hi, I noticed today in our poolroom with ~100 diskless clients the exact same behavior. The load was with very few clients over 10> with one CPU core. It is a KVM virtual host, so I thought, I have to give more CPU cores (but wondering why it was working for a long time, in VMware ESX host) for a better load, but searching for the high NFS CPU load ... Our clients needed minutes to boot, instead of 10secs before switching from my own kernel 3.4.35 to the main Debian 3.2.0-4-amd64 kernel. I switched back to my own kernel (I had under Vmware ESX) and added 8 cores to let my studies working again. So, I can absolutely confirm this bug. With 8 cores and my own kernel, I have now a load from 3-5. I will see, if I can test it tomorrow with the default one CPU core and own kernel, how much the load is. cu denny
Attachment:
signature.asc
Description: Message signed with OpenPGP using GPGMail