Conformance with Prometheus metric names and types

[bwerthmann]

Conformance to Prometheus metric name standards.

Suggested Prerequisite: #71

Compliance with metric name/TYPE standards improves operator/developer experience by enabling advanced features in Prometheus and related tools like promlens, Grafana, etc. Features such as context aware auto-complete / and PromQL linters.

Prometheus is unable to infer the type due to non compliance

Prometheus is able to infer the type

promtool has a linter for this

This tool compares the names and the types, but is not aware what the actual type is. Thus why #71 is suggested first.

curl -sS localhost:7777/metrics | ./prometheus-2.32.1.linux-amd64/promtool check metrics

nats_core_active_account_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_connection_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_core_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_gateway_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_gateway_inbound_msg_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_gateway_recv_msg_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_gateway_sent_msgs_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_recv_msgs_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_route_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_route_recv_msg_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_route_sent_msg_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_sent_msgs_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_slow_consumer_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_subs_count non-histogram and non-summary metrics should not have "_count" suffix
nats_core_total_connection_count non-histogram and non-summary metrics should not have "_count" suffix
nats_jetstream_advisory_count non-histogram and non-summary metrics should not have "_count" suffix
nats_latency_observations_count non-histogram and non-summary metrics should not have "_count" suffix
nats_survey_expected_count non-histogram and non-summary metrics should not have "_count" suffix
nats_survey_late_replies_count counter metrics should have "_total" suffix
nats_survey_late_replies_count non-histogram and non-summary metrics should not have "_count" suffix
nats_survey_no_replies_count counter metrics should have "_total" suffix
nats_survey_no_replies_count non-histogram and non-summary metrics should not have "_count" suffix
nats_survey_surveyed_count non-histogram and non-summary metrics should not have "_count" suffix

two other cases

nats_core_rtt_nanoseconds use base unit "seconds" instead of "nanoseconds"
nats_survey_nats_reconnects counter metrics should have "_total" suffix

---

Example counter metric: nats_core_total_connection_count

Example Scrape

# HELP nats_core_total_connection_count Total number of client connections serviced gauge
# TYPE nats_core_total_connection_count gauge
nats_core_total_connection_count{server_cluster="...",server_id="...",server_name="..."} 6345789

Linter for nats_core_total_connection_count

curl -sS localhost:7777/metrics | grep -F 'nats_core_total_connection_count' |head -n3 | ./prometheus-2.32.1.linux-amd64/promtool check metrics

nats_core_total_connection_count non-histogram and non-summary metrics should not have "_count" suffix

issue #71 will find this is a counter

curl -sS localhost:7777/metrics | grep -F 'nats_core_total_connection_count' |head -n3 |sed 's/gauge/counter/'| ./prometheus-2.32.1.linux-amd64/promtool check metrics

Two problems.

nats_core_total_connection_count counter metrics should have "_total" suffix
nats_core_total_connection_count non-histogram and non-summary metrics should not have "_count" suffix

Example "fixed" scrape

# HELP nats_core_total_connection_total Total number of client connections serviced counter
# TYPE nats_core_total_connection_total counter
nats_core_total_connection_total{server_cluster="...",server_id="...",server_name="..."} 42224222