Skip to content

Revert "admission: add per-group CPU time token metrics" #171136

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wenyihu6 wants to merge 2 commits into
base: master
Choose a base branch
from
Draft

Revert "admission: add per-group CPU time token metrics" #171136

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9f38fb2
Revert "admission: add per-group CPU time token metrics"
wenyihu6 May 28, 2026
34b011a
Revert "admission: garbage collect resource manager groups less frequ…
wenyihu6 May 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 4 additions & 40 deletions docs/generated/metrics/metrics.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13580,15 +13580,6 @@ layers:
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.admitted_count
exported_name: admission_cpu_time_tokens_admitted_count
description: Cumulative number of requests admitted per group by CPU time token admission control; use with wait_time_nanos to compute mean wait time via rate(wait_time) / rate(admitted_count)
y_axis_label: Requests
type: COUNTER
unit: COUNT
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.dampening_deficit_nanos
exported_name: admission_cpu_time_tokens_dampening_deficit_nanos
description: Cumulative time-weighted deficit from dampening CPU time token allocations during scheduler overload. Each tick increments by (1 - dampening_factor) * elapsed_nanos, where dampening_factor is in [floor, 1.0] (floor is 0.5). rate() / 1e9 yields the average fraction of token issuance suppressed; a rate of 0 means no dampening was active in the window, 5e8 means the floor was held for the full window
Expand Down Expand Up @@ -13627,7 +13618,7 @@ layers:
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.per_tenant.admitted_count
exported_name: admission_cpu_time_tokens_per_tenant_admitted_count
description: Cumulative number of requests admitted per tenant by CPU time token admission control; use with wait_time_nanos to compute mean wait time via rate(wait_time) / rate(admitted_count). Retained for compatibility with dashboards predating the admission.cpu_time_tokens.admitted_count family
description: Cumulative number of requests admitted per tenant by CPU time token admission control; use with wait_time_nanos to compute mean wait time via rate(wait_time) / rate(admitted_count)
y_axis_label: Requests
type: COUNTER
unit: COUNT
Expand All @@ -13636,7 +13627,7 @@ layers:
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.per_tenant.tokens_returned
exported_name: admission_cpu_time_tokens_per_tenant_tokens_returned
description: Cumulative CPU time tokens returned per tenant, for example when actual CPU usage was lower than the initial estimate; rate() gives the per-tenant token return rate. Retained for compatibility with dashboards predating the admission.cpu_time_tokens.tokens_returned family
description: Cumulative CPU time tokens returned per tenant, for example when actual CPU usage was lower than the initial estimate; rate() gives the per-tenant token return rate
y_axis_label: Tokens
type: COUNTER
unit: COUNT
Expand All @@ -13645,7 +13636,7 @@ layers:
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.per_tenant.tokens_used
exported_name: admission_cpu_time_tokens_per_tenant_tokens_used
description: Cumulative CPU time tokens consumed per tenant by admitted work; rate() gives the per-tenant token consumption rate. Retained for compatibility with dashboards predating the admission.cpu_time_tokens.tokens_used family
description: Cumulative CPU time tokens consumed per tenant by admitted work; rate() gives the per-tenant token consumption rate
y_axis_label: Tokens
type: COUNTER
unit: COUNT
Expand All @@ -13654,7 +13645,7 @@ layers:
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.per_tenant.wait_time_nanos
exported_name: admission_cpu_time_tokens_per_tenant_wait_time_nanos
