* [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore
@ 2022-06-01 8:12 Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-cluster 1/1] add pve2-metrics rrd (single metrics) Alexandre Derumier
` (9 more replies)
0 siblings, 10 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
Hi,
I'm still working on vm balancing/scheduling, and need some new metrics.
As suggested by Thomas, instead sending values to new rrd,
and then reparse rrd later in balancer, we broadcast:
- last minute && last 5 min average cpu/ram of qemu/lxc/node
- pressure average of cpu/mem/io for qemu/lxc/node
Theses values are broadcast each minute in 1 kvstore key "balancer-stats"
Size of broadcasted "balancer-stats" is around : 500bytes * number vms.
Size of in memory hash for average values is around : 6k (host+mem) * number vms.
(for last minute, we keep each X seconds iteration, for last 5min we keep last 5x1min average)
- last ksm value
This value is broadcast each iteration is a separate kvstore key "ksm"
pve-common:
Alexandre Derumier (2):
cgroup: get_pressure_stat: use controllers in get_path
cgroup: get_pressure_stat: add cpu full pressure
src/PVE/CGroup.pm | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
qemu-server:
Alexandre Derumier (3):
vmstatus: add hostcpu value
vmstatus: add hostmem value
vmstatus: add pressure stats
PVE/QemuServer.pm | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
pve-container:
Alexandre Derumier (1):
vmstatus: add pressure stats
src/PVE/LXC.pm | 2 ++
1 file changed, 2 insertions(+)
Alexandre Derumier (4):
pvestatd: add broadcast_balancer_stats
pvestatd: qemu/lxc/node : add pressure stats
pvestatd: qemu/lxc/node : add hostcpu/hostmem average stats
pvestatd: node : broadcast ksm value
PVE/Service/pvestatd.pm | 154 +++++++++++++++++++++++++++++++++++++++-
1 file changed, 153 insertions(+), 1 deletion(-)
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-cluster 1/1] add pve2-metrics rrd (single metrics)
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 1/2] cgroup: get_pressure_stat: use controllers in get_path Alexandre Derumier
` (8 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
This create 1 single rrd for each metric
allowed paths:
pve2-metrics/vms/<vmid>/<metric>
pve2-metrics/nodes/<node>/<metric>
pve2-metrics/storages/<node>/<storage>/<metric>
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
data/src/status.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 51 insertions(+)
diff --git a/data/src/status.c b/data/src/status.c
index 9bceaeb..75ab89e 100644
--- a/data/src/status.c
+++ b/data/src/status.c
@@ -1097,6 +1097,23 @@ static const char *rrd_def_storage[] = {
NULL,
};
+static const char *rrd_def_metric[] = {
+ "DS:metric:GAUGE:120:0:U",
+
+ "RRA:AVERAGE:0.5:1:70", // 1 min avg - one hour
+ "RRA:AVERAGE:0.5:30:70", // 30 min avg - one day
+ "RRA:AVERAGE:0.5:180:70", // 3 hour avg - one week
+ "RRA:AVERAGE:0.5:720:70", // 12 hour avg - one month
+ "RRA:AVERAGE:0.5:10080:70", // 7 day avg - ony year
+
+ "RRA:MAX:0.5:1:70", // 1 min max - one hour
+ "RRA:MAX:0.5:30:70", // 30 min max - one day
+ "RRA:MAX:0.5:180:70", // 3 hour max - one week
+ "RRA:MAX:0.5:720:70", // 12 hour max - one month
+ "RRA:MAX:0.5:10080:70", // 7 day max - ony year
+ NULL,
+};
+
#define RRDDIR "/var/lib/rrdcached/db"
static void
@@ -1173,6 +1190,40 @@ update_rrd_data(
