Announcement Announcement Module
Collapse
No announcement yet.
How to prevent duplicate with normal payload Page Title Module
Move Remove Collapse
X
Conversation Detail Module
Collapse
  • Filter
  • Time
  • Show
Clear All
new posts

  • How to prevent duplicate with normal payload

    I have a system which expose an Web Service. The Web Service define that if the 2 service calls have same checksum (I've used MD5 to calculate checksum of the HTTP Request body), so that's the duplicate method call and should not process.

    What I want is some ways to reject the new incoming message if it already existed in my channel. I've read through the document and only found out that there are prevent duplicates mechanism for things like Feed or File. Please point me if I missed it somewhere.

    Then I looked at the ChannelInterceptor and I think that I can implement the preSend to check the message, however, I worry that after the message passed my preSend, is it immediately store into the MessageStore (eg JDBC)? Say for I have 2 duplicate messages come in a short time, the 2nd message may got pass the preSend does to the 1st message not store into the mentioned MessageStore?

  • #2
    Hello

    In general you should implement Idempotent Receiver.
    Technically there is some component in the Spring Integration which is apropriate for your: Filter.
    Here your need to put several components before your business logic and implement some MessageSelector.
    And also I strongly don't recommend to use MessageStore in application logic. It is framework messaging persistence backend. The interference with his work can be dangerous.
    You said that you 'calculate checksum', so you may use in the MessageSelector#accept some global Set, when you just return from Set#add(checksum).
    Or you can add into your DB some simple table with checksum as PK and catch some ConstraintViolationException or DuplicateKeyException on INSERT call.
    However you should think about some purging logic on your checksum-store to prevent overhead, which can be based on some time-to-live.

    And we also already have some issue about Idempotent Receiver: https://jira.springsource.org/browse/INT-2426

    Is it what you are looking for?

    Artem Bilan
    Last edited by Artem Bilan; Apr 12th, 2012, 02:40 AM.

    Comment


    • #3
      Hi Cleric,

      Thanks for your reply.

      I think the MessageSelector backed by an simple database table is a good idea, cause I have several Web Service server with a front load balancer, so global Set will not be appropriate in this case.

      I will also think about the TTL to purge the checksums.

      Comment

      Working...
      X