5:52 p.m.
I'm drawing blanks when it comes to commit messages today. But this still has
time, because even though the patches are in QEMU upstream, but not a part of a
release yet. However nothing should break if we add support earlier...
Right? O:-)
v2:
- split patches
- add it to guestinfo instead of domstats
- some more macro deduplication
Martin Kletzander (4):
Add load average information type into virDomainGetGuestInfo
qemu_agent: Add qemuAgentGetLoadAvg()
qemu: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
virsh: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
docs/manpages/virsh.rst | 8 ++++-
include/libvirt/libvirt-domain.h | 1 +
src/libvirt-domain.c | 8 +++++
src/qemu/qemu_agent.c | 55 ++++++++++++++++++++++++++++++++
src/qemu/qemu_agent.h | 6 ++++
src/qemu/qemu_driver.c | 21 +++++++++++-
tests/qemuagenttest.c | 53 ++++++++++++++++++++++++++++++
tools/virsh-domain.c | 6 ++++
8 files changed, 156 insertions(+), 2 deletions(-)
--
2.48.1
5:52 p.m.
New subject: [PATCH v2 1/4] Add load average information type into virDomainGetGuestInfo
The public API part.
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
include/libvirt/libvirt-domain.h | 1 +
src/libvirt-domain.c | 8 ++++++++
2 files changed, 9 insertions(+)
diff --git a/include/libvirt/libvirt-domain.h b/include/libvirt/libvirt-domain.h
index f5420bca6e23..2d27f96be94d 100644
--- a/include/libvirt/libvirt-domain.h
+++ b/include/libvirt/libvirt-domain.h
@@ -6455,6 +6455,7 @@ typedef enum {
VIR_DOMAIN_GUEST_INFO_FILESYSTEM = (1 << 4), /* return filesystem information
(Since: 5.7.0) */
VIR_DOMAIN_GUEST_INFO_DISKS = (1 << 5), /* return disks information (Since:
7.0.0) */
VIR_DOMAIN_GUEST_INFO_INTERFACES = (1 << 6), /* return interfaces information
(Since: 7.10.0) */
+ VIR_DOMAIN_GUEST_INFO_LOAD = (1 << 7), /* return load averages (Since: 11.2.0)
*/
} virDomainGuestInfoTypes;
int virDomainGetGuestInfo(virDomainPtr domain,
diff --git a/src/libvirt-domain.c b/src/libvirt-domain.c
index 072cc3225579..ae15ab1e5a94 100644
--- a/src/libvirt-domain.c
+++ b/src/libvirt-domain.c
@@ -13292,6 +13292,14 @@ virDomainSetVcpu(virDomainPtr domain,
* "if.<num>.addr.<num1>.addr" - the IP address of addr
<num1>
* "if.<num>.addr.<num1>.prefix" - the prefix of IP address of
addr <num1>
*
+ * VIR_DOMAIN_GUEST_INFO_LOAD:
+ * Returns load (the number of processes in the runqueue or waiting for disk
+ * I/O) as double values:
+ *
+ * "load.1m" - load averaged over 1 minute
+ * "load.5m" - load averaged over 5 minutes
+ * "load.15m" - load averaged over 15 minutes
+ *
* Using 0 for @types returns all information groups supported by the given
* hypervisor.
*
--
2.48.1
5:52 p.m.
New subject: [PATCH v2 2/4] qemu_agent: Add qemuAgentGetLoadAvg()
With qemu guest agent 9.3 we are able to get the load averages with a
new command.
