We have a C8000 enclosure populated with eight C8220 sleds, each fitted with dual Intel E5-2680 2.7 GHz CPUs and 128 Gb of installed memory. One of these sleds however has always been slow - it takes hours to install any Linux OS onto it from a USB DVD drive when all the others installed quickly and in operation, it is slow to respond to commands or run compute jobs.
Since the original BIOS 1.1.11 was reporting the CPUs were running at 3500MHz instead of 2700, a BIOS update to the latest 2.3 version was done and has fixed this mis-reporting. The server is still slow and looking in the output from dmesg, I found these entries whch differ from all the other sleds:
Hierarchical RCU implementation.
[ 0.000000] RCU dyntick-idle grace-period acceleration is enabled.
[ 0.000000] RCU restricting CPUs from NR_CPUS=256 to nr_cpu_ids=120.
[ 0.000000] Offload RCU callbacks from all CPUs
[ 0.000000] Offload RCU callbacks from CPUs: 0-119.
and:
[ 0.000000] smpboot: Allowing 120 CPUs, 88 hotplug CPUs
[ 0.000000] nr_irqs_gsi: 88
The corresponding output from the other sleds is shown below:
Hierarchical RCU implementation.
[ 0.000000] RCU dyntick-idle grace-period acceleration is enabled.
[ 0.000000] RCU restricting CPUs from NR_CPUS=256 to nr_cpu_ids=128.
[ 0.000000] Offload RCU callbacks from all CPUs
[ 0.000000] Offload RCU callbacks from CPUs: 0-127.
and:
[ 0.000000] smpboot: Allowing 128 CPUs, 96 hotplug CPUs
[ 0.000000] nr_irqs_gsi: 88
Look to me like some CPU cores may be faulty - do i have a defective CPU?
Andy