create_rrd_file(filename, argcount, rrd_def_node);
}
+ } else if ((strncmp(key, "pve2-metrics/", 13) == 0)) {
+ const char *path = key + 13;
+ char **pa = g_strsplit(path, "/", -1);
+
+ if (!pa[0] || !pa[1] || !pa[2] || strlen(pa[1]) < 1 || strlen(pa[2]) < 1) {
+ goto keyerror;
+ } else {
+ if (strcmp(pa[0], "vms") == 0 || strcmp(pa[0], "nodes") == 0) {
+ if (pa[3]) {
+ goto keyerror;
+ }
+ } else if (strcmp(pa[0], "storages") == 0) {
+ if (pa[4] || !pa[3] || strlen(pa[3]) < 1) {
+ goto keyerror;
+ }
+ } else {
+ goto keyerror;
+ }
+ }
+ g_strfreev(pa);
+
+
+ filename = g_strdup_printf(RRDDIR "/%s", key);
+
+ if (!g_file_test(filename, G_FILE_TEST_EXISTS)) {
+
+ char *dir = g_path_get_dirname(filename);
+ g_mkdir_with_parents(dir, 0755);
+ g_free(dir);
+
+ int argcount = sizeof(rrd_def_metric)/sizeof(void*) - 1;
+ create_rrd_file(filename, argcount, rrd_def_metric);
+ }
+
} else if ((strncmp(key, "pve2-vm/", 8) == 0) ||
(strncmp(key, "pve2.3-vm/", 10) == 0)) {
const char *vmid;
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-common 1/2] cgroup: get_pressure_stat: use controllers in get_path
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-cluster 1/1] add pve2-metrics rrd (single metrics) Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/4] pvestatd: add broadcast_balancer_stats Alexandre Derumier
` (7 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
src/PVE/CGroup.pm | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/PVE/CGroup.pm b/src/PVE/CGroup.pm
index 44b3297..d3873fd 100644
--- a/src/PVE/CGroup.pm
+++ b/src/PVE/CGroup.pm
@@ -380,7 +380,8 @@ sub get_pressure_stat {
},
};
- my ($path, $version) = $self->get_path(undef, 1);
+ my ($path, $version) = $self->get_any_path(1, 'cpu', 'memory', 'io');
+
if (!defined($path)) {
return $res; # container or VM most likely isn't running, retrun zero stats
} elsif ($version == 1) {
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-manager 1/4] pvestatd: add broadcast_balancer_stats
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-cluster 1/1] add pve2-metrics rrd (single metrics) Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 1/2] cgroup: get_pressure_stat: use controllers in get_path Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/3] pvestatd: qemu/lxc/host : broadcast rrd pressure metrics Alexandre Derumier
` (6 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
broadcast each minute as it'll be average stats
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/Service/pvestatd.pm | 24 ++++++++++++++++++++++++
1 file changed, 24 insertions(+)
diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index 72445ec0..984877c1 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -79,6 +79,8 @@ sub hup {
my $cached_kvm_version = '';
my $next_flag_update_time;
my $failed_flag_update_delay_sec = 120;
+my $balancer_stats = {};
+my $last_balancer_broadcast_time = 0;
sub update_supported_cpuflags {
my $kvm_version = PVE::QemuServer::kvm_user_version();
@@ -491,6 +493,20 @@ sub update_sdn_status {
}
}
+sub broadcast_balancer_stats {
+
+ my $ctime = time();
+
+ if($ctime >= $last_balancer_broadcast_time + 60) {
+
+ PVE::Cluster::broadcast_node_kv(
+ 'balancer-stats',
+ encode_json($balancer_stats),
+ );
+ $last_balancer_broadcast_time = $ctime;
+ }
+}
+
my $broadcast_version_info_done = 0;