description: Cumulative nanoseconds of admission queue wait time per tenant in CPU time token admission control; use with admitted_count to compute mean wait time via rate(wait_time) / rate(admitted_count). Retained for compatibility with dashboards predating the admission.cpu_time_tokens.wait_time_nanos family
description: Cumulative nanoseconds of admission queue wait time per tenant in CPU time token admission control; use with admitted_count to compute mean wait time via rate(wait_time) / rate(admitted_count)
y_axis_label: Nanoseconds
type: COUNTER
unit: NANOSECONDS
Expand Down Expand Up @@ -13697,24 +13688,6 @@ layers:
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.tokens_returned
exported_name: admission_cpu_time_tokens_tokens_returned
description: Cumulative CPU time tokens returned per group, for example when actual CPU usage was lower than the initial estimate; rate() gives the per-group token return rate
y_axis_label: Tokens
type: COUNTER
unit: COUNT
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.tokens_used
exported_name: admission_cpu_time_tokens_tokens_used
description: Cumulative CPU time tokens consumed per group by admitted work; rate() gives the per-group token consumption rate
y_axis_label: Tokens
type: COUNTER
unit: COUNT
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.usage.consumed
exported_name: admission_cpu_time_tokens_usage_consumed
description: Cumulative number of CPU time tokens consumed (deducted from buckets) by admitted work
Expand All @@ -13733,15 +13706,6 @@ layers:
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.cpu_time_tokens.wait_time_nanos
exported_name: admission_cpu_time_tokens_wait_time_nanos
description: Cumulative nanoseconds of admission queue wait time per group in CPU time token admission control; use with admitted_count to compute mean wait time via rate(wait_time) / rate(admitted_count)
y_axis_label: Nanoseconds
type: COUNTER
unit: NANOSECONDS
aggregation: AVG
derivative: NON_NEGATIVE_DERIVATIVE
owner: cockroachdb/admission-control
- name: admission.elastic_cpu.acquired_nanos
exported_name: admission_elastic_cpu_acquired_nanos
description: Total CPU nanoseconds acquired by elastic work
Expand Down
4 changes: 0 additions & 4 deletions pkg/internal/metricscan/metric_owners.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,18 +26,14 @@ owners:
admission.cpu_time_tokens.refill.added.%s: cockroachdb/admission-control
admission.cpu_time_tokens.refill.removed.%s: cockroachdb/admission-control
admission_admitted_snapshot_bytes: cockroachdb/admission-control
admission_cpu_time_tokens_admitted_count: cockroachdb/admission-control
admission_cpu_time_tokens_dampening_deficit_nanos: cockroachdb/admission-control
admission_cpu_time_tokens_multiplier: cockroachdb/admission-control
admission_cpu_time_tokens_per_tenant_admitted_count: cockroachdb/admission-control
admission_cpu_time_tokens_per_tenant_tokens_returned: cockroachdb/admission-control
admission_cpu_time_tokens_per_tenant_tokens_used: cockroachdb/admission-control
admission_cpu_time_tokens_per_tenant_wait_time_nanos: cockroachdb/admission-control
admission_cpu_time_tokens_tokens_returned: cockroachdb/admission-control
admission_cpu_time_tokens_tokens_used: cockroachdb/admission-control
admission_cpu_time_tokens_usage_consumed: cockroachdb/admission-control
admission_cpu_time_tokens_usage_returned: cockroachdb/admission-control
admission_cpu_time_tokens_wait_time_nanos: cockroachdb/admission-control
admission_elastic_cpu_acquired_nanos: cockroachdb/admission-control
admission_elastic_cpu_available_nanos: cockroachdb/admission-control
admission_elastic_cpu_bypassed_utilization: cockroachdb/admission-control
Expand Down
4 changes: 0 additions & 4 deletions pkg/roachprod/agents/opentelemetry/files/cockroachdb_metrics.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,6 @@ admission_admitted_sql_sql_response_low_pri: admission.admitted.sql_sql_response
