compile silence matchers when the silence is added #4695

ultrotter · 2025-11-05T15:05:21Z

If QMatches doesn't update the map, multiple queries can execute
concurrently, with just a read lock. We have had this in production for
over a year, and it significantly improves performance for cases where
many silences need to be processed against many alerts.

This makes load time a bit slower, but a snapshot file with about
120.000 silences can still load in about 2 seconds, so we can accept
this, in exchange for the ability to run queries in parallel.

silence/silence.go

SuperQ · 2025-11-05T15:53:21Z

Looks like this needs a rebase.

silence/silence.go

If QMatches doesn't update the map, multiple queries can execute concurrently, with just a read lock. We have had this in production for over a year, and it significantly improves performance for cases where many silences need to be processed against many alerts. This makes load time a bit slower, but a snapshot file with about 120.000 silences can still load in about 2 seconds, so we can accept this, in exchange for the ability to run queries in parallel. goos: linux goarch: amd64 pkg: github.com/prometheus/alertmanager/cluster cpu: AMD EPYC Processor (with IBPB) │ bench-before.txt │ bench-after.txt │ │ sec/op │ sec/op vs base │ WriteTo-40 0.003794n ± ∞ ¹ 0.004227n ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 │ bench-before.txt │ bench-after.txt │ │ B/op │ B/op vs base │ WriteTo-40 0.000 ± ∞ ¹ 0.000 ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal │ bench-before.txt │ bench-after.txt │ │ allocs/op │ allocs/op vs base │ WriteTo-40 0.000 ± ∞ ¹ 0.000 ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal pkg: github.com/prometheus/alertmanager/inhibit │ bench-before.txt │ bench-after.txt │ │ sec/op │ sec/op vs base │ Mutes/1_inhibition_rule,_1_inhibiting_alert-40 1.248µ ± ∞ ¹ 1.340µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_1_inhibiting_alert-40 1.298µ ± ∞ ¹ 1.326µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1_inhibiting_alert-40 1.347µ ± ∞ ¹ 1.391µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_1_inhibiting_alert-40 1.498µ ± ∞ ¹ 1.578µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_1_inhibiting_alert-40 1.685µ ± ∞ ¹ 1.502µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10_inhibiting_alerts-40 1.443µ ± ∞ ¹ 1.521µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_100_inhibiting_alerts-40 1.453µ ± ∞ ¹ 1.561µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_1000_inhibiting_alerts-40 1.423µ ± ∞ ¹ 1.511µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10000_inhibiting_alerts-40 1.389µ ± ∞ ¹ 1.441µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1000_inhibiting_alerts-40 1.218µ ± ∞ ¹ 1.306µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_last_rule_matches-40 3.054µ ± ∞ ¹ 3.114µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_last_rule_matches-40 19.57µ ± ∞ ¹ 19.47µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_last_rule_matches-40 187.4µ ± ∞ ¹ 188.9µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_last_rule_matches-40 1.944m ± ∞ ¹ 1.955m ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 4.239µ 4.356µ +2.75% ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 │ bench-before.txt │ bench-after.txt │ │ B/op │ B/op vs base │ Mutes/1_inhibition_rule,_1_inhibiting_alert-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_1_inhibiting_alert-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1_inhibiting_alert-40 489.0 ± ∞ ¹ 489.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_1_inhibiting_alert-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_1_inhibiting_alert-40 489.0 ± ∞ ¹ 489.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10_inhibiting_alerts-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_100_inhibiting_alerts-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_1000_inhibiting_alerts-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10000_inhibiting_alerts-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1000_inhibiting_alerts-40 488.0 ± ∞ ¹ 488.