Known bugs for DIAL release 1.20 06sep05 --------------------- 1. 08jun05 Default DIAL version for dialroot is not 1.20. ******** Fixed in 1.20.01 --------------------- 2. 09jun05 LSF jobs that are killed after completeion are left in the running state. ******** Fixed in 1.20.02 --------------------- 3. 09jun05 Demo 6 (application atlasopt) fails with the BNL LSF sl3 service due to messages appearing when ATLAS libraries are loaded. --------------------- 4. 09jun05 Root command get_results() sometimes produces a cp error message. --------------------- 5. 13jun05 ATLAS jobs cannot be run with dcache input. Modify MagdaFileCatalog to allow selection based on an ordered list of file protocols. Rebuild root to include dcache and RFIO libraries. ******** Fixed in 1.20.04 with new build of root 4.03.02 --------------------- 6. 10apr05 Stop time reported for LSF jobs is the time DIAL realizes they are stopped rather than the time reported by LSF. --------------------- 6. 09aug05 DIAL scheduler service will crash at startup if the UID service does not authorize the certificate used to run the scheduler. --------------------- 7. 06sep05 Client error message fetching results: root [29] get_results() cp: missing destination file Try `cp --help' for more information. RootHistogramDataset::clone: File copy failed: /usatlas/magdacache003/dialstore/dial_lf1000106990 Job::copy_ctor: Result clone failed. ******** Does not happen in 1.21. --------------------- 8. 23sep05 Jobs failing at BNL because disks are not mounted on workers. See BNL ticket 1830. --------------------- 9. 23sep05 MagdaFileCatalog::put(...) does not check whether the call to magda_putfile succeeds. It incorrectly reports success after this command fails. --------------------- 10. 14oct05 DialWsClient uses error() to determine if the service is valid. But error indicates the status of the last call, not the validity. Thus service becomes invalid after a request for a nonexistent ID. ******** Fixed in 1.21.12 ---------------------