CCR2004: sudden ETHER 9-16 down

I have been using brand new CCR2004-16G-2S+ for about 2 weeks.
ROS 7.8, Firmware 7.8
All 16 ethernet ports are used.
Ether 1 - WAN provider #1 (static)
Ether 9 - WAN provider #2 (dhcp client)
Ether 2-8, 10-16 - local Network

There was strange issue today when ports 9-16 gone down suddenly and then restored up:

 09:48:13 interface,info ether9 link down
 09:48:13 interface,info ether10 link down
 09:48:13 interface,info ether11 link down
 09:48:13 interface,info ether12 link down
 09:48:13 interface,info ether13 link down
 09:48:13 interface,info ether14 link down
 09:48:13 interface,info ether15 link down
 09:48:13 interface,info ether16 link down
 09:48:13 dhcp,info dhcp-client on ether9 lost IP address 85.X.X.X - lease stopped locally
 09:48:15 interface,info ether15 link up (speed 100M, full duplex)
 09:48:16 interface,info ether9 link up (speed 1G, full duplex)
 09:48:16 interface,info ether11 link up (speed 1G, full duplex)
 09:48:16 interface,info ether12 link up (speed 1G, full duplex)
 09:48:16 interface,info ether13 link up (speed 1G, full duplex)
 09:48:16 interface,info ether14 link up (speed 1G, full duplex)
 09:48:16 interface,info ether16 link up (speed 1G, full duplex)

Here is CPU load graph (from Grafana)
ccr2004_cpu_load.jpg
Any version what happened?

Basically, the second 88E6191X switch chip restartet/reset, or the 10G link between it and the CPU was interrupted.
Block diagramm.

Are you using any hardware offloading functions on any of the ports 9 to 16?

I had no experience with hardware offloading yet, so I didn’t change default configuration concerning theses functions.
Just checked second switch settings and I see that it’s L3 hardware offloading is OFF
switch_offloading.png
switch_offloading_by_ports.png
May I get some advice concerning hardware offloading? In what cases should I use it?

L3 hardware offloading is not supported in CCR2004.