In the XML parser, a zero in the <shares> element is treated as if it was omitted completely. No value is written to the cgroups fs when the domain is started and the OS default is used. virDomainSetSchedulerParameters treated 0 as a valid value, which got changed to '2' by kernel. Treat 0 as 'not specified' instead, to be consistent with the XML and how other scheduler parameters deal with a 0 value, making it a no-op on live domains. Also clarify the documentation. --- docs/formatdomain.html.in | 2 +- src/lxc/lxc_driver.c | 2 +- src/qemu/qemu_driver.c | 2 +- tools/virsh.pod | 3 ++- 4 files changed, 5 insertions(+), 4 deletions(-) diff --git a/docs/formatdomain.html.in b/docs/formatdomain.html.in index a094524..9749d55 100644 --- a/docs/formatdomain.html.in +++ b/docs/formatdomain.html.in @@ -508,7 +508,7 @@ <dt><code>shares</code></dt> <dd> The optional <code>shares</code> element specifies the proportional - weighted share for the domain. If this is omitted, it defaults to + weighted share for the domain. If this is omitted or 0, it defaults to the OS provided defaults. NB, There is no unit for the value, it's a relative measure based on the setting of other VM, e.g. A VM configured with value diff --git a/src/lxc/lxc_driver.c b/src/lxc/lxc_driver.c index c499182..7ac3691 100644 --- a/src/lxc/lxc_driver.c +++ b/src/lxc/lxc_driver.c @@ -1732,7 +1732,7 @@ lxcDomainSetSchedulerParametersFlags(virDomainPtr dom, virTypedParameterPtr param = ¶ms[i]; if (STREQ(param->field, VIR_DOMAIN_SCHEDULER_CPU_SHARES)) { - if (flags & VIR_DOMAIN_AFFECT_LIVE) { + if (flags & VIR_DOMAIN_AFFECT_LIVE && params[i].value.ul) { if (virCgroupSetCpuShares(priv->cgroup, params[i].value.ul) < 0) goto cleanup; diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 45d11cd..eba1e1a 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -8583,7 +8583,7 @@ qemuDomainSetSchedulerParametersFlags(virDomainPtr dom, value_l = param->value.l; if (STREQ(param->field, VIR_DOMAIN_SCHEDULER_CPU_SHARES)) { - if (flags & VIR_DOMAIN_AFFECT_LIVE) { + if (flags & VIR_DOMAIN_AFFECT_LIVE && value_ul) { if (virCgroupSetCpuShares(priv->cgroup, value_ul) < 0) goto cleanup; vm->def->cputune.shares = value_ul; diff --git a/tools/virsh.pod b/tools/virsh.pod index c6a8be3..092b6ae 100644 --- a/tools/virsh.pod +++ b/tools/virsh.pod @@ -1393,7 +1393,8 @@ If I<--current> is specified, affect the current guest state. B<Note>: The cpu_shares parameter has a valid value range of 0-262144; Negative values are wrapped to positive, and larger values are capped at the maximum. Therefore, -1 is a useful shorthand for 262144. On the Linux kernel, the -values 0 and 1 are automatically converted to a minimal value of 2. +value 1 is automatically converted to a minimal value of 2. The value 0 +is the same as not specifying that parameter. B<Note>: The weight and cap parameters are defined only for the XEN_CREDIT scheduler and are now I<DEPRECATED>. -- 1.8.3.2