[SOLVED] Ubuntu VM "problem completing the installation"

[tranermr]

Hello all,

I am attempting to install Ubuntu 20.04 (though I am encountering the same issue on Ubuntu 18.04) on Proxmox 6.2-4, however the installation usually fails partway, and on rare occasions when it does complete the VM hangs on "Booting from Hard Disk"

I've tried peoples' recommendations to chang display to "VMware compatible", and I've also tried a couple different SCSI controllers (LSI 53C895A, VirtIO SCSI). Regardless of the settings I've tried, there is an inconsistent failure during installation, and the installed VM never boots regardless.

It seems like a fair number of people have encountered similar problems - any other suggestions or solutions people have?

Thanks and best!

Mike

 

[tranermr]

Update: I was able to install 18.04 and boot successfully using default BIOS, display, machine, and changing only the SCSI COntroller to the VMware PVSCSI. However, this doesn't seem to replicate for 20.04; I still have installation failures partway through.

 

[oguz]

hi,

can you post the VM configuration? (qm config VMID)

also which ISO are you using for the installation?

 

[tranermr]

Sure! For the 18.04 version that worked: (Notably, it only worked once - I've tried again with the same settings and it fails during the install)
qm config 102
bootdisk: scsi0
cores: 6
ide2: local:iso/ubuntu-18.04.4-live-server-amd64.iso,media=cdrom
memory: 10240
name: SteamHost
net0: virtio=46:99:90:25:95:C9,bridge=vmbr0,firewall=1
numa: 0
ostype: l26
scsi0: VM_SSD:vm-102-disk-0,size=10G
scsihw: pvscsi
smbios1: uuid=d1f964f8-8298-4e55-af3f-0d20cb00020d
sockets: 1
vmgenid: b2b667f7-9016-47dd-a7c1-da685d0bf269

For the 20.04 version that I haven't been able to install on any attempt:
qm config 100
bootdisk: scsi0
cores: 4
ide2: local:iso/ubuntu-20.04-live-server-amd64_1_.iso,media=cdrom
memory: 2048
name: Ubuntu20
net0: virtio=06:5D:76:E3:55:BE,bridge=vmbr1,firewall=1
numa: 0
ostype: l26
scsi0: VM_SSD:vm-100-disk-0,size=20G
scsihw: pvscsi
smbios1: uuid=8e64b729-ff6a-4d82-b9be-13ed29767129
sockets: 1
vga: vmware
vmgenid: f30ffce0-2220-4568-9424-d70925edb759

The ISOs that I am using are the server images downloaded from the Ubuntu website,
ubuntu-18.04.4-live-server-amd64.iso
ubuntu-20.04-live-server-amd64.iso

 

[fireon]

scsihw: pvscsi

This can be an bad idea.
Please use the default "pvscsi". We have here about 60 Proxmoxserver with a lot of Ubuntu's, CentOS, Windows and more running fine.
I have here an 20.04 Lubuntu LXQT Terminalserver with Nomachine Enterprise. Tested with this graphicscards:

• default graphics
• spice
• virtio-gpu
• serial terminal

Also running fine with bios and efi boot.
Maybe this help you to solve the problem

Here is one of my configs:

agent: 1
bios: ovmf
boot: cn
bootdisk: scsi0
cores: 4
efidisk0: SSD-local-zfs:vm-106-disk-1,size=1M
ide2: none,media=cdrom
memory: 4096
name: tuxi
net0: virtio=XXXXXXX,bridge=vmbr0,firewall=1,tag=40
numa: 1
onboot: 1
ostype: l26
scsi0: SSD-local-zfs:vm-106-disk-0,discard=on,size=32G,ssd=1
scsihw: pvscsi
serial0: socket
smbios1: uuid=XXXXXXXXXXXXXXXXXXXXXx
sockets: 1
vga: virtio
vmgenid: XXXXXXXXXXXXXXXXXXXx

1. If this did not help... on what Hardware to you exact have proxmox running?
2. Is Virtualization technology enabled in the Host BIOS?
3. Are running other VM's normaly, other OS's?
4. Post the Journallogs during the VM starts, maybe we can see some issue.

Thanks

 

[tranermr]

Thanks for your feedback and suggestions! I've tried it in the past with the default hwscsi setting (LSI 53C895A for me) but I encounter the same issue. I hadn't tried it with SPICE for a graphics card, so I reset to the default values except for the graphics, but same issue unfortunately:

bootdisk: scsi0
cores: 4
ide2: local:iso/ubuntu-20.04-live-server-amd64_1_.iso,media=cdrom
memory: 2048
name: Ubuntu20
net0: virtio=06:5D:76:E3:55:BE,bridge=vmbr1,firewall=1
numa: 0
ostype: l26
scsi0: VM_SSD:vm-100-disk-0,size=20G
smbios1: uuid=8e64b729-ff6a-4d82-b9be-13ed29767129
sockets: 1
vga: qxl,memory=64
vmgenid: f30ffce0-2220-4568-9424-d70925edb759

1. I'm using two old Dell Optiplex 980s (i7 processor, 16 GBs ram, 2x 1TB 5400 HD and 1x 256 GB SSD)
2. Yep! Intel Virtualization Technology is enabled (although VT for direct IO and Trusted Execution are not - should they be?
3. I've been able to install a copy of turnkey linux, but the test isos I've tried for Manjaro Architect and Fedora have failed, but I don't think in a useful way here (Manjaro can't see the internet, although I can ping google, so the installer won't proceed, and Fedora is unable to load the ISO). I don't have any plans to actually use either one though, so I haven't spent much time debugging them.
4. I'm not sure how to access the logs to which you are referring, but I'm happy to post them! Logs from Proxmox overall, or is there a log attached to the VM?

Thanks for your help!

 

[oguz]

maybe it's just the ISO file. have you verified the checksum to see if it's corrupted?

i can install ubuntu20 here on the latest version without any issues

 

[tranermr]

Yep, the SHA256 hashes for both the 18.04 and 20.04 ISOs match the ones provided by Ubuntu. It must be something with my configuration that is causing the instability during installation, but I'm stumped as to what it could be.

 

[fireon]

direct IO are not - should they be?

Activate. Here on my Supermicro it is enabled.

default hwscsi setting (LSI 53C895A for me)

Looks likes and really old installation with a lot Upgrades. Normaly it should work. I have an customer, they are upgrades over years. Since PVE 1.9

 

[tranermr]

Activate. Here on my Supermicro it is enabled.

Sure enough, that did it! Thanks! And wow - that's a downright historic install!

 

[carsten2]

I have the same problem installing a CentOS based Nethserver VM (https://www.nethserver.org/) on proxmox.
After settup the it hangs at "Booting from Hard Disk..." and consumes 100% of given CPUs.

Version: pve-manager/6.2-6/ee1d7754 (running kernel: 5.4.41-1-pve). Dell T20, all VT-settings enabled.
How to solve?

 

[fireon]

I have the same problem installing a CentOS based Nethserver VM (https://www.nethserver.org/) on proxmox.
After settup the it hangs at "Booting from Hard Disk..." and consumes 100% of given CPUs.

Version: pve-manager/6.2-6/ee1d7754 (running kernel: 5.4.41-1-pve). Dell T20, all VT-settings enabled.
How to solve?

Please open a new topic. This one here is solved.