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
src/qemu/qemu_agent.c | 55 +++++++++++++++++++++++++++++++++++++++++++
src/qemu/qemu_agent.h | 6 +++++
tests/qemuagenttest.c | 53 +++++++++++++++++++++++++++++++++++++++++
3 files changed, 114 insertions(+)
diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c
index 43fca86f10ed..27efb4b389ee 100644
--- a/src/qemu/qemu_agent.c
+++ b/src/qemu/qemu_agent.c
@@ -2568,3 +2568,58 @@ int qemuAgentGetDisks(qemuAgent *agent,
g_free(*disks);
return -1;
}
+
+
+int
+qemuAgentGetLoadAvg(qemuAgent *agent,
+ double *load1m,
+ double *load5m,
+ double *load15m,
+ bool report_unsupported)
+{
+ g_autoptr(virJSONValue) cmd = NULL;
+ g_autoptr(virJSONValue) reply = NULL;
+ virJSONValue *data = NULL;
+ int rc;
+
+ if (load1m)
+ *load1m = 0;
+
+ if (load5m)
+ *load5m = 0;
+
+ if (load15m)
+ *load15m = 0;
+
+ if (!(cmd = qemuAgentMakeCommand("guest-get-load", NULL)))
+ return -1;
+
+ if ((rc = qemuAgentCommandFull(agent, cmd, &reply, agent->timeout,
+ report_unsupported)) < 0)
+ return rc;
+
+ if (!(data = virJSONValueObjectGetObject(reply, "return"))) {
+ virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+ _("qemu agent didn't return an array of loads"));
+ return -1;
+ }
+
+#define GET_NUMBER_PARAM(param_) \
+ do { \
+ if (param_ && \
+ virJSONValueObjectGetNumberDouble(data, #param_, param_) < 0) { \
+ virReportError(VIR_ERR_INTERNAL_ERROR, \
+ _("'%1$s' missing in reply of
guest-get-load"), \
+ #param_); \
+ return -1; \
+ } \
+ } while (0)
+
+ GET_NUMBER_PARAM(load1m);
+ GET_NUMBER_PARAM(load5m);
+ GET_NUMBER_PARAM(load15m);
+
+#undef GET_NUMBER_PARAM
+
+ return 0;
+}
diff --git a/src/qemu/qemu_agent.h b/src/qemu/qemu_agent.h
index f98586e8f8ab..cd17a98d3924 100644
--- a/src/qemu/qemu_agent.h
+++ b/src/qemu/qemu_agent.h
@@ -195,3 +195,9 @@ int qemuAgentSSHRemoveAuthorizedKeys(qemuAgent *agent,
int qemuAgentGetDisks(qemuAgent *mon,
qemuAgentDiskInfo ***disks,
bool report_unsupported);
+
+int qemuAgentGetLoadAvg(qemuAgent *agent,
+ double *load1m,
+ double *load5m,
+ double *load15m,
+ bool report_unsupported);
diff --git a/tests/qemuagenttest.c b/tests/qemuagenttest.c
index 328788024197..566571cf1107 100644
--- a/tests/qemuagenttest.c
+++ b/tests/qemuagenttest.c
@@ -1356,6 +1356,58 @@ testQemuAgentTimezone(const void *data)
virTypedParamsFree(params, nparams);
return ret;
}
+
+
+static const char testQemuAgentGetLoadAvgResponse[] =
+ "{"
+ " \"return\": {"
+ " \"load15m\": 0.03564453125,"
+ " \"load5m\": 0.064453125,"
+ " \"load1m\": 0.00390625"
+ " }"
+ "}";
+
+static int
+testQemuAgentGetLoadAvg(const void *data)
+{
+ virDomainXMLOption *xmlopt = (virDomainXMLOption *)data;
+ g_autoptr(qemuMonitorTest) test = qemuMonitorTestNewAgent(xmlopt);
+ double load1m = 0;
+ double load5m = 0;
+ double load15m = 0;
+
+ if (!test)
+ return -1;
+
+ if (qemuMonitorTestAddAgentSyncResponse(test) < 0)
+ return -1;
+
+ if (qemuMonitorTestAddItem(test, "guest-get-load",
+ testQemuAgentGetLoadAvgResponse) < 0)
+ return -1;
+
+ if (qemuAgentGetLoadAvg(qemuMonitorTestGetAgent(test),
+ &load1m, &load5m, &load15m, true) < 0)
+ return -1;
+
+#define VALIDATE_LOAD(value_, expected_) \
+ do { \
+ if (value_ != expected_) { \
+ virReportError(VIR_ERR_INTERNAL_ERROR, \
+ "Expected " #value_ " '%.11f', got
'%.11f'", \
+ expected_, value_); \
+ return -1; \
+ } \
+ } while (0)
+
+ VALIDATE_LOAD(load1m, 0.00390625);
+ VALIDATE_LOAD(load5m, 0.064453125);
+ VALIDATE_LOAD(load15m, 0.03564453125);
+
+ return 0;
+}
+
+
static int
mymain(void)
{
@@ -1392,6 +1444,7 @@ mymain(void)
DO_TEST(Timezone);
DO_TEST(SSHKeys);
DO_TEST(GetDisks);
+ DO_TEST(GetLoadAvg);
DO_TEST(Timeout); /* Timeout should always be called last */
--
2.48.1
7 p.m.
