When using AppFlow and HDX Insight we were continuously having issues with our Netscalers. They would crash and reboot causing all connections to be dropped from the primary node. The secondary Netscaler would take over and users would be able to reconnect however with AppFlow enabled failover is not seamless for the users.
Citrix Support looked at our crash dump files and determined we encountered a bug in the firmware.
This appears to be BUG0534821 in which the packet engine is sent a signal 6.
The PE is stuck in a look attempting to process invalid HDX/Insight data from a Receiver client.
The PE (packet engine) gets stuck in a look attempting to process invalid HDX/Insight data from a Receiver client. A recommended workaround for this issue is disabling Appflow globally until the fix is included in an upcoming release which most likely appears to be the next release -10.5-56.12.
(This firmware is now outdated and 10.5 56.22 is the recommended and Safe Harbor firmware that you should be running.)