my sub broadcast_version_info : prototype() {
if (!$broadcast_version_info_done) {
@@ -507,6 +523,8 @@ sub update_status {
# correct list in case of an unexpected crash.
my $rpcenv = PVE::RPCEnvironment::get();
+ $balancer_stats = {};
+
eval {
my $tlist = $rpcenv->active_workers();
PVE::Cluster::broadcast_tasklist($tlist);
@@ -534,6 +552,12 @@ sub update_status {
$err = $@;
syslog('err', "lxc status update error: $err") if $err;
+ eval {
+ broadcast_balancer_stats();
+ };
+ $err = $@;
+ syslog('err', "balancer stats broadcast error: $err") if $err;
+
eval {
rebalance_lxc_containers();
};
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-manager 1/3] pvestatd: qemu/lxc/host : broadcast rrd pressure metrics
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (2 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/4] pvestatd: add broadcast_balancer_stats Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 2/2] cgroup: get_pressure_stat: add cpu full pressure Alexandre Derumier
` (5 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
only "some" values for now, not sure we need full values
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/Service/pvestatd.pm | 35 +++++++++++++++++++++++++++++++++++
1 file changed, 35 insertions(+)
diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index b1e71ec8..832d9dc5 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -132,6 +132,7 @@ sub update_node_status {
my $stat = PVE::ProcFSTools::read_proc_stat();
my $cpuinfo = PVE::ProcFSTools::read_cpuinfo();
my $maxcpu = $cpuinfo->{cpus};
+ my $pressure = PVE::ProcFSTools::read_pressure();
update_supported_cpuflags();
@@ -168,10 +169,13 @@ sub update_node_status {
memory => $meminfo,
blockstat => $dinfo,
nics => $netdev,
+ pressure => $pressure,
};
$node_metric->{cpustat}->@{qw(avg1 avg5 avg15)} = ($avg1, $avg5, $avg15);
$node_metric->{cpustat}->{cpus} = $maxcpu;
+ broadcast_rrd_pressure($ctime, $node_metric, "pve2-metrics/nodes/$nodename");
+
my $transactions = PVE::ExtMetric::transactions_start($status_cfg);
PVE::ExtMetric::update_all($transactions, 'node', $nodename, $node_metric, $ctime);
PVE::ExtMetric::transactions_finish($transactions);
@@ -232,12 +236,41 @@ sub update_qemu_status {
}
PVE::Cluster::broadcast_rrd("pve2.3-vm/$vmid", $data);
+ broadcast_rrd_pressure($ctime, $d, "pve2-metrics/vms/$vmid");
+
PVE::ExtMetric::update_all($transactions, 'qemu', $vmid, $d, $ctime, $nodename);
}
PVE::ExtMetric::transactions_finish($transactions);
}
+sub broadcast_rrd_pressure {
+ my ($ctime, $d, $path) = @_;
+
+ return if !defined($d->{pressure});
+
+ my $pressure = $d->{pressure};
+
+ foreach my $type (keys %{$pressure}) {
+ my $pressuretype = $pressure->{$type};
+
+ foreach my $kind (keys %{$pressuretype}) {
+ next if $kind ne 'some';
+ my $pressurekind = $pressuretype->{$kind};
+
+ foreach my $avg (keys %{$pressurekind}) {
+ next if $avg eq 'total';
+ my $value = $pressurekind->{$avg};
+ my $metric = "pressure_".$type."_".$kind."_".$avg;
+ my $data = $generate_rrd_string->([$ctime, $value]);
+ PVE::Cluster::broadcast_rrd("$path/$metric", $data);
+ $d->{$metric} = $value;
+ }
+ }
+ }
+ delete $d->{pressure};
+}
+
sub remove_stale_lxc_consoles {
my $vmstatus = PVE::LXC::vmstatus();
@@ -441,6 +474,8 @@ sub update_lxc_status {
}
PVE::Cluster::broadcast_rrd("pve2.3-vm/$vmid", $data);
+ broadcast_rrd_pressure($ctime, $d, "pve2-metrics/vms/$vmid");
+
PVE::ExtMetric::update_all($transactions, 'lxc', $vmid, $d, $ctime, $nodename);
}
PVE::ExtMetric::transactions_finish($transactions);
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-common 2/2] cgroup: get_pressure_stat: add cpu full pressure
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (3 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/3] pvestatd: qemu/lxc/host : broadcast rrd pressure metrics Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 2/4] pvestatd: qemu/lxc/node : add pressure stats Alexandre Derumier
` (4 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
available since kernel 5.13
https://lore.kernel.org/all/20210303034659.91735-2-zhouchengming@bytedance.com/T/#u
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
src/PVE/CGroup.pm | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/PVE/CGroup.pm b/src/PVE/CGroup.pm
index d3873fd..bc5b8c8 100644
--- a/src/PVE/CGroup.pm
+++ b/src/PVE/CGroup.pm
@@ -368,7 +368,8 @@ sub get_pressure_stat {
my $res = {
cpu => {
- some => { avg10 => 0, avg60 => 0, avg300 => 0 }
