Symptom
IOS-XR kernel panic is seen upon while doing replacing and then rolling back the entire router running configuration and node reloaded on its own.
Conditions
These are the conditions:
1. Cisco IOS XR 64 bit
2. Commit replace of entire config( this removes entire running configuration)
3. Rollback configuration last 1 ( this rollbacks the last config on the node) .
Kernel panic was seen during this trigger and node reloaded on its own.
Workaround
Kernel panic can be disabled by using the following command:
echo 0 > /proc/sys/kernel/hung_task_panic
Further Problem Description
Kernel panic is seen upon doing following steps.
1. commit replace of entire config( this removes entire running configuration)
2. Rollback configuration last 1 ( this rollbacks the last config on the node).
kernel panic was seen during this trigger and node reloaded on its own.
The problem might be occurring because :
One of the IOS-XR applications is sending a message to TCP process. This results in a LWM IPC from client process to TCP. And TCP should normally reply back to client after it services the client request. In this case, for some reason TCP never replies back.
It looks like, the issue happens because when ?commit replace? is done all configuration is deleted and the interface over which the socket connection is established gets deleted. This is current thought on this, the exact root cause is not known yet.
