Hi,
I was trying to set up a testing environment for some netflow and ids tools on kvm based VMs. The system was running on a vanilla proxmox 1.9 (with the latest updates) installed on a supermicro server with a quad-core cpu and intel e1000 nics. Everything looked good and worked fine until I put "some load" on the nics - around 80-100kpps. The proxmox host ran for a variable amount of time (1-20 minutes) and then crashed reproduceable with a kernel panic taking the VMs with it. After some googling I found some resembling error reports for the e1000 drivers - most problems have been fixed with an upgrade to newer versions of the intel drivers. A vanilla ubuntu 10.04 lts ran absolutely stable in the same environment indicating that they already use fixed drivers. It seems, that the debian kernel on which proxmox is based still has this bug. The betas of proxmox 2.0 are alsp crashing in the test scenario.
Unfortunatly, I don't have any further information on this right now - the test system is already deleted. Maybe a developer could have a look at the currently used drivers in the kernel and check if an update would be possible.
I was trying to set up a testing environment for some netflow and ids tools on kvm based VMs. The system was running on a vanilla proxmox 1.9 (with the latest updates) installed on a supermicro server with a quad-core cpu and intel e1000 nics. Everything looked good and worked fine until I put "some load" on the nics - around 80-100kpps. The proxmox host ran for a variable amount of time (1-20 minutes) and then crashed reproduceable with a kernel panic taking the VMs with it. After some googling I found some resembling error reports for the e1000 drivers - most problems have been fixed with an upgrade to newer versions of the intel drivers. A vanilla ubuntu 10.04 lts ran absolutely stable in the same environment indicating that they already use fixed drivers. It seems, that the debian kernel on which proxmox is based still has this bug. The betas of proxmox 2.0 are alsp crashing in the test scenario.
Unfortunatly, I don't have any further information on this right now - the test system is already deleted. Maybe a developer could have a look at the currently used drivers in the kernel and check if an update would be possible.
