none
Too many running instances - Biztalk 2010 RRS feed

  • Question

  • Hello all,

    I have this scenario:

    (In all these steps, the files are moved between several folders.)

    1 - Orchestration retrieves edi attachment from email box (the file is put in a folder)

    2 - For each file , creates a record in a sql table used for monitoring

    3 - Does some validations in an Orchestration, updates status field in the monitoring table

    4 - If successfull, transforms the file in another format and puts the file in a folder

    5 - If validations not sucessfull :

    a : I have a site that reads the monitoring table, users have a button to "RESUME" processing of the file (changes the status field in the monitoring table)

    b: Every 60 minutes, there's an Orchestration that gets the files from the FilesWithError folder and sends only the files that have been RESUMED back to the folder where the Validation Orchestration reads from

    c: Files that have not been RESUMED are put back to the folder FilesWithError 

    My problem is that if I check in the monitoring table for the files that have error states, I get around 3 000 but in the RUNNING SERVICE INSTANCES -> DEHYDRATED ORCHESTRATIONS I have around 40 000 instances.

    So in the FilesWithError folder I was expecting to see around 3000 files but what I see is around 40 000 files.

    If I open some of the files I see that many are refering to information processed long time ago (I match some ids in the files to ids I store in the monitoring table)

    It seems to me that everytime the Orchestration that checks if the files have been RESUMED runs, it creates another instance of the file but when it validates with success it moves only the original file to the Success folder, and the "clones" stay in the FilesWithError folder.

    Another thing is, all the files (the 40 000) have a Modified Date of today, so I don't have any easy way of identifying which are part of the real 3 000 with errors and which are clones, without opening each one to check the values inside and match with the info in the monitoring table.

    Any ideia why the "clones" are created? (I think it's in the 5C step but don't understand why)

    Thanks.



    Tuesday, April 9, 2013 11:40 AM

All replies

  • We faced the same issue, the major problem because of the subscription.

    Need to check the subscription of each orchestration. This might reduce the number of errors.


    Regards, Aboorva Raja R Please remember to mark the replies as answers if they help and unmark them if they provide no help.

    Monday, March 21, 2016 9:39 AM
  • Nelson,

    There's quite a bit going on here. If you suspect something with step 5c), it means there is an implementation logic bug in your flow. Ideally, you should try to replicate this with a single input file in dev/test environment and try and reproduce the error there. If it's not happening, try the same with 5/10 input files, and see if you can catch a logical error with the flow.

    Thanls
    Arindam


    Monday, March 21, 2016 11:05 AM
  • So, I can tell you for sure, the problem is not with BizTalk Server...it's your app :(

    Somewhere in the flow, messages are getting mis-routed/over subscribed.

    While you probably could figure this out in PROD, I think you would find the answer much more quickly if you reproduce in a DEV or TEST kit.  Just pick a few incoming files that are stuck and you can trace them without all the noise.

    One common mistake in such cases is filtering only on MessageType when you really mean MessageType from a specific source.

    Monday, March 21, 2016 12:11 PM
  • Hi Nelson,

    One area I would like to you to check -- what is the retry count on Send Port? Try setting Retry count to zero.

    And do let us know what you discover.

    Monday, March 21, 2016 2:22 PM