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_last_rule_matches-40 472.0 ± ∞ ¹ 472.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_last_rule_matches-40 472.0 ± ∞ ¹ 472.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_last_rule_matches-40 473.0 ± ∞ ¹ 473.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_last_rule_matches-40 481.0 ± ∞ ¹ 481.0 ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 484.2 484.2 +0.00% ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal │ bench-before.txt │ bench-after.txt │ │ allocs/op │ allocs/op vs base │ Mutes/1_inhibition_rule,_1_inhibiting_alert-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_1_inhibiting_alert-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1_inhibiting_alert-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_1_inhibiting_alert-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_1_inhibiting_alert-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10_inhibiting_alerts-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_100_inhibiting_alerts-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_1000_inhibiting_alerts-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1_inhibition_rule,_10000_inhibiting_alerts-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_1000_inhibiting_alerts-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_inhibition_rules,_last_rule_matches-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_inhibition_rules,_last_rule_matches-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_inhibition_rules,_last_rule_matches-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_inhibition_rules,_last_rule_matches-40 10.00 ± ∞ ¹ 10.00 ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 10.00 10.00 +0.00% ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal pkg: github.com/prometheus/alertmanager/matcher/parse │ bench-before.txt │ bench-after.txt │ │ sec/op │ sec/op vs base │ ParseSimple-40 1.046µ ± ∞ ¹ 1.096µ ± ∞ ¹ ~ (p=1.000 n=1) ² ParseComplex-40 4.855µ ± ∞ ¹ 4.769µ ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 2.254µ 2.286µ +1.45% ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 │ bench-before.txt │ bench-after.txt │ │ B/op │ B/op vs base │ ParseSimple-40 232.0 ± ∞ ¹ 232.0 ± ∞ ¹ ~ (p=1.000 n=1) ² ParseComplex-40 2.602Ki ± ∞ ¹ 2.602Ki ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 786.2 786.2 +0.00% ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal │ bench-before.txt │ bench-after.txt │ │ allocs/op │ allocs/op vs base │ ParseSimple-40 7.000 ± ∞ ¹ 7.000 ± ∞ ¹ ~ (p=1.000 n=1) ² ParseComplex-40 42.00 ± ∞ ¹ 42.00 ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 17.15 17.15 +0.00% ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal pkg: github.com/prometheus/alertmanager/notify │ bench-before.txt │ bench-after.txt │ │ sec/op │ sec/op vs base │ HashAlert-40 239.8n ± ∞ ¹ 244.2n ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 │ bench-before.txt │ bench-after.txt │ │ B/op │ B/op vs base │ HashAlert-40 72.00 ± ∞ ¹ 72.00 ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal │ bench-before.txt │ bench-after.txt │ │ allocs/op │ allocs/op vs base │ HashAlert-40 2.000 ± ∞ ¹ 2.000 ± ∞ ¹ ~ (p=1.000 n=1) ² ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal pkg: github.com/prometheus/alertmanager/silence │ bench-before.txt │ bench-after.txt │ │ sec/op │ sec/op vs base │ Mutes/1_silence_mutes_alert-40 10.50µ ± ∞ ¹ 10.23µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_silences_mute_alert-40 12.15µ ± ∞ ¹ 13.85µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_silences_mute_alert-40 35.81µ ± ∞ ¹ 39.36µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_silences_mute_alert-40 430.7µ ± ∞ ¹ 474.4µ ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_silences_mute_alert-40 5.054m ± ∞ ¹ 5.211m ± ∞ ¹ ~ (p=1.000 n=1) ² Query/100_silences-40 31.09µ ± ∞ ¹ Query/1000_silences-40 314.3µ ± ∞ ¹ Query/10000_silences-40 6.531m ± ∞ ¹ QueryParallel/100_silences,_1_goroutine-40 