Search…
Metrics
The cortex get and cortex get API_NAME commands display the request time (averaged over the past 2 weeks) and response code counts (summed over the past 2 weeks) for your APIs:
1
cortex get
2
3
env api status up-to-date requested last update avg request 2XX
4
cortex iris-classifier live 1 1 17m 24ms 1223
5
cortex text-generator live 1 1 8m 180ms 433
6
cortex image-classifier-resnet50 live 2 2 1h 32ms 1121126
Copied!
The cortex get API_NAME command also provides a link to a Grafana dashboard:
dashboard

Metrics in the dashboard

Panel
Description
Note
Request Rate
Request rate, computed over every minute, of an API
In Flight Request
Active in-flight requests for an API.
In-flight requests are recorded every 10 seconds, which will correspond to the minimum resolution.
Active Replicas
Active replicas for an API
2XX Responses
Request rate, computed over a minute, for responses with status code 2XX of an API
4XX Responses
Request rate, computed over a minute, for responses with status code 4XX of an API
5XX Responses
Request rate, computed over a minute, for responses with status code 5XX of an API
p99 Latency
99th percentile latency, computed over a minute, for an API
Value might not be accurate because the histogram buckets are not dynamically set.
p90 Latency
90th percentile latency, computed over a minute, for an API
Value might not be accurate because the histogram buckets are not dynamically set.
p50 Latency
50th percentile latency, computed over a minute, for an API
Value might not be accurate because the histogram buckets are not dynamically set.
Average Latency
Average latency, computed over a minute, for an API
Last modified 15d ago
Copy link