Sequel Server question

[MrChuckles]

Does anyone know of a way to monitor for a new file (which will be in TXT format) that will be loaded to the JDE server from our Mainframe and then, subsequently execute (or Trigger) a call to launch a table conversion (TC) UBE to process the in-bound TXT file?

I recognize that this is outside of JDE, but we have no CNC person on staff that can answer this question and I'm just looking for information.

From what I can determine, we may be able to use the 'LaunchBatchApplication' function to start the UBE, but here are my two main question:

1.) How can I monitor for the new TXT file on the JDE server?

and

2.) Once a new file exists on the JDE server, what is required to use the 'LaunchBatchApplication' to start the TC UBE? In other words, how do I go about using 'LaunchBatchApplication'? I would assume the call would be contained within a Windows batch script, but what is the syntax of the script and any parms that are needed? If anybody has a sample code with the call, that would be appreciated.

Last, do I need to change any settings in the JDE.INI file if we use this method?

Thanks for your help.



OneWorld Xe SP23; Sequel Server 2003

 

[Solutionkid]

First, I'm an App guy, not tech; but,
When I need my techs to set up a similar scenario they put a custom UBE into the JDE Scheduler and set the run cycle at about 30 minutes (more often if needed). The UBE runs and looks at the text file for data; if it finds data it executes and clears the text data; if no data it ends. Of course at the appropriate time at night the scheduler stops running so all the regular overnight stuff gets done.
Sems simple to a guy like me.

 

[MrChuckles]

Donald,

Thank you for the response.

I agree that that is the quick and simple way to do it, however, because of security issues with the data contained in the TXT file, we do not want to have that file sitting on the server any longer than absolutely necessary. Thus, we want to invoke the call to the table conversion (TC) UBE immediately after the TXT file has been placed on the JDE server so that the file is processed and cleared ASAP.

I'm trying to track down someone who has some Sequel Server experience that might know of a way to monitor for when the TXT file appears on the JDE server and then cause a 'trigger' to occur so that the table conversion UBE is immediately invoked.

If anyone else has any ideas, I'd appreciate any comments.

Regards

 

[AllisonD]

You probably want a UBE running Async to wake up and check for data at whatever frequency you deem necessary. We have several that run at a very low priority in an iSeries environment and wake up at 5 second intervals. I'm not a developer so I can't speak to the specifics of the Apps but do a search here for 'async' and on the Developers site and you'll likely get a load of information to begin your research from.

 

[altquark]

You could have a Batch file that runs on the machine in question. Because you're a SQL Server user and using Intel Application Servers - you can use some of the sysinternals tools (http://technet.microsoft.com/en-us/sysinternals/default.aspx) from Microsoft and put together a batch script that checks for the existence of the file, and if the file exists - launch runube on the application server and then delete the file in the script.

The advantage of using a batch process instead of a UBE kicking off every 5 seconds is that your F986110 won't be filled up...nor will you impact other JDE jobs (even lower priority jobs impact other jobs !)

The disadvantage is that your batch file will have to have the password probably encoded. However, you could always compile the batch process as an executable - and then use SRVANY to make it into a service !!!

Just some thoughts...

 

[AllisonD]

Thanks Jon,

No matter how hard I try to avoid it, I still manage to learn new things he re. I know this will be subjective but what would be a high record count f or the F986110? Best see what the impact has been since you brought up a p otential problem. And you are right, low level jobs still impact performan ce and other JDE jobs but so far the impact has been very minimal.

From: [email protected] [mailto:[email protected]] On B ehalf Of altquark
Sent: Wednesday, January 14, 2009 2:15 PM
To: [email protected]
Subject: Re: Sequel Server question

You could have a Batch file that runs on the machine in question. Because y ou're a SQL Server user and using Intel Application Servers - you can use s ome of the sysinternals tools (http://technet.microsoft.com/en-us/sysintern als/default.aspx) from Microsoft and put together a batch script that check s for the existence of the file, and if the file exists - launch runube on the application server and then delete the file in the script.

The advantage of using a batch process instead of a UBE kicking off every 5 seconds is that your F986110 won't be filled up...nor will you impact othe r JDE jobs (even lower priority jobs impact other jobs !)

The disadvantage is that your batch file will have to have the password pro bably encoded. However, you could always compile the batch process as an ex ecutable - and then use SRVANY to make it into a service !!!

Just some thoughts...

Jon Steel EnterpriseOne Technical Specialist erpSOURCING LLC http://www.erp sourcing.com [email protected]<mailto:[email protected]> 24/7 Assistance - (904) 382 57 01

 

[altquark]

Re: RE: Sequel Server question

[ QUOTE ]
Thanks Jon,<br><br>No matter how hard I try to avoid it, I still manage to learn new things he re. I know this will be subjective but what would be a high record count f or the F986110? Best see what the impact has been since you brought up a p otential problem. And you are right, low level jobs still impact performan ce and other JDE jobs but so far the impact has been very minimal.

[/ QUOTE ]

Realistically, in my opinion, a high record count for the F986110 is anything that creates issues for whatever platform you're on.

The number of records isn't really the issue - its the number of PDF files stored in the PRINTQUEUE location - but since they're linked, then the number of records can become an issue. The number of records can also be an issue if it impacts the time for users to query the F986110...

I guess I just prefer a nice neat submitted job table !

However, what is really an issue with a job constantly being submitted every 5 seconds isn't necessarily the job itself - its the startup of that job. Remember, each time a UBE is submitted, it has to go through the bootstrap procedure - logging into OneWorld as that user, getting security records, data source information, OCM information - all of that stuff that you see at the beginning of all JDEDEBUG logs. If it then does an "if exist" of a text file, finds it doesn't exist and shuts down - thats a LOT of processing that occurred for no reason. Especially every 5 seconds !!!!

My idea of a batch process to check to see if the file exists is just a different way to do the same process - but the impact on the JDE environment is certainly a lot less. Certainly you're hardcoding somewhat - but some companies have been known to create some sort of DOS based "Crontab" that does something like this based on tables in SQL - a much easier method to manage.

Anyhows - hopefully this works for you...

 

[AllisonD]

RE: RE: Sequel Server question

Thanks Jon,

I'm always open to new ways of looking at things. An old dog needs new tri cks to keep the kibble coming in. I'm not new to JDE per say but I am new to full time CNC. I've danced on the edges as a CNC backup and as the iSer ies Admin, now I do both. I appreciate the mini class and will take a clos er look at the process.

From: [email protected] [mailto:[email protected]] On B ehalf Of altquark
Sent: Wednesday, January 14, 2009 5:13 PM
To: [email protected]
Subject: Re: RE: Sequel Server question

Quote:

 

[jdeKrnl]

Here are my suggestions for a robust solution:

1.) How can I monitor for the new TXT file on the JDE server?

Why rely on pooling when you can build a real-time, notification-based integration?

Take a look at CDirectoryChangeWatcher (http://www.codeproject.com/KB/files/directorychangewatcher.aspx) or "Obtaining Directory Change Notifications" (http://msdn.microsoft.com/en-us/library/aa365261(VS.85).aspx) for applications to monitor Windows directory changes (i.e., monitor creation of new file).

2.) Once a new file exists on the JDE server, what is required to use the 'LaunchBatchApplication' to start the TC UBE?

Just submit a RUNUBEXML doc to the server. Take a look at XML invoker testDriver.cpp that is shipped with E1. You can download it from Doc ID 651936.1 in MetaLink3 or see attachment.

Step 3 (optional): Post final code in download area of JDEList so we can all benefit from it

Cheers,