38.32µ ± ∞ ¹ 18.78µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_2_goroutines-40 37.57µ ± ∞ ¹ 17.50µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_4_goroutines-40 39.29µ ± ∞ ¹ 17.52µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_8_goroutines-40 37.88µ ± ∞ ¹ 18.83µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_1_goroutine-40 351.65µ ± ∞ ¹ 46.81µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_2_goroutines-40 335.03µ ± ∞ ¹ 44.99µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_4_goroutines-40 362.03µ ± ∞ ¹ 46.17µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_8_goroutines-40 338.11µ ± ∞ ¹ 46.33µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_1_goroutine-40 6979.9µ ± ∞ ¹ 354.1µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_2_goroutines-40 6454.0µ ± ∞ ¹ 363.8µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_4_goroutines-40 6607.6µ ± ∞ ¹ 346.9µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_8_goroutines-40 9666.4µ ± ∞ ¹ 409.3µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_10%_add_rate-40 493.8µ ± ∞ ¹ 154.1µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_1%_add_rate-40 373.19µ ± ∞ ¹ 70.64µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_0.1%_add_rate-40 358.16µ ± ∞ ¹ 49.32µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/10000_initial_silences,_1%_add_rate-40 7028.4µ ± ∞ ¹ 524.5µ ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/10000_initial_silences,_0.1%_add_rate-40 8035.2µ ± ∞ ¹ 374.5µ ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/100_silences,_4_goroutines-40 53.23µ ± ∞ ¹ 39.80µ ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/1000_silences,_4_goroutines-40 484.7µ ± ∞ ¹ 173.0µ ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/10000_silences,_4_goroutines-40 4.185m ± ∞ ¹ 1.112m ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/10000_silences,_8_goroutines-40 4.365m ± ∞ ¹ 1.077m ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 451.9µ 109.5µ -76.10% ³ ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 ³ benchmark set differs from baseline; geomeans may not be comparable │ bench-before.txt │ bench-after.txt │ │ B/op │ B/op vs base │ Mutes/1_silence_mutes_alert-40 3.835Ki ± ∞ ¹ 3.718Ki ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_silences_mute_alert-40 6.740Ki ± ∞ ¹ 6.799Ki ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_silences_mute_alert-40 35.99Ki ± ∞ ¹ 35.87Ki ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_silences_mute_alert-40 308.2Ki ± ∞ ¹ 308.1Ki ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_silences_mute_alert-40 3.536Mi ± ∞ ¹ 3.527Mi ± ∞ ¹ ~ (p=1.000 n=1) ² Query/100_silences-40 6.142Ki ± ∞ ¹ Query/1000_silences-40 41.30Ki ± ∞ ¹ Query/10000_silences-40 525.2Ki ± ∞ ¹ QueryParallel/100_silences,_1_goroutine-40 6.052Ki ± ∞ ¹ 6.136Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_2_goroutines-40 6.048Ki ± ∞ ¹ 6.049Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_4_goroutines-40 6.048Ki ± ∞ ¹ 6.050Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_8_goroutines-40 6.053Ki ± ∞ ¹ 6.039Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_1_goroutine-40 41.30Ki ± ∞ ¹ 41.42Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_2_goroutines-40 41.27Ki ± ∞ ¹ 41.42Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_4_goroutines-40 41.32Ki ± ∞ ¹ 41.34Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_8_goroutines-40 41.29Ki ± ∞ ¹ 41.44Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_1_goroutine-40 525.0Ki ± ∞ ¹ 524.9Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_2_goroutines-40 524.9Ki ± ∞ ¹ 524.9Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_4_goroutines-40 524.9Ki ± ∞ ¹ 525.0Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_8_goroutines-40 525.0Ki ± ∞ ¹ 524.9Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_10%_add_rate-40 100.7Ki ± ∞ ¹ 226.4Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_1%_add_rate-40 61.88Ki ± ∞ ¹ 86.27Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_0.1%_add_rate-40 