NUMA Placement Failed, Performance Might Be Affected

Home » CentOS-Virt » NUMA Placement Failed, Performance Might Be Affected
CentOS-Virt 2 Comments

I just did a yum update on a CentOS 7 / Xen 4.6 server which took me from kernel-3.18.34-20.el7.x86_64 -> kernel-3.18.44-20.el7.x86_64

After rebooting, the following notice is printed immediately upon xl create’ing a domain: libxl: notice: libxl_numa.c:499:libxl__get_numa_candidate: NUMA placement failed, performance might be affected

Indeed performance is significantly degraded. This issue didn’t happen on the older kernel. Any ideas where it came from or how it can be fixed?

# cat /proc/cpuinfo 
processor : 0
vendor_id : GenuineIntel cpu family : 6
model : 63
model name : Intel(R) Xeon(R) CPU E5-2637 v3 @ 3.50GHz stepping : 2
microcode : 0x36
cpu MHz : 3500.090
cache size : 15360 KB
physical id : 0
siblings : 1
core id : 0
cpu cores : 1
apicid : 0
initial apicid : 0
fpu : yes fpu_exception : yes cpuid level : 15
wp : yes flags : fpu de tsc msr pae mce cx8 apic sep mca cmov pat clflush acpi mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc arch_perfmon rep_good nopl nonstop_tsc eagerfpu pni pclmulqdq monitor est ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm ida arat epb pln pts dtherm fsgsbase bmi1 avx2 bmi2 erms xsaveopt bugs :
bogomips : 7000.18
clflush size : 64
cache_alignment : 64
address sizes : 46 bits physical, 48 bits virtual power management:

# xl info host                   : ****.localdomain release                : 3.18.44-20.el7.x86_64
version                : #1 SMP Tue Oct 25 11:42:52 UTC 2016
machine                : x86_64
nr_cpus                : 1
max_cpu_id             : 15
nr_nodes               : 2
cores_per_socket       : 1
threads_per_core       : 1
cpu_mhz                : 3500
hw_caps                : bfebfbff:2c100800:00000000:00007f00:77fefbff:00000000:00000021:000037ab virt_caps              : hvm hvm_directio total_memory           : 130958
free_memory            : 28003
sharing_freed_memory   : 0
sharing_used_memory    : 0
outstanding_claims     : 0
free_cpus              : 0
xen_major              : 4
xen_minor              : 6
xen_extra              : .3-7.el7
xen_version            : 4.6.3-7.el7
xen_caps               : xen-3.0-x86_64 xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p hvm-3.0-x86_64 
xen_scheduler          : credit xen_pagesize           : 4096
platform_params        : virt_start=0xffff800000000000
xen_changeset          : Wed Feb 15 02:12:22 2017 -0600 git:c359c64-dirty xen_commandline        : placeholder dom0_mem=1024M,max:1024M cpuinfo com1=115200,8n1 console=com1,tty loglvl=all guest_loglvl=all cc_compiler            : gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-11)
cc_compile_by          : mockbuild cc_compile_domain      : CentOS.org cc_compile_date        : Wed Feb 15 12:18:43 UTC 2017
xend_config_format     : 4

Thanks in advance,

Chris

2 thoughts on - NUMA Placement Failed, Performance Might Be Affected

  • Solved.

    For the archives:

    Noticed in xl info output that only 1 core was recognised, even though nr_cpus should show 16 on this box.

    Tried rebooting and selecting the old .34 kernel, console printed “smpboot: do_boot_cpu failed(-1) to wakeup CPU#1” and CentOS failed to boot. Powered down the box fully and started back up with the latest .44 kernel and everything is working fine now.

    Not enough of an expert to know if this is a bug in the kernel, a problem with the BIOS, bad hardware, a completely random stroke of bad luck or a cosmic sign. Happy it works now. If anyone has experienced this and has any suggestions they would be received with gratitude.

    Chris

  • Chris,

    Check your cpuinfo from flag which did not indicate any ‘VMX’ support for your CPU virtualization support on the board.

    Xlord

    From: CentOS-virt [mailto:CentOS-virt-bounces@CentOS.org] On Behalf Of Chris Wik Sent: Wednesday, March 1, 2017 7:07 AM
    To: Discussion about the virtualization on CentOS
    Subject: Re: [CentOS-virt] NUMA placement failed, performance might be affected

    Solved.

    For the archives:

    Noticed in xl info output that only 1 core was recognised, even though nr_cpus should show 16 on this box.

    Tried rebooting and selecting the old .34 kernel, console printed “smpboot: do_boot_cpu failed(-1) to wakeup CPU#1” and CentOS failed to boot. Powered down the box fully and started back up with the latest .44 kernel and everything is working fine now.

    Not enough of an expert to know if this is a bug in the kernel, a problem with the BIOS, bad hardware, a completely random stroke of bad luck or a cosmic sign. Happy it works now. If anyone has experienced this and has any suggestions they would be received with gratitude.

    Chris

LEAVE A COMMENT