Proxmox v4 & v5 incompatible with HP Microserver GEN8

[emanuelebruno]

Hi all,
in my previous post I said that changing CPU to the HP Microserver (from G1610T to E3-1230V2 or anything else) meets into rebooting issue.
It is wrong! I managed to reproduce the problem by doing so:
I created a virtual machine with windows 7, and I've created two virtual machines that start from a "clonezilla" iso: the first reads the disk to make a copy via lan (disk to disk); the second receives data via lan and writes to disk.

After about 3 minutes the server restarts suddenly.

The server doesn't reboot suddendly if the virtual machine with Windows 7 stay off.

I hope I've been clear and you can understand how to solve the problem.

Thank you,
E. Bruno.

 

[morph027]

Na, might be something different. I'm running 2 Microservers without any issues.

 

[emanuelebruno]

Na, might be something different. I'm running 2 Microservers without any issues.

The fact that it never happened does not mean that your servers are not affected by the same problem

 

[morph027]

Sure

Any chance to get the kernel logs when this is happening? dmesg? journalctl?

 

[sigxcpu]

I am running a gen8 microserver using Proxmox 3 and then 4 for two years now. CPU is E3-1265L V2. In the past I've had VMs on Windows 7 & Server 2012R2. Never had an issue.

 

[LnxBil]

Best would be a crash dump. I'm unfamiliar with HP microservers, but do they also have iLO? What about the event log there? Maybe a NMI or some machine exception?

 

[emanuelebruno]

LnxBil and morph027 are right! This problem needs more investigation (perhaps it doesn't happen only with HP microserver) but I don't know how to debug... can you give instructions?

 

[vooze]

Are you sure you don't have watchdog enabled in BIOS?

 

[emanuelebruno]

I don't know how to disalbe wathdog...
In this video I show Proxmox rebooting problem... if you want to know if your server is affected by same problem, why don't you try it?

 

[dcsapak]

what is your

pveversion -v

 

[morph027]

Also, if you're using E3-1230V2, which has TDP of 69 W (Microserver only can handle a TDP of 35W), the whole system might just get too hot and triggers a NMI. You can watch your ILO for info on sensors.

 

[emanuelebruno]

what is your

pveversion -v

I tested with Proxmox VE 5.0 beta1.ISO and Proxmox VE 4.4-eb2d6f1e-2.ISO (no "apt-get update && upgrade" was done...)

 

[emanuelebruno]

Also, if you're using E3-1230V2, which has TDP of 69 W (Microserver only can handle a TDP of 35W), the whole system might just get too hot and triggers a NMI. You can watch your ILO for info on sensors.

The HP Microserver that you see in the video is a new one with the original G1610T CPU... (no CPU upgrade was done...)

 

[emanuelebruno]

I am running a gen8 microserver using Proxmox 3 and then 4 for two years now. CPU is E3-1265L V2. In the past I've had VMs on Windows 7 & Server 2012R2. Never had an issue.

I have found a particular situation in wich the server Proxmox v4 & v5 crashes "always"... it means that there is a stability problem (kernel ? bios ? network ? hard disk access?)

I can't stay calm if I think that I server can reboot under stress condition.

It is vital that virtualization software (PROXMOX) is a rock, otherwise all active virtual machines could be compromised and this should not happen

 

[morph027]

Thats true. Really weird issue anyway. I'm regularily putting decent load at my lap server and did not observe any crash like this. What are the last messages in journalctl -lf when the reboot occurs?

 

[dcsapak]

do you use zfs? how much ram do you have and how much ram did you assign to vms ?

 

[emanuelebruno]

do you use zfs? how much ram do you have and how much ram did you assign to vms ?

Hi dcsapak, the server is 4 GB, Windows 7 has 1 GB, the second and the third kvm have 512 MB; filesystem zfs only one drive (4 TB) .

 

[emanuelebruno]

Thats true. Really weird issue anyway. I'm regularily putting decent load at my lap server and did not observe any crash like this. What are the last messages in journalctl -lf when the reboot occurs?

How can I see these messages?

 

[dcsapak]

Hi dcsapak, the server is 4 GB, Windows 7 has 1 GB, the second and the third kvm have 512 MB; filesystem zfs only one drive (4 TB) .

I guess this too little, as far as i remember, zfs has a default minimum arc setting of 3GB,
also if you have swap also on zfs, it can happen that your system has high memory pressure -> tries to swap (on zfs) -> increases the arc -> higher memory pressure -> etc..

so i would either limit the arc usage further down(i do not recommend this solution) (see https://pve.proxmox.com/wiki/ZFS_on_Linux#_limit_zfs_memory_usage)
use less ram in your vms, or upgrade the host

 

[LnxBil]

Again: configure crashdump http://lmgtfy.com/?q=jessie+crash+dump