+ some => { avg10 => 0, avg60 => 0, avg300 => 0 },
+ full => { avg10 => 0, avg60 => 0, avg300 => 0 }
},
memory => {
some => { avg10 => 0, avg60 => 0, avg300 => 0 },
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-manager 2/4] pvestatd: qemu/lxc/node : add pressure stats
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (4 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 2/2] cgroup: get_pressure_stat: add cpu full pressure Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value Alexandre Derumier
` (3 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/Service/pvestatd.pm | 34 ++++++++++++++++++++++++++++++++++
1 file changed, 34 insertions(+)
diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index 984877c1..dd578d6b 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -134,6 +134,7 @@ sub update_node_status {
my $stat = PVE::ProcFSTools::read_proc_stat();
my $cpuinfo = PVE::ProcFSTools::read_cpuinfo();
my $maxcpu = $cpuinfo->{cpus};
+ my $pressure = PVE::ProcFSTools::read_pressure();
update_supported_cpuflags();
@@ -170,10 +171,13 @@ sub update_node_status {
memory => $meminfo,
blockstat => $dinfo,
nics => $netdev,
+ pressure => $pressure,
};
$node_metric->{cpustat}->@{qw(avg1 avg5 avg15)} = ($avg1, $avg5, $avg15);
$node_metric->{cpustat}->{cpus} = $maxcpu;
+ compute_pressure($ctime, $node_metric, 'node', $nodename);
+
my $transactions = PVE::ExtMetric::transactions_start($status_cfg);
PVE::ExtMetric::update_all($transactions, 'node', $nodename, $node_metric, $ctime);
PVE::ExtMetric::transactions_finish($transactions);
@@ -227,6 +231,8 @@ sub update_qemu_status {
[$d->{uptime}, $d->{name}, $status, $template, $ctime, $d->{cpus}, $d->{cpu},
$d->{maxmem}, $d->{mem}, $d->{maxdisk}, $d->{disk},
$d->{netin}, $d->{netout}, $d->{diskread}, $d->{diskwrite}]);
+
+ compute_pressure($ctime, $d, 'qemu', $vmid);
} else {
$data = $generate_rrd_string->(
[0, $d->{name}, $status, $template, $ctime, $d->{cpus}, undef,
@@ -436,6 +442,7 @@ sub update_lxc_status {
$d->{maxdisk}, $d->{disk},
$d->{netin}, $d->{netout},
$d->{diskread}, $d->{diskwrite}]);
+ compute_pressure($ctime, $d, 'lxc', $vmid);
} else {
$data = $generate_rrd_string->(
[0, $d->{name}, $d->{status}, $template, $ctime, $d->{cpus}, undef,
@@ -507,6 +514,33 @@ sub broadcast_balancer_stats {
}
}
+sub compute_pressure {
+ my ($ctime, $d, $objecttype, $id) = @_;
+
+ return if !defined($d->{pressure});
+ my $pressure = $d->{pressure};
+
+ foreach my $type (keys %{$pressure}) {
+ my $pressuretype = $pressure->{$type};
+
+ foreach my $kind (keys %{$pressuretype}) {
+ next if $kind ne 'some';
+ my $pressurekind = $pressuretype->{$kind};
+
+ foreach my $avg (keys %{$pressurekind}) {
+ next if $avg eq 'total';
+ my $value = $pressurekind->{$avg};
+ #for externel metric
+ my $metric = "pressure_".$type."_".$kind."_".$avg;
+ $d->{$metric} = $value;
+ next if $avg eq 'avg10';
+ $balancer_stats->{$objecttype}->{$id}->{pressure}->{$type}->{$kind}->{$avg} = $value;
+ }
+ }
+ }
+ delete $d->{pressure};
+}
+
my $broadcast_version_info_done = 0;
my sub broadcast_version_info : prototype() {
if (!$broadcast_version_info_done) {
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (5 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 2/4] pvestatd: qemu/lxc/node : add pressure stats Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 3/4] pvestatd: qemu/lxc/node : add hostcpu/hostmem average stats Alexandre Derumier
` (2 subsequent siblings)
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/QemuServer.pm | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 9441cf2..4fc183e 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -2933,6 +2933,11 @@ sub vmstatus {
if ($pstat->{vsize}) {
$d->{mem} = int(($pstat->{rss}/$pstat->{vsize})*$d->{maxmem});
}
+ if (defined(my $hostmemstat = $cgroups->get_memory_stat())) {
+ $d->{hostmem} = $hostmemstat->{mem};
+ } else {
+ $d->{hostmem} = 0;
+ }
my $old = $last_proc_pid_stat->{$pid};
if (!$old) {
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-manager 3/4] pvestatd: qemu/lxc/node : add hostcpu/hostmem average stats
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (6 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 3/3] vmstatus: add pressure stats Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 4/4] pvestatd: node : broadcast ksm value Alexandre Derumier
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
we aggregate each last X second stats to 1min average
we aggragate each last 5 1min average to 5min average.