New subject: [PATCH v2 2/4] qemu_agent: Add qemuAgentGetLoadAvg()
On Tue, Feb 25, 2025 at 15:52:30 +0100, Martin Kletzander wrote:
With qemu guest agent 9.3 we are able to get the load averages with
a
new command.
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
src/qemu/qemu_agent.c | 55 +++++++++++++++++++++++++++++++++++++++++++
src/qemu/qemu_agent.h | 6 +++++
tests/qemuagenttest.c | 53 +++++++++++++++++++++++++++++++++++++++++
3 files changed, 114 insertions(+)
diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c
index 43fca86f10ed..27efb4b389ee 100644
--- a/src/qemu/qemu_agent.c
+++ b/src/qemu/qemu_agent.c
@@ -2568,3 +2568,58 @@ int qemuAgentGetDisks(qemuAgent *agent,
g_free(*disks);
return -1;
}
+
+
+int
+qemuAgentGetLoadAvg(qemuAgent *agent,
+ double *load1m,
+ double *load5m,
+ double *load15m,
+ bool report_unsupported)
+{
+ g_autoptr(virJSONValue) cmd = NULL;
+ g_autoptr(virJSONValue) reply = NULL;
+ virJSONValue *data = NULL;
+ int rc;
+
+ if (load1m)
+ *load1m = 0;
+
+ if (load5m)
+ *load5m = 0;
+
+ if (load15m)
+ *load15m = 0;
Clearing these is pointless ...
+
+ if (!(cmd = qemuAgentMakeCommand("guest-get-load", NULL)))
+ return -1;
+
+ if ((rc = qemuAgentCommandFull(agent, cmd, &reply, agent->timeout,
+ report_unsupported)) < 0)
+ return rc;
+
+ if (!(data = virJSONValueObjectGetObject(reply, "return"))) {
+ virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
+ _("qemu agent didn't return an array of loads"));
+ return -1;
+ }
+
+#define GET_NUMBER_PARAM(param_) \
+ do { \
+ if (param_ && \
+ virJSONValueObjectGetNumberDouble(data, #param_, param_) < 0) { \
... as they are unconditionally overwritten ...
+ virReportError(VIR_ERR_INTERNAL_ERROR, \
+ _("'%1$s' missing in reply of
guest-get-load"), \
How about: parameter '%1%s' missing ...
+ #param_); \
+ return -1; \
... or error is returned.
+ } \
+ } while (0)
+
+ GET_NUMBER_PARAM(load1m);
+ GET_NUMBER_PARAM(load5m);
+ GET_NUMBER_PARAM(load15m);
5:52 p.m.
