When the user runs ipfix-rita stop, it is fair to say that the user expects to be able to analyze any data collected during the current day. We currently do not support this.
If we naively mark the database's import_finished field on exit, then the system will become unstable if the user starts the program within the same day. This is due to the fact that IPFIX-RITA does not check if the import_finished flag is set to true before it starts writing out results.
In order to support clean shutdowns, we need to do one of two things:
Check the import_finished flag before writing, if it is set, create a new database, otherwise carry on as usual.
Check the import_finished flag before writing, if it is set, check if RITA has started analysis (currently impossible), if it has started, create a new database, otherwise unset the import_finished flag and carry on as usual.
Since the latter requires a change in RITA and the latter could build on the prior, I suggest we implement the first option to start.
When the user runs
ipfix-rita stop
, it is fair to say that the user expects to be able to analyze any data collected during the current day. We currently do not support this.If we naively mark the database's
import_finished
field on exit, then the system will become unstable if the user starts the program within the same day. This is due to the fact that IPFIX-RITA does not check if theimport_finished
flag is set to true before it starts writing out results.In order to support clean shutdowns, we need to do one of two things:
import_finished
flag before writing, if it is set, create a new database, otherwise carry on as usual.import_finished
flag before writing, if it is set, check if RITA has started analysis (currently impossible), if it has started, create a new database, otherwise unset the import_finished flag and carry on as usual.Since the latter requires a change in RITA and the latter could build on the prior, I suggest we implement the first option to start.