Charts and metrics

Charts by resource type

The Metrics card on a resource's Overview shows:

Resource	Charts
Apps	CPU, Memory, Network In, Network Out, Disk, HTTP 2xx, HTTP 3xx, HTTP 4xx, HTTP Server Errors
Container instances	CPU, Memory, Network In, Network Out, Disk
PostgreSQL, SQL Server, Qdrant and Valkey databases	CPU, Memory, Network In, Network Out, Disk
Static web apps	HTTP 2xx, HTTP 3xx, HTTP 4xx, HTTP Server Errors
App slots	CPU, Memory, Network In, Network Out, Disk — but in this version the card shows no data: a slot's pods are measured as part of the app

Function namespaces, S3 storage accounts and secret vaults have no Metrics card. A function's page has metrics of its own — see Watch it run.

What each chart measures

Chart	Measures	Unit	Stored series
CPU	CPU time used by the resource's containers, added up over its pods, as a share of their CPU limits	% of limit; cores when no limit is declared	container_cpu_usage_seconds_total, as a rate
Memory	Working-set memory of the resource's containers, added up over its pods, as a share of their memory limits	% of limit; bytes when no limit is declared	container_memory_working_set_bytes, averaged
Disk	Used space of the resource's persistent volumes, added up over its volumes, as a share of their capacity	% of capacity	kubelet_volume_stats_used_bytes over kubelet_volume_stats_capacity_bytes
Network In	Bytes the resource's pods received	bytes per second	container_network_receive_bytes_total, as a rate
Network Out	Bytes the resource's pods sent	bytes per second	container_network_transmit_bytes_total, as a rate
HTTP 2xx, HTTP 3xx, HTTP 4xx	Requests the ingress answered with a status in that class	requests per point	The ingress controller's request counter
HTTP Server Errors	Requests answered with a 5xx status	requests per point	The ingress controller's request counter

• How the percentages are calculated is explained in Percent of what.
• The CPU and memory percentages leave out the pods that build an app's image.
• An app's charts include its slots.
• 1xx responses are not charted, so the four HTTP charts do not add up to all traffic.
• When one HTTP class has traffic in the range, the others are drawn as zero where they had none.

Stored series

These are the series stored for a resource, for the metric parameter of the metrics route — see Through the API. Everything else the container and volume statistics report is dropped.

Series	What it is
container_cpu_usage_seconds_total	CPU time used, in seconds; a counter
container_memory_working_set_bytes	Working-set memory; what the Memory chart shows
container_memory_rss	Resident (RSS) memory
container_network_receive_bytes_total, container_network_transmit_bytes_total	Bytes received and sent, per pod; counters
container_fs_reads_bytes_total, container_fs_writes_bytes_total	Bytes the container read and wrote; counters
container_fs_usage_bytes	Space used by the container's own writable layer
container_start_time_seconds	When the container started
container_oom_events_total	Out-of-memory events; a counter
container_cpu_cfs_periods_total, container_cpu_cfs_throttled_periods_total	CPU scheduling periods, and those in which the container was throttled; counters
kubelet_volume_stats_used_bytes, kubelet_volume_stats_capacity_bytes, kubelet_volume_stats_available_bytes	Used, total and available space of each persistent volume

Use Rate for the counters. The route adds the series up over the resource's pods and containers, except container_start_time_seconds, which it returns per series.