vm avgstats are resetted when vm is stopped or removed.
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/Service/pvestatd.pm | 94 ++++++++++++++++++++++++++++++++++++++++-
1 file changed, 93 insertions(+), 1 deletion(-)
diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index dd578d6b..a33b21cd 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -81,6 +81,7 @@ my $next_flag_update_time;
my $failed_flag_update_delay_sec = 120;
my $balancer_stats = {};
my $last_balancer_broadcast_time = 0;
+my $avgstats = {};
sub update_supported_cpuflags {
my $kvm_version = PVE::QemuServer::kvm_user_version();
@@ -177,6 +178,11 @@ sub update_node_status {
$node_metric->{cpustat}->{cpus} = $maxcpu;
compute_pressure($ctime, $node_metric, 'node', $nodename);
+ my $avg_metrics = {
+ cpu => $stat->{cpu},
+ mem => $meminfo->{memused},
+ };
+ compute_avg_metrics($ctime, $avg_metrics, 'node', $nodename);
my $transactions = PVE::ExtMetric::transactions_start($status_cfg);
PVE::ExtMetric::update_all($transactions, 'node', $nodename, $node_metric, $ctime);
@@ -233,6 +239,11 @@ sub update_qemu_status {
$d->{netin}, $d->{netout}, $d->{diskread}, $d->{diskwrite}]);
compute_pressure($ctime, $d, 'qemu', $vmid);
+ my $avg_metrics = {
+ cpu => $d->{hostcpu},
+ mem => $d->{hostmem},
+ };
+ compute_avg_metrics($ctime, $avg_metrics, 'qemu', $vmid);
} else {
$data = $generate_rrd_string->(
[0, $d->{name}, $status, $template, $ctime, $d->{cpus}, undef,
@@ -242,7 +253,7 @@ sub update_qemu_status {
PVE::ExtMetric::update_all($transactions, 'qemu', $vmid, $d, $ctime, $nodename);
}
-
+ delete_old_qemu_avgstats($vmstatus);
PVE::ExtMetric::transactions_finish($transactions);
}
@@ -443,6 +454,11 @@ sub update_lxc_status {
$d->{netin}, $d->{netout},
$d->{diskread}, $d->{diskwrite}]);
compute_pressure($ctime, $d, 'lxc', $vmid);
+ my $avg_metrics = {
+ cpu => $d->{cpu},
+ mem => $d->{mem},
+ };
+ compute_avg_metrics($ctime, $avg_metrics, 'lxc', $vmid);
} else {
$data = $generate_rrd_string->(
[0, $d->{name}, $d->{status}, $template, $ctime, $d->{cpus}, undef,
@@ -452,6 +468,7 @@ sub update_lxc_status {
PVE::ExtMetric::update_all($transactions, 'lxc', $vmid, $d, $ctime, $nodename);
}
+ delete_old_lxc_avgstats($vmstatus);
PVE::ExtMetric::transactions_finish($transactions);
}
@@ -541,6 +558,81 @@ sub compute_pressure {
delete $d->{pressure};
}
+sub compute_avg_metrics {
+ my ($ctime, $avg_metrics, $objectype, $id) = @_;
+
+ foreach my $metric (keys %$avg_metrics) {
+ my $value = $avg_metrics->{$metric};
+ next if !defined($value);
+
+ my $stats = $avgstats->{$objectype}->{$id}->{$metric} || {};
+ $stats->{series}->{60}->{$ctime} = $value;
+ $stats->{avg60} = 0 if !defined($stats->{avg60});
+ $stats->{avg300} = 0 if !defined($stats->{avg300});
