Bug 1269715

Libvirtd could possibly reuse the outdated advice received from numad at domain startup. Even though the domain would now require more memory than originally requested, there could be a chance the allocation will succeed. On the other hand, allowing this might trick people into thinking the combination of numad and memory hotplug does anything useful.

(In reply to Ján Tomko from comment #1)
> Libvirtd could possibly reuse the outdated advice received from numad at
> domain startup. Even though the domain would now require more memory than
> originally requested, there could be a chance the allocation will succeed.
> On the other hand, allowing this might trick people into thinking the
> combination of numad and memory hotplug does anything useful.

Hi Jan,

Thanks your quick reply and i found i can hit this error when start a guest, 

1.
# virsh dumpxml rhel7.0-rhel
<domain type='kvm'>
  <name>rhel7.0-rhel</name>
  <uuid>67c7a123-5415-4136-af62-a2ee098ba6cd</uuid>
  <maxMemory slots='16' unit='KiB'>15243264</maxMemory>
  <memory unit='KiB'>1667072</memory>
...
    <memory model='dimm'>
      <target>
        <size unit='KiB'>131072</size>
        <node>0</node>
      </target>
    </memory>
...

2.
# virsh start rhel7.0-rhel
error: Failed to start domain rhel7.0-rhel
error: internal error: Advice from numad is needed in case of automatic numa placement

3. check the debug log:

2015-10-08 09:04:41.822+0000: 16772: debug : virCommandRunAsync:2428 : About to run /bin/numad -w 2:1628

we can found libvirt used 1628M (1667072KiB, and already include memory device), why we forbid start the guest in this case ? and how about the vcpus hot-plug when set vcpu placement is auto and numatune placement is auto (we didn't forbid hot-plug vcpu in that case ) ?

Thanks in advance for your reply.

I'll have a look to see whether it makes sense or how we should forbid using auto placement with resource hotplug.

The original report is not really a bug, since the numad advice is not valid at the point of hotplug. The issue described in Comment 2 is valid though.

Fixed upstream:

commit 25c39f76b80a5453551e8242e99ebc1986ed0d77
Author: Peter Krempa <pkrempa>
Date:   Thu Mar 24 16:05:11 2016 +0100

    qemu: command: Pass numad nodeset when formatting memory devices at boot
    
    When starting up a VM libvirtd asks numad to place the VM in case of
    automatic nodeset. The nodeset would not be passed to the memory device
    formatter and the user would get an error.
    

I might later add a patch that improves the error message in case of the original issue which isn't a bug.

Verify this bug with libvirt-2.0.0-4.el7.x86_64:

1. prepare a guest like this:
...
  <maxMemory slots='16' unit='KiB'>15242882</maxMemory>
  <memory unit='KiB'>1179648</memory>
  <currentMemory unit='KiB'>1179648</currentMemory>
...
  <vcpu placement='auto' current='6'>9</vcpu>
  <numatune>
    <memory mode='strict' placement='auto'/>
  </numatune>
...
  <cpu>
    <numa>
      <cell id='0' cpus='0-2' memory='524288' unit='KiB'/>
      <cell id='1' cpus='3-5' memory='524288' unit='KiB'/>
    </numa>
  </cpu>
...
    <memory model='dimm'>
      <target>
        <size unit='KiB'>131072</size>
        <node>0</node>
      </target>
    </memory>
...

2. start guest
# virsh start r7
Domain r7 started

3. check qemu cmd line

# virsh numatune r7
numa_mode      : strict
numa_nodeset   : 0-1

# ps aux|grep qemu
... -object memory-backend-ram,id=memdimm0,size=134217728,host-nodes=0-1,policy=bind -device pc-dimm,node=0,memdev=memdimm0,id=dimm0
...

4. libvirt still report old error when attach memdevice for a automatic nodeset guest

# cat memdevice.xml 
    <memory model='dimm'>
      <target>
        <size unit='KiB'>131072</size>
        <node>0</node>
      </target>
    </memory>

# virsh attach-device r7 memdevice.xml 
error: Failed to attach device from memdevice.xml
error: internal error: Advice from numad is needed in case of automatic numa placement

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHSA-2016-2577.html