Sending encrypted messages via TCP/IP

[Arslan Khan]

We use standard tcp_acknak.pdl to send encrypted messages internally to other sites/interfaces. Once in a while, we notice duplicates on the receiving end. As these are encrypted messages, I am not sure whether we can use mlp_tcp2.pdl with standard ack/nak tcl procs.

Is there a way we can avoid duplication on the receiving end?

Thanks.

Topic

[Jim Kosloskey]

Do you resend messages upon timeot on the outbound thread?

That may be what is happening.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Arslan Khan]

We do not have any ack/nack enabled on the outbound thread, thus no await replies option activated.

Default entry of “-1” under Retries and Interval of “10” is being used under “Outbound Data” section of “Outbound” tab.

[Jim Kosloskey]

Does your Outbound SMAT show messages being duplicated?

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Arslan Khan]

As these are encrypted messages, we do not enable SMAT on these outbound threads. We do have SMAT on the receiving side which shows the duplicates.

[Jim Kosloskey]

Without getting into a lot of discussion, I am going to suggest you deploy acknowledgment handling with these internal connections. Wiat for a reply before you send the next message.

I don’t think the encryption is the cause of the issue.

We have acknowledgment handling with all of our internal threads (and we have a lot) and have not observed duplication as long as we are configured to exchange acknowledgments properly.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Arslan Khan]

We did enable basic ACK/NACK but were still getting duplicates. Based on the info we have so far, We think that duplication may be happening at the protocol level.

[Jim Kosloskey]

Have you tried using the standard mlp PDL along with recover 33(or 58 or whatever) procs defined on the outbound thread as well as a proc on the inbound thread to generate the ack rather than using a PDL to do the ack?

If you have not had the outbound SMAT on how do you know duplicates are not being generated on the outbound?

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Arslan Khan]

We did enable the standard ACK/NAK procs but still encountered the same duplicate issue.

Also, we did extensive trouble shooting earlier to eliminate the option of client sending the duplicates.

[Jon Melin]

Are you doing any filtering of messages from source to destination, such as suppressing messages on the same route? I’ve had correctly coded filtering, not filter suddenly and start sending duplicates down the route. I’d look at that.

Thank you

[Arslan Khan]

As these are encryted messages, we are not doing any filtering. These are straight raw thru.

[David Barr]

You can’t use MLLP for sending arbitray binary data. It’s possible that this data could contain 0x1C 0x0D in the data, and that is the message termination sequence for MLLP. You could use PROTOCOL:tcpip with length encoding or base64 encode your data before sending it over MLLP.

If you already know that your data can’t contain 0x1C then you can ignore this post.

[Arslan Khan]

We are using PROTOCOL:tcpip with length encoding but still see the duplicates. We’ll try to use base64 encoding and see whether this fixes the issue.

Thanks.

[Author]

Replies