<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Hello all,<br>
<br>
I was recently trying out NUMA placement for my guests on both
x86_64 and ppc64 machines. When booting a guest on the x86_64
machine, the following specs were valid (obviously, just notable
excepts from the xml):<br>
<div class="de1"><br>
<memory unit='KiB'><span class="nu0">8388608</span></memory><br>
<currentMemory unit='KiB'><span class="nu0">8388608</span></currentMemory></div>
<div class="de2"> <vcpu placement='static'><span class="nu0">4</span></vcpu></div>
...<br>
<cpu><br>
<topology sockets='4' cores='1' threads='1'/><br>
<numa><br>
<cell cpus='0-2' memory='6144'/><br>
<cell cpus='3' memory='2048'/><br>
</numa><br>
</cpu><br>
<br>
However, on ppc64 this causes the following error:<br>
<br>
error: Failed to create domain from sample_guest.xml<br>
error: internal error: early end of file from monitor: possible
problem:<br>
2014-09-11T18:44:25.502140Z qemu-system-ppc64: total memory for NUMA
nodes (8388608) should equal RAM size (200000000)<br>
<br>
The 200000000 is actually 8192 MB in bytes and hexidecimal. This is
apparently just an issue with the error message. <br>
<br>
The following specs work on ppc64:<br>
<cpu><br>
<topology sockets='4' cores='1' threads='1'/><br>
<numa><br>
<cell cpus='0-2' memory='6291456'/><br>
<cell cpus='3' memory='2097152'/><br>
</numa><br>
</cpu><br>
<br>
Note that the memory for each cell is 6144*1024 and 2048*1024
respectively. The issue is that the memory size for each NUMA cell
should be specified in KiB, not MB (<a class="moz-txt-link-freetext"
href="http://libvirt.org/formatdomain.html#resPartition">http://libvirt.org/formatdomain.html#resPartition</a>
"<code>memory</code> specifies the node memory in kibibyte").<br>
<br>
In short, it seems that specifying NUMA cell memory in MB works on
x86_64 but not on ppc64. Does anyone have any insight to what's
causing this, or if I'm misunderstanding something? Any help is
appreciated, thank you!<br>
<br>
Regards,<br>
Michael Turek<br>
</body>
</html>