VMware Cloud Community
jneedham83
Contributor
Contributor
Jump to solution

Creating Guests on Ampere HR330A via Fusion Pro 12 causes kickstart to hang

I don't know if anybody even does this, but I ran into this very weird problem with the Fusion12 Pro M1 version a while back and just verified that the x64 version of FP12Pro also creates guests that hang when kicked.

I am kicking ROCKY87, but it first occurred with CENTOS84 guests creating by FusionPro12/arm64e. so I wrote if off as a port-specific bug as FP12Pro/arm64e port could still have bugs.

The boot will hang right after 'Starting Basic System' and dracut initqueue hook, but those are just the last messages encountered.

It might also seem like it is hung doing "Started cancel waiting for multipath siblings of sda", as that can be the "last" message encountered. Setting nompath kernel argument simply removes this message  

It is definitely kick related. If I boot from the ISO that is vmfs-local, no problem. This is the ISO i use to create the /var/lib/tftpboot/directories where vmlinuz and initrd.img originate when pulled via tftp. 

The boot will hang right after 'Starting Basic System' and probably at or near dracut/initqueue, but that is a guess. It might also seem like it is hung doing dmpath "cancel path on sibling of sda", as that can be the "last" message encountered.

 

Tags (1)
Reply
0 Kudos
1 Solution

Accepted Solutions
jneedham83
Contributor
Contributor
Jump to solution

 

The issue is that FPro12 creates the NIC with adapter-type VMXNET3.

When switched to e1000e, the kickstart install proceeds.  This might be specific to this hardware platform

I had to use the web console to edit the adapter-type as FPro12 doesn't support this degree of editing of an ESXi guest.

I suspect that most don't use FPro12 to create and edit guests on an ESX server (use either the Web console or vCenter), but using FPro12 has been easy and usually works well with only a couple of servers.

 

 

View solution in original post

Reply
0 Kudos
3 Replies
cyprienlaplace
VMware Employee
VMware Employee
Jump to solution

Hi, this sounds like an interesting issue, and I have a lot of questions.

The first one would be the configuration of the VM(s). I would imagine you are using the very same virtual hardware configuration when you boot the ISO from the vmfs-local?

Then have you tried the same kick boot from a VM on Fusion12 Pro M1? The virtualization support is quite different between the fling and Fusion, so I am curious if that changes anything.

And then, I have no idea what linux is doing/waiting when it stalls. And I don't know those distros well enough to be able to help finding out 😕

Reply
0 Kudos
jneedham83
Contributor
Contributor
Jump to solution

 

The issue is that FPro12 creates the NIC with adapter-type VMXNET3.

When switched to e1000e, the kickstart install proceeds.  This might be specific to this hardware platform

I had to use the web console to edit the adapter-type as FPro12 doesn't support this degree of editing of an ESXi guest.

I suspect that most don't use FPro12 to create and edit guests on an ESX server (use either the Web console or vCenter), but using FPro12 has been easy and usually works well with only a couple of servers.

 

 

Reply
0 Kudos
jneedham83
Contributor
Contributor
Jump to solution

Also, make sure the USB controller is defined as USB3.1, not USB2 or the console window will not respond. 

In Summary: e1000e and USB3.1 are things to verify after you have created the guest with FPro12 (in case you do this on an ARM ESXi Fling server). A lot of if, if, if, so a corner case

 

Reply
0 Kudos