+ $stats->{last_compute_time} = $ctime if !defined($stats->{last_compute_time});
+
+ #compute avg each minute
+
+ if($stats->{last_compute_time} && $ctime >= $stats->{last_compute_time} + 60) {
+ $stats->{avg60} = compute_avg($stats->{series}->{60}, $ctime, 60);
+ $stats->{series}->{300}->{$ctime} = $stats->{avg60};
+ $stats->{avg300} = compute_avg($stats->{series}->{300}, $ctime, 300);
+ $stats->{last_compute_time} = $ctime;
+ }
+
+ $balancer_stats->{$objectype}->{$id}->{$metric}->{avg60} = $stats->{avg60};
+ $balancer_stats->{$objectype}->{$id}->{$metric}->{avg300} = $stats->{avg300};
+ $avgstats->{$objectype}->{$id}->{$metric} = $stats;
+ }
+}
+
+sub compute_avg {
+ my ($series, $ctime, $delta) = @_;
+
+ my $total = 0;
+ my $count = 0;
+ my $to_delete;
+
+ foreach my $time (keys %$series) {
+ if ($ctime - $delta >= $time) {
+ $to_delete->{$time} = 1;
+ next;
+ }
+ $count++;
+ $total += $series->{$time};
+ }
+
+ delete %$series{keys %{$to_delete}};
+
+ my $avg = $total / $count if $count > 0;
+ return $avg;
+}
+
+sub delete_old_qemu_avgstats {
+ my ($vmstatus) = @_;
+
+ my $stats = $avgstats->{'qemu'};
+
+ my $to_delete;
+ #delete stats for removed vm, or stopped vm
+ foreach my $vmid (keys %$stats) {
+ $to_delete->{$vmid} = 1 if !defined($vmstatus->{$vmid}) || !$vmstatus->{$vmid}->{pid};
+ }
+ delete %$stats{keys %{$to_delete}};
+}
+
+sub delete_old_lxc_avgstats {
+ my ($vmstatus) = @_;
+
+ my $stats = $avgstats->{'lxc'};
+
+ my $to_delete;
+ #delete stats for removed ct, or stopped ct
+ foreach my $vmid (keys %$stats) {
+ $to_delete->{$vmid} = 1 if !defined($vmstatus->{$vmid}) || $vmstatus->{$vmid}->{status} ne 'running';
+ }
+ delete %$stats{keys %{$to_delete}};
+}
my $broadcast_version_info_done = 0;
my sub broadcast_version_info : prototype() {
if (!$broadcast_version_info_done) {
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH qemu-server 3/3] vmstatus: add pressure stats
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (7 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 3/4] pvestatd: qemu/lxc/node : add hostcpu/hostmem average stats Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 4/4] pvestatd: node : broadcast ksm value Alexandre Derumier
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/QemuServer.pm | 2 ++
1 file changed, 2 insertions(+)
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 4fc183e..09f3a0c 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -2971,6 +2971,8 @@ sub vmstatus {
$d->{cpu} = $old->{cpu};
$d->{hostcpu} = $old->{hostcpu};
}
+
+ $d->{pressure} = $cgroups->get_pressure_stat();
}
return $res if !$full;
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH pve-manager 4/4] pvestatd: node : broadcast ksm value
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
` (8 preceding siblings ...)
