Information about the nodes/partitions is stored within the file set, so if job is changed to use different number of nodes, it can cause errors even when overwriting the file. When one job writes to a file set or dataset and another job then reads them, it needs to be read using same number of nodes in APT_CONFIG_FILE to prevent partition errors. When you create a partitioned file set or data set on multiple nodes, DataStage actually create multiple partition files, one per node. The user changing configuration file had expected no problem because they chose File Set stage option to overwrite the file each time, thus expected file with old configuration to be replaced with file using new node configuration, but the above error occurred.
Message: FS_DSPD860A_010_T3_APPL_DIM_Update: Error when checking operator: Node name "node2" not in config fileĮrror when checking operator: Node name "node3" not in config file Invalid fileset /data/DSPD860A_T3_APPL_DIM_Insert20150504.fs: problem with partitioning Message: main_program: Invalid fileset /data/DSPD860A_T3_APPL_DIM_Update20150504.fs: problem with partitioning If user or administrator either changes APT_CONFIGURATION_FILE to point to a new *.apt node configuration file, or changes the definition of the current configuration file to a definition containing fewer nodes, then the job may fail on next run with error for each node that is no longer in the configuration file.
A DataStage job containing a File Set stage is setup to overwrite the File Set each job run.
0 kommentar(er)
