On 08/24/2015 07:44 PM, Falko Trojahn wrote:
Hi all, after upgrading our 3 nodes proxmox 3.4 system (using wheezy) from stock 3.2.0-4-amd64 debian kernel to 3.10.0-11-pve, we have only on one node and only on it's cpu0 approx. 93..100% load constantly with ksoftirqd.
ksoftirq should normally use more than a tiny percent of the cpu, your system is under high interrupt load which causes this behavior.
Often it's related to a situation when a high speed network card receives a very large number of packets in a short time frame.
Driver of your network card misconfigured/messed up? What does:
cat /proc/interrupts
outputs?
As our 3 systems are basically the same (mainboard is Supermicro X8DT3-F, running Intel 82598EB and 82599EB dual network controllers), I'd like to get suggestions how to track down why this is only on one system and perhaps what we can do to prevent this. Thank you very much and best regards Falko _______________________________________________ pve-user mailing list pve-user@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
_______________________________________________ pve-user mailing list pve-user@pve.proxmox.com http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
