Installing R82.10 on Promox

[PhoneBoy]

When I attempted to install R82.10 on Promox, after First Time Wizard, the VM would boot loop.
By default on my VM, the Processor Type was set to x86-64-v2-AES.
This appeared to work ok on R82, but does not on R82.10, which uses a different Linux kernel. 

In my case, I set the processor type to "host" which resolved my boot loop issue.
As we only support Intel hardware, you might need to change this to something else if your host machine uses AMD 🙂 

Here are the settings I am using for a standalone (management and gateway on same system) VM.

[the_rock]

Good to know!

[Bob_Zimmerman]

Just curious, but does x86-64-v4 work? That would be a better option for AMD users if so.

[_Jelle]

I had exactly the same issue on our Proxmox environment, i can confirm setting the CPU Type to host fixes the boot loop.

CPU installed on our environment:

CPU(s)
72 x Intel(R) Xeon(R) Gold 6254 CPU @ 3.10GHz (2 Sockets)

@Bob_Zimmerman , I can confirm setting the CPU Type to x86-64-v4 with an Intel processor does not fix the boot loop.

[Bob_Zimmerman]

Fascinating! I wonder what is going on that the generic feature flag sets don't work.

[Don_Paterson]

I can't see official support yet for private cloud hypervisors and R82.10.

I added a reply to this related thread about it:

https://community.checkpoint.com/t5/Management/R82-10-ISO-Install-fails-on-VMWare-Vcenter-7/m-p/2667...

[Hugo_vd_Kooij]

The puzzling thing here is. It only applies to gateways.

I first installed a R82.10 MDS server and that one runs on the default CPU settings without an issue.

Then I installed a R82.10 gateway and it ran into issues AFTER I ran the first time wizard. Not before. So something in the First time wizard actually changes the behaviour of the OS in such a way that the boot loop occurs.

[PhoneBoy]

MDS (or even Standalone management) doesn't load the various kernel modules needed for Firewall.
Those modules would be active after FTW.

[Timothy_Hall]

Any chance that the e1000 network driver is in use?  This driver was deprecated in R82.10, and when I tried to use it on an SMS in VMWare it still worked anyway.  But on a gateway e1000 would seem to still work until after the FTW was completed, then it would stop working after the reboot.

[Tom_Kendrick]

Just to add to this, I too had a R82.10 Gateway VM in a reboot loop, post FTW. I tried the CPU fix, but that alone didn't fix it for me.

The cause in my lab seemed to be one of my interfaces was configured with Jumbo Frames - inherited from the parent vmbrX interface. 

After a brief call with one of the Dev's responsible for our DPDK/USIM_x86 code - we changed the interface in the VM in Proxmox to use 1500 MTU and it booted.

Worth checking in your labs too, if you run into the same.

Tom

[Oliver_Fink]

What is this "CPU fix"? I set the CPU to "Host" and could implement ElasticXL with R82.10 in Proxmox as a charm.

[PhoneBoy]

That's the fix (changing the CPU type in Promox/KVM)

[WiliRGasparetto]

Good to know Phoneboy  I'll need to upload a VM like this this week, and this will help me.