Proxmox dead?

[TinusBlaauw]

Hi all,

I have just upgraded to new hardware and thought I would see what it can do...

I restored by 7 backed up VM's, and decided to click on 'start' for each individual VM, without waiting for the previous one to finish starting up.

BIG MISTAKE. Result is I lost connection to :8006; and even with a reboot, it appeared to have been damaged.

Only solution was then to do a complete reinstall of the proxmox.

Has anybody else experienced this?

Tinus

 

[LnxBil]

More information (especially corresponding log entries, real screen shots for kernel panics and such) would help.

Try to check if you enabled virtualization in your bios, sometimes this is deactivated by default

 

[TinusBlaauw]

Hi

Virtualization is enabled in the bios. I can't access the webgui; neither SSH. If I go to the machine and try and log in via the cli; it takes AGES.

I have a cloned VM, and seems it is the culprit. However, I can not reinstall the system, as I have my email server on it too, and the backup that I have is now old (albeit only a few days)...

THIS IS SO FRUSTRATING!

 

[LnxBil]

But it'll login eventually? Try w and top and see what is going on.

Do you have a support contract? This would come in handy now :-D

 

[TinusBlaauw]

Sent from my SM-G900F using Tapatalk

 

[TinusBlaauw]

No, doesn't even log in....

 

[TinusBlaauw]

OK, so now the webgui finally started responding - except it does not accept 'root' and the password anymore?!

 

[TinusBlaauw]

Progress ---- pulled the network cables - suddenly I can log in from the console without a problem.

So where to from here?????

 

[LnxBil]

First, run free and then stop all VMs (if running) and rerun free. Your message shows that the swapper is taking too long.

Next thing would be

dmesg | grep -i error

 

[TinusBlaauw]

Run free?

Sent from my SM-G900F using Tapatalk

 

[TinusBlaauw]

Before after. Log

Sent from my SM-G900F using Tapatalk

 

[TinusBlaauw]

Sent from my SM-G900F using Tapatalk

 

[kobuki]

You have allocated 24 GB of ram for KVM VMs, while you only have 16 GB in your HN. It's not going to work. Well, I might, for some time, with extensive swapping which in turn will render the server unusable. Revise memory allocation for your VMs. No wonder you run into serious problems. Leave at least a GB or two for the OS itself.

 

[TinusBlaauw]

You have allocated 24 GB of ram for KVM VMs, while you only have 16 GB in your HN. It's not going to work. Well, I might, for some time, with extensive swapping which in turn will render the server unusable. Revise memory allocation for your VMs. No wonder you run into serious problems. Leave at least a GB or two for the OS itself.

I used to run esxi and this was not a problem, as it dynamically changed the ram assigned, etc.... So this is not the case then with proxmox???

Sent from my SM-G900F using Tapatalk

 

[kobuki]

It's also possible with Proxmox. Did you setup dynamic memory allocation for the VMs in question when creating them (you can change it later)? Even if you did, what do you think will happen when all VMs start to consume all of their RAM?

 

[TinusBlaauw]

It's also possible with Proxmox. Did you setup dynamic memory allocation for the VMs in question when creating them (you can change it later)? Even if you did, what do you think will happen when all VMs start to consume all of their RAM?

I am not sure. Is it somewhere in the menus? It seems the problem is always arising on a cloned vm made from a freebsd 10 image...

Sent from my SM-G900F using Tapatalk

 

[udo]

Hi,
simply edit your VM-configs (like "vi /etc/pve/qemu-server/104.conf") and remove the line "onboot: 1".

Udo

 

[m.ardito]

I used to run esxi and this was not a problem, as it dynamically changed the ram assigned, etc.... So this is not the case then with proxmox???

proxmox (really kvm under proxmox) can use KSM an balooning. They must be enabled for the vm and balooning supported by drivers into the vm.
see https://pve.proxmox.com/wiki/KSM

but those are tools to let VMs share in some way the same physical ram.
Your pve host needs its own, for many reasons, and it's better if it can count on it to be always available.
So reserve some GB for pve node, as kobuki said, and configure the other VM to share the remaining.

KSM is powerful, and VM needs not to be aware of it: if most VMs use many identical ram pages (ie: if they are running very similar OS versions and configurations), but it takes a while to guess what can be shared..
so, start VM sharing ram with delays or, at node start they will claim all configured ram all together...

Marco