Hmm, I feel your pain but I don't think a Public Forum article will give you the level of attention you deserve.
This needs to be escalated towards Qlik and probably your sales rep is the best channel for this at this point as support apparently has not been able to provide the escalation you need.
I'd recommend just removing this topic, or re-writing in a generic form like:
"Hey folks, every few weeks we are experiencing data corruption on our Kafka targets using Replicate 2021.5
The way the corruption manifests itself it that an additional column that we have added to tag the origin of the data displays as value the name of the target Kafka topic (defined in a Global Rule) in a totally incomprehensible way. "Start processing changes from" is a workaround but we need a real solution,
We have raised this as a support issue(s) to Qlik but without resolution thus far and were wondering whether other might have seen something similar."
Formulated that way I'd answer:
- Was this ever reproducible at will?
- Please consider upgrading to 2022.11 Sp's released on Dec 6th - one never knows... and it gives you a stronger position when escalating.
- Perhaps other can recognize the 'incomprehensible" pattern. Please provide a sample (textual!, or text of dump) of the data which was expected and the data which was received.
- When the corruption happens does it involve a single 'row' / topic or are there multiple occurences and if so, how does that affect the corrupted part.
- WAG - sometimes corruptions occur when all stars aligns in a funny manner. Like a message being exactly 2 million (hex) bytes or 1 million minus 1 or plus 1. Very hard to figure out but something to set up traces/logs for perhaps. Was the payload of the corrupted messages special in any way?
Good luck!
Hein.
