Fileset Local HELP needed

[Yvonne Gaffney]

I am currently working on a backload process.  I am saving production data which will be sent via the Cloverleaf to another system via a TCP/IP connection.  

Here is my problem.  When reading the data from the file it is loading up as pending on the inbound thread and seems like it is taking forever to move to the outbound thread.  What can I do to speed things up?

The backload data is being saved locally and I have set up the Fileset Local protocol using a Directory Parse.  Under scheduling I have the read interval at 5; max messages at 1; and the scan interval at 30.  The back load messages are HL7 messages that are bing modified using Tcl procs.

Whatever help someone could give would be greatly appreciated.  Thanks.

Topic

[William Rowley]

Increase the Max Messages parameter. What you have now is telling the engine to process 1 message (Max Messages) every 5 seconds (Read Interval).

[Yvonne Gaffney]

That looks like it did some good!  Thanks 😀

[Nora Katz]

How do you determine the max number of messages to set?  5, 10 15, 20?  Is there a limit ?  How do you set the other parameters (scan and read intervals)

Nora

[Michael Lacriola]

Assume you have 100 messages sitting in a directory and no other messages come in for 2 hours in this example.

The read interval is how often to read if there are still messages to read in a directory. The max messages is how many messages to “get” when you are reading, and the scan is how long to wait when they are no more messages to read.

That being said, parms set to 5, 10, and 30 for read, max msgs, and scan interval results in 10 messages every 5 seconds. So if you have 100 messages sitting there, it would take 50 seconds to read all of them to be processed.

Actual setting of the values takes a little tweaking based on how many messages you are receiving and other threads running. Setting it to low will result in a backup of messages for that thread, reading too many at once may choke other thread processes from doing their work.

I usually start of with 5, 5, and 30 and see what there leads. That’s a message a second and then waiting 30 seconds when no more messages to process. It’s easy to tweak because you only need to stop the thread to change the values. you can easily do this in a live environment generally without hurting other processes.

Hope this helps.

[Nora Katz]

Thanks for the explanation.  I’ll try it.  I’m not sure whether it would help me.  What happens in my case is that I have two differnet threads processing different files and one of the files get processed completely and other file does not go to completion.  When I requeue the second file by itself, the file gets processed completely.

Do the parameters in one thread have to be coordinated with the parameters in the other thread?

[Author]

Replies