) [details] Bzipped multi-container log from GRAM recoverability tester The following jobs did not recover properly: 5ae92872-b126-11d9-9561-000d61215ff0 (transferId=29047) 50141146-b126-11d9-ba4b-000d61215ff0 (transferId=29044)
I noticed that the NPE's don't actually show up in the container log for some reason. There were exactly two of them, though, so this looks like a fairly consistent occurance.
Changed target milestone to 4.0.1 as other gram recoverability bugs
Created an attachment (id=583) [details] patch to fix NPE Works as expected in my tests.
The patch worked for me too. The recoverability tester succeeded for the first time. The code is checked into the gram_recoverability_tester_branch branch. Stu doesn't want to check in this code for 4.0.0 RC3, so this will be included in 4.0.1 in about a month and a clear release notes statement will be added to 4.0.0 that makes it known that a patch is waiting and ready.
