On Fri, 20 May 2011 11:42:26 +0100
"Daniel P. Berrange" <berrange(a)redhat.com> wrote:
On Tue, May 17, 2011 at 04:00:07PM +0900, Minoru Usui wrote:
> virNodeGetCPUTimeParameters: Implement virsh support
>
> Signed-off-by: Minoru Usui <usui(a)mxm.nes.nec.co.jp>
> ---
> tools/virsh.c | 107 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
> tools/virsh.pod | 4 ++
> 2 files changed, 111 insertions(+), 0 deletions(-)
>
> diff --git a/tools/virsh.c b/tools/virsh.c
> index a38750f..7c9fa29 100644
> --- a/tools/virsh.c
> +++ b/tools/virsh.c
> @@ -3453,6 +3453,112 @@ cmdNodeinfo(vshControl *ctl, const vshCmd *cmd
ATTRIBUTE_UNUSED)
> }
>
> /*
> + * "nodecputime" command
> + */
> +static const vshCmdInfo info_nodecputime[] = {
> + {"help", N_("Prints cpu utilization of the node.")},
> + {"desc", N_("Returns cpu utilization of the node.(%)")},
> + {NULL, NULL}
> +};
> +
> +static bool
> +cmdNodeCPUTimeParameters(vshControl *ctl, const vshCmd *cmd ATTRIBUTE_UNUSED)
> +{
> + int i, j;
> + bool flag_utilization = false;
> + virCPUTimeParameterPtr params;
> + int nparams = 0;
> + bool ret = false;
> + struct cpu_time {
> + unsigned long long user;
> + unsigned long long sys;
> + unsigned long long idle;
> + unsigned long long iowait;
> + unsigned long long util;
> + } cpu_time[2];
> + double user_time, sys_time, idle_time, iowait_time, total_time;
> + double usage;
> +
> + if (!vshConnectionUsability(ctl, ctl->conn))
> + return false;
> +
> + if (virNodeGetCPUTimeParameters(ctl->conn, NULL, &nparams, 0) != 0) {
> + vshError(ctl, "%s",
> + _("Unable to get number of cpu time parameters"));
> + return false;
> + }
> + if (nparams == 0) {
> + /* nothing to output */
> + return true;
> + }
> +
> + memset(cpu_time, 0, sizeof(struct cpu_time) * 2);
> + params = vshCalloc(ctl, nparams, sizeof(*params));
> + for (i = 0; i < 2; i++) {
> + if (virNodeGetCPUTimeParameters(ctl->conn, params, &nparams, 0) !=
0) {
> + vshError(ctl, "%s", _("Unable to get node cpu time
parameters"));
> + goto cleanup;
> + }
> +
> + for (j = 0; j < nparams; j++) {
> + unsigned long long value = params[j].value;
> +
> + if (strcmp(params[j].field, VIR_CPU_TIME_KERNEL) == 0)
> + cpu_time[i].sys = value;
> +
> + if (strcmp(params[j].field, VIR_CPU_TIME_USER) == 0)
> + cpu_time[i].user = value;
> +
> + if (strcmp(params[j].field, VIR_CPU_TIME_IDLE) == 0)
> + cpu_time[i].idle = value;
> +
> + if (strcmp(params[j].field, VIR_CPU_TIME_IOWAIT) == 0)
> + cpu_time[i].iowait = value;
> +
> + if (strcmp(params[j].field, VIR_CPU_TIME_UTILIZATION) == 0) {
> + cpu_time[i].util = value;
> + flag_utilization = true;
> + break;
> + }
> + }
> +
> + sleep(1);
> + }
> +
> + if (flag_utilization == true) {
> + usage = cpu_time[0].util;
> +
> + vshPrint(ctl, "%-15s %5.1lf%%\n", _("usage:"), usage);
> + vshPrint(ctl, "%-15s %5.1lf%%\n", _("idle :"), 100 -
usage);
> + } else {
> + user_time = cpu_time[1].user - cpu_time[0].user;
> + sys_time = cpu_time[1].sys - cpu_time[0].sys;
> + idle_time = cpu_time[1].idle - cpu_time[0].idle;
> + iowait_time = cpu_time[1].iowait - cpu_time[0].iowait;
> + total_time = user_time + sys_time + idle_time + iowait_time;
> +
> + usage = (user_time + sys_time) / total_time * 100;
> +
> + vshPrint(ctl, "%-15s %5.1lf%%\n",
> + _("usage:"), usage);
> + vshPrint(ctl, "%-15s %5.1lf%%\n",
> + _(" user :"), user_time / total_time * 100);
> + vshPrint(ctl, "%-15s %5.1lf%%\n",
> + _(" system:"), sys_time / total_time * 100);
> + vshPrint(ctl, "%-15s %5.1lf%%\n",
> + _("idle :"), idle_time / total_time * 100);
> + vshPrint(ctl, "%-15s %5.1lf%%\n",
> + _("iowait:"), iowait_time / total_time * 100);
> + }
> +
> + ret = true;
> +
> + cleanup:
> + VIR_FREE(params);
> + return ret;
> +}
I'm not sure this is the best way todo this command. I think we want
a way to see the raw/absolute values without a sleep() in there by
default. Only do the repeated calls+sleep+delta calculation if asked
by some flag
Please see the following thread.
http://www.mail-archive.com/libvir-list@redhat.com/msg36821.html
I received same comment for v2 patch from you,
and I asked you the question but you didn't answer.
Could I ask you a same question?
If we don't care the ESX hypervisor,
I'll change this virsh subcommand prints out raw/absolute values.
But I think it is not legal for libvirt.
> +
> +/*
> * "capabilities" command
> */
> static const vshCmdInfo info_capabilities[] = {
> @@ -10908,6 +11014,7 @@ static const vshCmdDef hostAndHypervisorCmds[] = {
> {"freecell", cmdFreecell, opts_freecell, info_freecell},
> {"hostname", cmdHostname, NULL, info_hostname},
> {"nodeinfo", cmdNodeinfo, NULL, info_nodeinfo},
> + {"nodecputime", cmdNodeCPUTimeParameters, NULL, info_nodecputime},
> {"qemu-monitor-command", cmdQemuMonitorCommand,
opts_qemu_monitor_command, info_qemu_monitor_command},
> {"sysinfo", cmdSysinfo, NULL, info_sysinfo},
> {"uri", cmdURI, NULL, info_uri},
> diff --git a/tools/virsh.pod b/tools/virsh.pod
> index d11a0e3..fe0755f 100644
> --- a/tools/virsh.pod
> +++ b/tools/virsh.pod
> @@ -237,6 +237,10 @@ Print the XML representation of the hypervisor sysinfo, if
available.
> Returns basic information about the node, like number and type of CPU,
> and size of the physical memory.
>
> +=item B<nodecputime>
> +
> +Returns cpu utilization of the node.
> +
> =item B<capabilities>
>
> Print an XML document describing the capabilities of the hypervisor
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|
--
Minoru Usui <usui(a)mxm.nes.nec.co.jp>