UFL doesn't support the usual hot/warn/cold failover, but it is possible to set up two instances of the UFL interface pointing at the same data folder.
Thus if one fails, the other instance will continue to process the files.
Does this match your needs for a failover?
Yes this match my requirement.
Thanks for information.
Thanks and Regards,
1 of 1 people found this helpful
Thank you for writing in PI Square !
The PI UFL interface does not support automated failover. However, to ensure that data continues to be written if an instance of the interface stops running, you can configure multiple instances of PI UFL to read the same data files or POP3 server. Multiple instances cannot read the same serial port.
Configure the instances identically and enable the Send data to PI Archive (/LB) option so that data is written directly to the PI Data Archive. For efficiency, enable the Read before Overwrite /RBO parameter, which prevents an instance from writing an already written value.
You can fin the above information in page 12, section "Fail-Over" from the manual: PI Interface for Universal File and Stream Loading (UFL) (x64) User Guide
Please let us know if you need further help or clarification,
Thanks for the information.
Thanks and Regards,
sorry to interject on your subject, but I am trying the same thing.
In PI ICU, however, when I choose "Send data to PI Archive" it greys out "Read before Overwrite" and vice versa.
Is there a way around this?
Thank you for posting in PI Square , any question is welcomed at any time in the thread.
What you see it’s the correct behavior.
We are mentioning the possibility of using one option or the other. What it is recommended for efficiency in this case is to is to use " Read before Overwrite " /RBO . This mode of operation will do an archive read first (to see if the value exists at the given timestamp) and will send the new value only if it is different. Also, this mode only works when Location5=1 and neither Send data to PI Archive (/LB) nor Laboratory Snapshots /lbs start up parameters are set. The reason is that /lb and /lbs mean sending data in bulks and some events thus may still not be in PI Archives when the reading occurs.