44.36Ki ± ∞ ¹ 46.04Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/10000_initial_silences,_1%_add_rate-40 1058.7Ki ± ∞ ¹ 533.3Ki ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/10000_initial_silences,_0.1%_add_rate-40 1118.4Ki ± ∞ ¹ 525.1Ki ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/100_silences,_4_goroutines-40 36.10Ki ± ∞ ¹ 36.05Ki ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/1000_silences,_4_goroutines-40 307.6Ki ± ∞ ¹ 307.4Ki ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/10000_silences,_4_goroutines-40 3.487Mi ± ∞ ¹ 3.517Mi ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/10000_silences,_8_goroutines-40 3.492Mi ± ∞ ¹ 3.514Mi ± ∞ ¹ ~ (p=1.000 n=1) ² geomean 94.43Ki 100.4Ki -0.98% ³ ¹ need >= 6 samples for confidence interval at level 0.95 ² need >= 4 samples to detect a difference at alpha level 0.05 ³ benchmark set differs from baseline; geomeans may not be comparable │ bench-before.txt │ bench-after.txt │ │ allocs/op │ allocs/op vs base │ Mutes/1_silence_mutes_alert-40 39.00 ± ∞ ¹ 39.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10_silences_mute_alert-40 59.00 ± ∞ ¹ 59.00 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/100_silences_mute_alert-40 158.0 ± ∞ ¹ 158.0 ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/1000_silences_mute_alert-40 1.071k ± ∞ ¹ 1.071k ± ∞ ¹ ~ (p=1.000 n=1) ² Mutes/10000_silences_mute_alert-40 10.18k ± ∞ ¹ 10.10k ± ∞ ¹ ~ (p=1.000 n=1) ³ Query/100_silences-40 45.00 ± ∞ ¹ Query/1000_silences-40 141.0 ± ∞ ¹ Query/10000_silences-40 1.051k ± ∞ ¹ QueryParallel/100_silences,_1_goroutine-40 42.00 ± ∞ ¹ 42.00 ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_2_goroutines-40 42.00 ± ∞ ¹ 42.00 ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_4_goroutines-40 42.00 ± ∞ ¹ 42.00 ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/100_silences,_8_goroutines-40 42.00 ± ∞ ¹ 42.00 ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/1000_silences,_1_goroutine-40 138.0 ± ∞ ¹ 139.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/1000_silences,_2_goroutines-40 138.0 ± ∞ ¹ 139.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/1000_silences,_4_goroutines-40 138.0 ± ∞ ¹ 139.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/1000_silences,_8_goroutines-40 138.0 ± ∞ ¹ 139.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/10000_silences,_1_goroutine-40 1.049k ± ∞ ¹ 1.050k ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/10000_silences,_2_goroutines-40 1.049k ± ∞ ¹ 1.049k ± ∞ ¹ ~ (p=1.000 n=1) ² QueryParallel/10000_silences,_4_goroutines-40 1.049k ± ∞ ¹ 1.050k ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryParallel/10000_silences,_8_goroutines-40 1.049k ± ∞ ¹ 1.049k ± ∞ ¹ ~ (p=1.000 n=1) ² QueryWithConcurrentAdds/1000_initial_silences,_10%_add_rate-40 400.0 ± ∞ ¹ 876.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryWithConcurrentAdds/1000_initial_silences,_1%_add_rate-40 204.0 ± ∞ ¹ 294.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryWithConcurrentAdds/1000_initial_silences,_0.1%_add_rate-40 175.0 ± ∞ ¹ 162.0 ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryWithConcurrentAdds/10000_initial_silences,_1%_add_rate-40 8.363k ± ∞ ¹ 1.065k ± ∞ ¹ ~ (p=1.000 n=1) ³ QueryWithConcurrentAdds/10000_initial_silences,_0.1%_add_rate-40 9.138k ± ∞ ¹ 1.052k ± ∞ ¹ ~ (p=1.000 n=1) ³ MutesParallel/100_silences,_4_goroutines-40 158.0 ± ∞ ¹ 158.0 ± ∞ ¹ ~ (p=1.000 n=1) ² MutesParallel/1000_silences,_4_goroutines-40 1.073k ± ∞ ¹ 1.072k ± ∞ ¹ ~ (p=1.000 n=1) ³ MutesParallel/10000_silences,_4_goroutines-40 10.18k ± ∞ ¹ 10.10k ± ∞ ¹ ~ (p=1.000 n=1) ³ MutesParallel/10000_silences,_8_goroutines-40 10.16k ± ∞ ¹ 10.10k ± ∞ ¹ ~ (p=1.000 n=1) ³ geomean 380.8 366.0 -11.39% ⁴ ¹ need >= 6 samples for confidence interval at level 0.95 ² all samples are equal ³ need >= 4 samples to detect a difference at alpha level 0.05 ⁴ benchmark set differs from baseline; geomeans may not be comparable Signed-off-by: Guido Trotter <guido@hudson-trading.com>