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 3/3] vmstatus: add pressure stats Alexandre Derumier
@ 2022-06-01 8:12 ` Alexandre Derumier
9 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-06-01 8:12 UTC (permalink / raw)
To: pve-devel; +Cc: t.lamprecht, Alexandre Derumier
only last value is needed for balancing algorithm
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/Service/pvestatd.pm | 2 ++
1 file changed, 2 insertions(+)
diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
index a33b21cd..dfaef7ce 100755
--- a/PVE/Service/pvestatd.pm
+++ b/PVE/Service/pvestatd.pm
@@ -184,6 +184,8 @@ sub update_node_status {
};
compute_avg_metrics($ctime, $avg_metrics, 'node', $nodename);
+ PVE::Cluster::broadcast_node_kv("ksm", $meminfo->{memshared});
+
my $transactions = PVE::ExtMetric::transactions_start($status_cfg);
PVE::ExtMetric::update_all($transactions, 'node', $nodename, $node_metric, $ctime);
PVE::ExtMetric::transactions_finish($transactions);
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value
2022-05-25 6:52 [pve-devel] [PATCH-SERIES cluster/common/qemu-server/lxc/manager] add new metric stats in single rrd Alexandre Derumier
@ 2022-05-25 6:52 ` Alexandre Derumier
0 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2022-05-25 6:52 UTC (permalink / raw)
To: pve-devel
Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
---
PVE/QemuServer.pm | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 9441cf2..4fc183e 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -2933,6 +2933,11 @@ sub vmstatus {
if ($pstat->{vsize}) {
$d->{mem} = int(($pstat->{rss}/$pstat->{vsize})*$d->{maxmem});
}
+ if (defined(my $hostmemstat = $cgroups->get_memory_stat())) {
+ $d->{hostmem} = $hostmemstat->{mem};
+ } else {
+ $d->{hostmem} = 0;
+ }
my $old = $last_proc_pid_stat->{$pid};
if (!$old) {
--
2.30.2
^ permalink raw reply [flat|nested] 13+ messages in thread
* [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value
2020-11-04 1:41 [pve-devel] [PATCH qemu-server 0/3] vmstatus: add pressure + hostcpu/hostmem Alexandre Derumier
@ 2020-11-04 1:41 ` Alexandre Derumier
0 siblings, 0 replies; 13+ messages in thread
From: Alexandre Derumier @ 2020-11-04 1:41 UTC (permalink / raw)
To: pve-devel
---
PVE/QemuServer.pm | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 0c2b1db..64d3635 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -2658,6 +2658,11 @@ sub vmstatus {
if ($pstat->{vsize}) {
$d->{mem} = int(($pstat->{rss}/$pstat->{vsize})*$d->{maxmem});
}
+ if (defined(my $hostmemstat = $cgroups->get_memory_stat())) {
+ $d->{hostmem} = $hostmemstat->{mem};
+ } else {
+ $d->{hostmem} = 0;
+ }
my $old = $last_proc_pid_stat->{$pid};
if (!$old) {
--
2.20.1
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2022-06-01 8:13 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-06-01 8:12 [pve-devel] [PATCH-SERIES common/qemu-server/lxc/manager] brodcast new metric stats in kvstore Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-cluster 1/1] add pve2-metrics rrd (single metrics) Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 1/2] cgroup: get_pressure_stat: use controllers in get_path Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/4] pvestatd: add broadcast_balancer_stats Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 1/3] pvestatd: qemu/lxc/host : broadcast rrd pressure metrics Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-common 2/2] cgroup: get_pressure_stat: add cpu full pressure Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 2/4] pvestatd: qemu/lxc/node : add pressure stats Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 3/4] pvestatd: qemu/lxc/node : add hostcpu/hostmem average stats Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH qemu-server 3/3] vmstatus: add pressure stats Alexandre Derumier
2022-06-01 8:12 ` [pve-devel] [PATCH pve-manager 4/4] pvestatd: node : broadcast ksm value Alexandre Derumier
-- strict thread matches above, loose matches on Subject: below --
2022-05-25 6:52 [pve-devel] [PATCH-SERIES cluster/common/qemu-server/lxc/manager] add new metric stats in single rrd Alexandre Derumier
2022-05-25 6:52 ` [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value Alexandre Derumier
2020-11-04 1:41 [pve-devel] [PATCH qemu-server 0/3] vmstatus: add pressure + hostcpu/hostmem Alexandre Derumier
2020-11-04 1:41 ` [pve-devel] [PATCH qemu-server 2/3] vmstatus: add hostmem value Alexandre Derumier
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox