OneFS 7.2.x: Isilon node could reboots with Fatal double fault in proc: sshd with out recording any stack

OneFS 7.2.x: Isilon node could reboot unexpectedly with Fatal double fault in proc: sshd with out recording any stack when certain rare conditions are met:

The following may be displayed in /var/log/messages:

/boot/kernel.amd64/kernel: panic @ time <TIMESTAMP>, thread 0xffffff015e91a5b0: Fatal double fault
/boot/kernel.amd64/kernel: cpuid = #
/boot/kernel.amd64/kernel: Panic occurred in module kernel loaded at 0xffffffff80200000:
/boot/kernel.amd64/kernel:
/boot/kernel.amd64/kernel: Stack: --------------------------------------------------
/boot/kernel.amd64/kernel: Stack not recorded (starting frame pointer invalid) for proc: sshd (9999) (userspace included: false)

Please note that this issue can only occur when ALL OF THE THREE conditions listed in the cause section of this KB article are met.

This is a rare case scenario which could be hit due to a known issue ONLY WHEN ALL of these three conditions are met:

- The Source Based Routing (SBR) is enabled on the cluster
- And the network in use has network frame Maximum Transfer Unit (MTU) set as smaller than that of the Isilon cluster side.
- And connect via secure shell (SSH) from a client whose Internet Protocol (IP) address and connection is going to make use of the SBR on the cluster.

Work around: 
If all of the three conditions listed in the cause section of this KB article are met & you experience an unexpected node reboot with messages listed in the "Issue" section above, then the recommended work-around is to disable Source Based Routing (SBR) in the cluster until the OneFS Operating System can be upgraded to the recommended code level listed below.

To check whether SBR is enabled:
   # isi networks | grep Source
   Source Based Rout... Enabled

To disable SBR:
   # isi networks sbr disable

Permanent Solution:
Upgrade the Isilon OneFS Operating System to to OneFS 8.0 or later.