New subject: [PATCH v2 3/4] qemu: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
src/qemu/qemu_driver.c | 21 ++++++++++++++++++++-
1 file changed, 20 insertions(+), 1 deletion(-)
diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c
index 80c918312b8f..3b8b2c0a1eda 100644
--- a/src/qemu/qemu_driver.c
+++ b/src/qemu/qemu_driver.c
@@ -19181,7 +19181,8 @@ static const unsigned int qemuDomainGetGuestInfoSupportedTypes =
VIR_DOMAIN_GUEST_INFO_HOSTNAME |
VIR_DOMAIN_GUEST_INFO_FILESYSTEM |
VIR_DOMAIN_GUEST_INFO_DISKS |
- VIR_DOMAIN_GUEST_INFO_INTERFACES;
+ VIR_DOMAIN_GUEST_INFO_INTERFACES |
+ VIR_DOMAIN_GUEST_INFO_LOAD;
static int
qemuDomainGetGuestInfoCheckSupport(unsigned int types,
@@ -19468,6 +19469,10 @@ qemuDomainGetGuestInfo(virDomainPtr dom,
qemuAgentDiskInfo **agentdiskinfo = NULL;
virDomainInterfacePtr *ifaces = NULL;
size_t nifaces = 0;
+ double load1m = 0;
+ double load5m = 0;
+ double load15m = 0;
+ bool format_load = false;
size_t i;
virCheckFlags(0, -1);
@@ -19538,6 +19543,14 @@ qemuDomainGetGuestInfo(virDomainPtr dom,
nifaces = rc;
}
+ if (supportedTypes & VIR_DOMAIN_GUEST_INFO_LOAD) {
+ rc = qemuAgentGetLoadAvg(agent, &load1m, &load5m, &load15m,
report_unsupported);
+ if (rc == -1)
+ goto exitagent;
+ if (rc >= 0)
+ format_load = true;
+ }
+
qemuDomainObjExitAgent(vm, agent);
virDomainObjEndAgentJob(vm);
@@ -19564,6 +19577,12 @@ qemuDomainGetGuestInfo(virDomainPtr dom,
virDomainInterfaceFormatParams(ifaces, nifaces, params, nparams,
&maxparams);
}
+ if (format_load) {
+ virTypedParamsAddDouble(params, nparams, &maxparams, "load.1m",
load1m);
+ virTypedParamsAddDouble(params, nparams, &maxparams, "load.5m",
load5m);
+ virTypedParamsAddDouble(params, nparams, &maxparams, "load.15m",
load15m);
+ }
+
ret = 0;
cleanup:
--
2.48.1
5:52 p.m.
New subject: [PATCH v2 4/4] virsh: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
Resolves: https://issues.redhat.com/browse/RHEL-71883
Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com>
---
docs/manpages/virsh.rst | 8 +++++++-
tools/virsh-domain.c | 6 ++++++
2 files changed, 13 insertions(+), 1 deletion(-)
diff --git a/docs/manpages/virsh.rst b/docs/manpages/virsh.rst
index 06c2802b3f9f..f159c40631ec 100644
--- a/docs/manpages/virsh.rst
+++ b/docs/manpages/virsh.rst
@@ -2930,7 +2930,7 @@ Success is always reported in this case.
You can limit the types of information that are returned by specifying one or
more flags. Available information types flags are *--user*, *--os*,
-*--timezone*, *--hostname*, *--filesystem*, *--disk* and *--interface*.
+*--timezone*, *--hostname*, *--filesystem*, *--disk*, *--interface* and *--load*.
If an explicitly requested information type is not supported by the guest agent
at that point, the processes will provide an exit code of 1.
@@ -3009,6 +3009,12 @@ returned:
* ``if.<num>.addr.<num1>.addr`` - the IP address of addr <num1>
* ``if.<num>.addr.<num1>.prefix`` - the prefix of IP address of addr
<num1>
+*--load* returns:
+* ``load.1m`` - average load in guest for last 1 minute
+* ``load.5m`` - average load in guest for last 5 minutes
+* ``load.15m`` - average load in guest for last 15 minutes
+
+
guestvcpus
----------
diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c
index cc5ae6053606..93c34c497133 100644
--- a/tools/virsh-domain.c
+++ b/tools/virsh-domain.c
@@ -13095,6 +13095,10 @@ static const vshCmdOptDef opts_guestinfo[] = {
.type = VSH_OT_BOOL,
.help = N_("report interface information"),
},
+ {.name = "load",
+ .type = VSH_OT_BOOL,
+ .help = N_("report load averages information"),
+ },
{.name = NULL}
};
@@ -13122,6 +13126,8 @@ cmdGuestInfo(vshControl *ctl, const vshCmd *cmd)
types |= VIR_DOMAIN_GUEST_INFO_DISKS;
if (vshCommandOptBool(cmd, "interface"))
types |= VIR_DOMAIN_GUEST_INFO_INTERFACES;
+ if (vshCommandOptBool(cmd, "load"))
+ types |= VIR_DOMAIN_GUEST_INFO_LOAD;
if (!(dom = virshCommandOptDomain(ctl, cmd, NULL)))
return false;
--
2.48.1
7:02 p.m.
New subject: [PATCH v2 0/4] Add support for getting load averages from guest
agent
On Tue, Feb 25, 2025 at 15:52:28 +0100, Martin Kletzander wrote:
I'm drawing blanks when it comes to commit messages today. But
this still has
time, because even though the patches are in QEMU upstream, but not a part of a
release yet. However nothing should break if we add support earlier...
