We are currently migrating from UNIX to LINUX
We are using MQ channel to transmit messages from LINUX to MF for processing

We have isolated that the issue is a transmission of the message sent from a remote queue in LINUX to a local queue in MF
This we prove by running our application that Puts messages in the queue and viewing the queue in the LOCAL queue in MF
We encountered two situations
1. When two messages were sent one of the message goes to the dead letter queue
2. Some messages are losing a byte when transmitted to the MF Local Queue

The MQ support group mentioned that there is a conversion happening during the transmission that is set. They have changed the character set which enabled all messages to be transmitted but we are still seeing corruption on the message in the MF

Our messages were written and we used the iconv function before putting the message in the MQ Remote queue in LINUX

The messages do not look the same when it hits the MF LOCAL queue

This set-up is also set-up in UNIX and the same messages does not behave the same. No corruption is seen

What things do we look at to ccmpare UNIX and LINUX set-ups on our MQ channel that can achieve successful message transmission. Is that defined by the character set in LINUX MQ set-ups ???

Assuming MF = mainframe (IBM style), then if the MF is running an IBM OS then it uses EBCDIC and UNIX & Linux use ASCII.
I would check at which end translation is taking place; maybe you've accidentally set it to translate twice, once at each end?
Are the cfgs on Unix & Linux exactly the same?

I am asking the MQ support group that set-up the MQ channel to double check the cfgs???

Since I have international data coming in i use the iconv utility in LINUX to translit and used the iso-815 to handle the double byte characters.. do I ask my MQ support to not convert to EBCIDIC in MF in the receiving queue????

I am struggling since my MQ support is not doing some work on what conversions need to happen..

Hope you can assist

Never used MQ & very rarely communicated with an IBM MF; I just know a little bit about the potential issues there.
I would Linux and Unix to have virtually identical cfgs/setups. Check also any pre-processing which may happen on the working Unix system.

thanks chris... would look in that direction