def api_call_latency(title, metric, verb, scope, threshold):
return d.Graph(
title=title,
targets=[
d.Target(expr=str(threshold), legendFormat="threshold"),
d.Target(
expr='quantile_over_time(0.99, %(metric)s{quantile="0.99", verb=~"%(verb)s", scope=~"%(scope)s"}[12h])'
% {"metric": metric, "verb": verb, "scope": scope}
),
],
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
)
def api_call_latency(title, verb, scope, threshold):
return d.Graph(
title=title,
targets=[
g.Target(expr=str(threshold), legendFormat="threshold"),
g.Target(
expr='apiserver:apiserver_request_latency_1m:histogram_quantile{quantile="0.99", verb=~"%(verb)s", scope=~"%(scope)s"}'
% {"verb": verb, "scope": scope},
legendFormat="{{verb}} {{scope}}/{{resource}}",
),
],
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
)
def api_call_latency(title, verb, scope, threshold):
return d.Graph(
title=title,
targets=[
g.Target(expr=str(threshold), legendFormat="threshold"),
g.Target(
expr=d.one_line(expression % {
"verb": verb,
"scope": scope
}),
# TODO(github.com/grafana/grafana/issues/19410): uncomment once fixed
# legendFormat="{{verb}} {{scope}}/{{resource}}",
),
],
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
)
def api_call_latency(title, verb, scope, threshold):
return d.Graph(
title=title,
targets=[
g.Target(expr=str(threshold), legendFormat="threshold"),
g.Target(
expr=d.one_line("""
apiserver:apiserver_request_latency_1m:histogram_quantile{
quantile="0.99",
verb=~"%(verb)s",
scope=~"%(scope)s",
resource=~"${resource:regex}s*",
}""" % {
"verb": verb,
"scope": scope
}),
# TODO(github.com/grafana/grafana/issues/19410): uncomment once fixed
# legendFormat="{{verb}} {{scope}}/{{resource}}",
),
],
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
)
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from grafanalib import core as g
import defaults as d
DNS_LATENCY_PANEL = [
d.Graph(
title="In-cluster DNS latency SLI",
targets=d.show_quantiles(
("quantile_over_time(" + "0.99, " +
'probes:dns_lookup_latency:histogram_quantile{{quantile="{quantile}"}}[24h])'
),
legend="{{quantile}}",
),
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
nullPointMode="null",
),
d.Graph(
title="DNS latency",
targets=d.show_quantiles(
'probes:dns_lookup_latency:histogram_quantile{{quantile="{quantile}"}}',
legend="{{quantile}}",
),
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
nullPointMode="null",
),
d.Graph(
d.simple_graph(
"etcd_disk_backend_commit_duration_seconds",
"histogram_quantile(0.99, sum(rate(etcd_disk_backend_commit_duration_seconds[1m])) by (le, instance))",
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
),
d.simple_graph(
"etcd wal fsync duration",
"histogram_quantile(1.0, sum(rate(etcd_disk_wal_fsync_duration_seconds_bucket[1m])) by (le, endpoint))",
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
),
d.Graph(
title="etcd compaction max pause",
points=True,
lines=False,
targets=[
g.Target(
expr=
"histogram_quantile(1.0, sum(rate(etcd_debugging_mvcc_db_compaction_pause_duration_milliseconds_bucket[1m])) by (le, instance))"
)
],
yAxes=g.single_y_axis(format=g.MILLISECONDS_FORMAT),
),
d.simple_graph(
"etcd objects",
"sum(etcd_object_counts) by (resource, instance)",
legend="{{instance}}: {{resource}}",
),
d.simple_graph(
"etcd db size",
[
"etcd_mvcc_db_total_size_in_bytes",
"etcd_mvcc_db_total_size_in_use_in_bytes",
),
d.simple_graph(
"etcd compaction pause num chunks",
"delta(etcd_debugging_mvcc_db_compaction_pause_duration_milliseconds_count[1m])",
),
d.simple_graph(
"etcd_disk_backend_commit_duration_seconds",
"histogram_quantile(0.99, sum(rate(etcd_disk_backend_commit_duration_seconds[1m])) by (le, instance))",
g.single_y_axis(format=g.SECONDS_FORMAT),
),
d.Graph(
title="etcd compaction max pause",
points=True,
lines=False,
targets=[
g.Target(
expr=
"histogram_quantile(1.0, sum(rate(etcd_debugging_mvcc_db_compaction_pause_duration_milliseconds_bucket[1m])) by (le, instance))"
)
],
yAxes=g.single_y_axis(format=g.MILLISECONDS_FORMAT),
),
d.simple_graph("etcd objects",
"sum(etcd_object_counts) by (resource, instance)"),
d.simple_graph(
"etcd db size",
[
"etcd_mvcc_db_total_size_in_bytes",
"etcd_mvcc_db_total_size_in_use_in_bytes",
"etcd_server_quota_backend_bytes",
],
g.single_y_axis(format=g.BYTES_FORMAT),
from grafanalib import core as g
import defaults as d
NETWORK_PROGRAMMING_PANEL = [
d.Graph(
title="SLI: Network programming latency",
description=(
"NetworkProgrammingLatency is defined as the time it took to "
+ "program the network - from the time the service or pod has "
+ "changed to the time the change was propagated and the proper "
+ "kube-proxy rules were synced. Exported for each endpoints object "
+ "that were part of the rules sync."
),
targets=d.show_quantiles(
(
"quantile_over_time("
+ "0.99, "
+ 'kubeproxy:kubeproxy_network_programming_duration:histogram_quantile{{quantile="{quantile}"}}[24h])'
),
legend="{{quantile}}",
),
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
),
d.Graph(
title="Network programming latency",
description=(
"NetworkProgrammingLatency is defined as the time it took to "
+ "program the network - from the time the service or pod has "
+ "changed to the time the change was propagated and the proper "
q = "{:.2f}".format(quantile)
l = legend or q
targets.append(
g.Target(expr=queryTemplate.format(quantile=q), legendFormat=l))
return targets
NETWORK_PROGRAMMING_PANEL = [
d.Graph(
title="SLI: Network programming latency",
description=(
"NetworkProgrammingLatency is defined as the time it took to " +
"program the network - from the time the service or pod has " +
"changed to the time the change was propagated and the proper " +
"kube-proxy rules were synced. Exported for each endpoints object "
+ "that were part of the rules sync."),
targets=show_quantiles(
("quantile_over_time(" + "0.99, " +
'kubeproxy:kubeproxy_network_programming_duration:histogram_quantile{{quantile="{quantile}"}}[24h])'
),
legend="{{quantile}}",
),
yAxes=g.single_y_axis(format=g.SECONDS_FORMAT),
),
d.Graph(
title="Network programming latency",
description=(
"NetworkProgrammingLatency is defined as the time it took to " +
"program the network - from the time the service or pod has " +
"changed to the time the change was propagated and the proper " +
"kube-proxy rules were synced. Exported for each endpoints object "
+ "that were part of the rules sync."),