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

…e cache Before the silences would be in the status, but querying would result in query errors. Now they are still in the status (since we have to calculate after calling Merge), but we also know that some broken ones exist. At load snapshot time we can also avoid importing those, since they would not be usable anyway, and we have different labels to distinguish the cases/behaviors. Signed-off-by: Guido Trotter <guido@hudson-trading.com>

ultrotter · 2025-11-05T16:26:48Z

Looks like this needs a rebase.

Done!

rajagopalanand

Overall LGTM. Added just a nit comment

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

ultrotter · 2025-11-05T17:27:15Z

Overall LGTM. Added just a nit comment

Thanks! What's the comment? I only saw the suggestion for the function name, which I apply, but I can't see anything else... Is there another one? Thanks!

rajagopalanand · 2025-11-05T18:35:30Z

Overall LGTM. Added just a nit comment

Thanks! What's the comment? I only saw the suggestion for the function name, which I apply, but I can't see anything else... Is there another one? Thanks!

The function name suggestion was the nit comment

silence/silence.go

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

SuperQ reviewed Nov 5, 2025

View reviewed changes

silence/silence.go Outdated Show resolved Hide resolved

siavashs reviewed Nov 5, 2025

View reviewed changes

silence/silence.go Outdated Show resolved Hide resolved

silence/silence.go Outdated Show resolved Hide resolved

Guido Trotter added 3 commits November 5, 2025 11:23

Make 'get' private

5387548

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

ultrotter force-pushed the precompilesilencematcher branch from 602f97c to 70f47b1 Compare November 5, 2025 16:23

rajagopalanand approved these changes Nov 5, 2025

View reviewed changes

Rename silenceAdded to cacheSilence

fc41f8a

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

SuperQ approved these changes Nov 5, 2025

View reviewed changes

SuperQ reviewed Nov 5, 2025

View reviewed changes

silence/silence.go Outdated Show resolved Hide resolved

silence/silence.go Outdated Show resolved Hide resolved

Use matcherCompile.*Error metrics with pre-set labels

599be5d

Signed-off-by: Guido Trotter <guido@hudson-trading.com>

SuperQ merged commit 6ceb38d into prometheus:main Nov 5, 2025
7 checks passed

ultrotter deleted the precompilesilencematcher branch November 8, 2025 09:36

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

compile silence matchers when the silence is added #4695

compile silence matchers when the silence is added #4695

Uh oh!

ultrotter commented Nov 5, 2025

Uh oh!

Uh oh!

SuperQ commented Nov 5, 2025

Uh oh!

Uh oh!

Uh oh!

ultrotter commented Nov 5, 2025

Uh oh!

rajagopalanand left a comment

Uh oh!

ultrotter commented Nov 5, 2025

Uh oh!

rajagopalanand commented Nov 5, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

compile silence matchers when the silence is added #4695

compile silence matchers when the silence is added #4695

Uh oh!

Conversation

ultrotter commented Nov 5, 2025

Uh oh!

Uh oh!

SuperQ commented Nov 5, 2025

Uh oh!

Uh oh!

Uh oh!

ultrotter commented Nov 5, 2025

Uh oh!

rajagopalanand left a comment

Choose a reason for hiding this comment

Uh oh!

ultrotter commented Nov 5, 2025

Uh oh!

rajagopalanand commented Nov 5, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants