9.18.3.1 Disk Collection Agent - Real-Time
The Disk collection agent collects files from a local file system and inserts them into a workflow.
The agent's source directory is continuously scanned for all files that match a filter expression. The matched files are then temporarily moved to an automatically created subdirectory before processing. This is to prevent that the same set of files are collected by multiple workflows.
The collected files are routed to the workflow as partial data sets, i e part of a bytearray or a decoded UDR, via FileSend
types. These UDRs contain the original data and a sequence number. When all the partial data sets have been sent by the agent, it will send a FileEnd
UDR to the workflow. The workflow must acknowledge the reception by routing back the same UDR type, or a timeout error will occur. For further information about timeout errors, see 9.18.3.1.1 Disk Collection Agent Configuration - Real-Time.
UDR Flow
When a file has been successfully processed by the workflow, the agent offers the possibility of moving, renaming, or removing the original file.
If the agent fails to read, move or decode a file, a FileError
UDR type is routed to the workflow. In case of decoding errors, the erroneous file is also sent to DR_ERROR_DIR
in the source directory.
If the EC is terminated while the workflow is running, the temporary files will remain on the file system. When the workflow is restarted, these files will be moved to the original source directory.
Note!
Since partial data is routed to the workflow, it may be duplicated when temporary files are moved back to the source directory after a workflow restart.
If you require to run the batch-based Disk collection agent in a real-time workflow, see 9.87 Batch-Based Real-Time Agents.