admission_admitted_sql_sql_response_normal_pri: admission.admitted.sql_sql_response.normal_pri
admission_admitted_sql_sql_response_user_high_pri: admission.admitted.sql_sql_response.user_high_pri
admission_admitted_sql_sql_response_user_low_pri: admission.admitted.sql_sql_response.user_low_pri
admission_cpu_time_tokens_admitted_count: admission.cpu_time_tokens.admitted_count
admission_cpu_time_tokens_dampening_deficit_nanos: admission.cpu_time_tokens.dampening_deficit_nanos
admission_cpu_time_tokens_exhausted_duration_nanos_can_burst: admission.cpu_time_tokens.exhausted_duration_nanos.can_burst
admission_cpu_time_tokens_exhausted_duration_nanos_no_burst: admission.cpu_time_tokens.exhausted_duration_nanos.no_burst
Expand All @@ -123,11 +122,8 @@ admission_cpu_time_tokens_refill_added_can_burst: admission.cpu_time_tokens.refi
admission_cpu_time_tokens_refill_added_no_burst: admission.cpu_time_tokens.refill.added.no_burst
admission_cpu_time_tokens_refill_removed_can_burst: admission.cpu_time_tokens.refill.removed.can_burst
admission_cpu_time_tokens_refill_removed_no_burst: admission.cpu_time_tokens.refill.removed.no_burst
admission_cpu_time_tokens_tokens_returned: admission.cpu_time_tokens.tokens_returned
admission_cpu_time_tokens_tokens_used: admission.cpu_time_tokens.tokens_used
admission_cpu_time_tokens_usage_consumed: admission.cpu_time_tokens.usage.consumed
admission_cpu_time_tokens_usage_returned: admission.cpu_time_tokens.usage.returned
admission_cpu_time_tokens_wait_time_nanos: admission.cpu_time_tokens.wait_time_nanos
admission_elastic_cpu_acquired_nanos: admission.elastic_cpu.acquired_nanos
admission_elastic_cpu_available_nanos: admission.elastic_cpu.available_nanos
admission_elastic_cpu_bypassed_utilization: admission.elastic_cpu_bypassed.utilization
Expand Down
1 change: 0 additions & 1 deletion pkg/util/admission/BUILD.bazel
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -123,7 +123,6 @@ go_test(
"@com_github_cockroachdb_redact//:redact",
"@com_github_cockroachdb_tokenbucket//:tokenbucket",
"@com_github_guptarohit_asciigraph//:asciigraph",
"@com_github_prometheus_client_model//go",
"@com_github_stretchr_testify//require",
],
)
16 changes: 4 additions & 12 deletions pkg/util/admission/cpu_time_token_grant_coordinator.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -238,18 +238,10 @@ func makeCPUTimeTokenGrantCoordinator(
wqOpts := makeWorkQueueOptions(KVWork)
wqOpts.mode = usesCPUTimeTokens
wqOpts.perGroupAggMetrics = &groupAggMetrics{
primary: &groupAggMetricSet{
admittedCount: metrics.AdmittedCount,
waitTimeNanos: metrics.WaitTimeNanos,
tokensUsed: metrics.TokensUsed,
tokensReturned: metrics.TokensReturned,
},
legacy: &groupAggMetricSet{
admittedCount: metrics.LegacyAdmittedCountPerTenant,
waitTimeNanos: metrics.LegacyWaitTimeNanosPerTenant,
tokensUsed: metrics.LegacyTokensUsedPerTenant,
tokensReturned: metrics.LegacyTokensReturnedPerTenant,
},
admittedCount: metrics.AdmittedCountPerTenant,
waitTimeNanos: metrics.WaitTimeNanosPerTenant,
tokensUsed: metrics.TokensUsedPerTenant,
tokensReturned: metrics.TokensReturnedPerTenant,
}
wqOpts.configHolder = configHolder
wqOpts.groupKeyForWorkInfo = cpuTimeTokenGroupKeyForWorkInfo
Expand Down
4 changes: 2 additions & 2 deletions pkg/util/admission/cpu_time_token_granter.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,9 +159,9 @@ func (stg *cpuTimeTokenGranter) tookWithoutPermissionLocked(count int64) {
// compute rate(consumed) - rate(returned) over arbitrary windows
// (1m, 30m, etc.).
if count > 0 {
stg.metrics.UsageConsumed.Inc(count)
stg.metrics.TokensConsumed.Inc(count)
} else {
stg.metrics.UsageReturned.Inc(-count)
stg.metrics.TokensReturned.Inc(-count)
}
now := stg.timeSource.Now()
for qual := range stg.mu.buckets {
Expand Down
Loading
Loading