This process traverses through the archive directories and checks if the archive is old enough to be deleted. The process init_afd will automatically restart it when it terminates for whatever reason. When files are archived by the process sf_xxx, they are stored in the following directory structure (beginning at $AFD_WORK_DIR):
archive/hollywood/donald/n/archive_name/ | | | | | | | +-------> This unique directory name | | | is created when files are | | | being archived. | | +---------------> Due to the restriction that | | some unix system allow for | | only 1000 directories, sub- | | directories starting with | | zero are being created. | +--------------------> Remote user name. +----------------------------> Remote host name.
The time when the archive should be deleted is in the archive name, which has the following format:
nnnnnnnnnn_jjjj | | | +-------> Job Identifier +---------------> time when archive should be deleted. The format is unix time in seconds.
Files stored in here have the following file name:
ttttttttt_uuuu_ssss_<original file name> | | | | | | | +---> This is the original file name | | | part as send to the destination. | | +--------------> Split job counter. | +-------------------> A unique number. +--------------------------> Time when this file was picked up by the AMG.
Archive watch will walk through this directory structure every 10 minutes (ARCHIVE_STEP_TIME) and see if any archive has reached its deletion time. If you lower this value more archives ie. directories will be created. This has the disadvantage that there will be a high number of directories created and deleted again later.
Copyright © 1997 - 2005 by H.Kiehl Holger.Kiehl@dwd.de Last updated: 08.07.2005 |
Index | Home |