Right? O:-)
No it doesn't qualify to be pushed during freeze ;)
v2:
- split patches
- add it to guestinfo instead of domstats
- some more macro deduplication
Martin Kletzander (4):
Add load average information type into virDomainGetGuestInfo
qemu_agent: Add qemuAgentGetLoadAvg()
qemu: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
virsh: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
Reviewed-by: Peter Krempa <pkrempa(a)redhat.com>
7:17 p.m.
New subject: [PATCH v2 0/4] Add support for getting load averages from guest
agent
On Tue, Feb 25, 2025 at 05:02:35PM +0100, Peter Krempa wrote:
On Tue, Feb 25, 2025 at 15:52:28 +0100, Martin Kletzander wrote:
> I'm drawing blanks when it comes to commit messages today. But this still has
> time, because even though the patches are in QEMU upstream, but not a part of a
> release yet. However nothing should break if we add support earlier...
> Right? O:-)
No it doesn't qualify to be pushed during freeze ;)
I meant earlier as in "before it's released in qemu" not "during
freeze", I even specifically mention 11.2.0 in the docs ;)
>
> v2:
> - split patches
> - add it to guestinfo instead of domstats
> - some more macro deduplication
>
> Martin Kletzander (4):
> Add load average information type into virDomainGetGuestInfo
> qemu_agent: Add qemuAgentGetLoadAvg()
> qemu: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
> virsh: Add support for VIR_DOMAIN_GUEST_INFO_LOAD
Reviewed-by: Peter Krempa <pkrempa(a)redhat.com>
Thanks, I'll fixup what you suggested and wait for after the release.
7:45 p.m.
New subject: [PATCH v2 0/4] Add support for getting load averages from guest
agent
On Tue, Feb 25, 2025 at 17:17:46 +0100, Martin Kletzander wrote:
On Tue, Feb 25, 2025 at 05:02:35PM +0100, Peter Krempa wrote:
> On Tue, Feb 25, 2025 at 15:52:28 +0100, Martin Kletzander wrote:
> > I'm drawing blanks when it comes to commit messages today. But this still
has
> > time, because even though the patches are in QEMU upstream, but not a part of
a
> > release yet. However nothing should break if we add support earlier...
> > Right? O:-)
>
> No it doesn't qualify to be pushed during freeze ;)
>
I meant earlier as in "before it's released in qemu" not "during
freeze", I even specifically mention 11.2.0 in the docs ;)
You see cover-letters are pointless because nobody reads them properly.
Anyways we do integration with pushed-but-not-yet-released qemu stuff
all the time so there's definitely precedent. And stuff gets changed;
libvirt can still adapt it without breaking compatibility.
8:05 p.m.
New subject: [PATCH v2 0/4] Add support for getting load averages from guest
agent
On a Tuesday in 2025, Peter Krempa wrote:
On Tue, Feb 25, 2025 at 17:17:46 +0100, Martin Kletzander wrote:
> On Tue, Feb 25, 2025 at 05:02:35PM +0100, Peter Krempa wrote:
> > On Tue, Feb 25, 2025 at 15:52:28 +0100, Martin Kletzander wrote:
> > > I'm drawing blanks when it comes to commit messages today. But this
still has
> > > time, because even though the patches are in QEMU upstream, but not a part
of a
> > > release yet. However nothing should break if we add support earlier...
> > > Right? O:-)
> >
> > No it doesn't qualify to be pushed during freeze ;)
> >
>
> I meant earlier as in "before it's released in qemu" not "during
> freeze", I even specifically mention 11.2.0 in the docs ;)
You see cover-letters are pointless because nobody reads them properly.
Anyways we do integration with pushed-but-not-yet-released qemu stuff
all the time so there's definitely precedent. And stuff gets changed;
libvirt can still adapt it without breaking compatibility.
I think that only happened in a handful of cases in the past.
Also, there's still almost 5 weeks until the release of 11.2.0 so plenty
of time to change stuff.
Jano
16
days inactive
16
days old
9 comments
3 participants
participants (3)
-
Ján Tomko -
Martin Kletzander -
Peter Krempa