Improve range cache efficiency

cmd/pint/tests/0025_config.txt

@@ -16,7 +16,24 @@ level=info msg="Loading configuration file" path=.pint.hcl
       "uri": "https://",
       "timeout": "2m",
       "concurrency": 16,
+      "cache": 10000,
       "required": true
+    },
+    {
+      "name": "custom-concurrency",
+      "uri": "https://",
+      "timeout": "2m",
+      "concurrency": 32,
+      "cache": 10000,
+      "required": false
+    },
+    {
+      "name": "custom-cache",
+      "uri": "https://",
+      "timeout": "10s",
+      "concurrency": 16,
+      "cache": 32,
+      "required": false
     }
   ],
   "checks": {
@@ -133,6 +150,19 @@ prometheus "prom" {
   required = true
 }
 
+prometheus "custom-concurrency" {
+  uri         = "https://"
+  timeout     = "2m"
+  required    = false
+  concurrency = 32
+}
+
+prometheus "custom-cache" {
+  uri      = "https://"
+  timeout  = "10s"
+  cache    = 32
+}
+
 checks {
   disabled = ["promql/fragile"]
 }

cmd/pint/tests/0041_watch.txt

@@ -6,10 +6,8 @@ pint.ok --no-color -l debug watch --interval=5s --listen=127.0.0.1:6041 --pidfil
 stderr 'level=info msg="Pidfile created" path=pint.pid'
 stderr 'level=info msg="Started HTTP server" address=127.0.0.1:6041'
 stderr 'level=info msg="Will continuously run checks until terminated" interval=5s'
-stderr 'level=debug msg="Clearing cache"'
 stderr 'level=debug msg="Running checks"'
 stderr 'level=error msg="Got an error when running checks" error="no matching files"'
-stderr 'level=debug msg="Clearing cache"'
 stderr 'level=debug msg="Running checks"'
 stderr 'level=error msg="Got an error when running checks" error="no matching files"'
 stderr 'level=info msg="Shutting down"'

cmd/pint/watch.go

@@ -181,10 +181,6 @@ func startTimer(ctx context.Context, cfg config.Config, workers int, interval ti
 		for {
 			select {
 			case <-ticker.C:
-				log.Debug().Msg("Clearing cache")
-				for _, prom := range cfg.PrometheusServers {
-					prom.ClearCache()
-				}
 				log.Debug().Msg("Running checks")
 				if !wasBootstrapped {
 					ticker.Reset(interval)

docs/changelog.md

@@ -2,10 +2,19 @@
 
 ## v0.23.0
 
+### Fixed
+
+- Improved range query cache efficiency.
+
 ### Added
 
 - Added extra global configuration for `promql/series` check.
   See check [documentation](checks/promql/series.md) for details.
+- `prometheus` server definition in `pint` config file can now accept optional
+  `cache` field (defaults to 10000) to allow fine tuning of built-in Prometheus
+  API query caching.
+- Added `pint_prometheus_cache_size` metric that exposes the number of entries
+  currently in the query cache.
 
 ## v0.22.2
 

docs/configuration.md

@@ -158,6 +158,7 @@ prometheus "$name" {
   failover    = ["https://...", ...]
   timeout     = "60s"
   concurrency = 16
+  cache       = 10000
   required    = true|false
   paths       = ["...", ...]
 }
@@ -177,6 +178,7 @@ prometheus "$name" {
 - `timeout` - timeout to be used for API requests.
 - `concurrency` - how many concurrent requests can pint send to this Prometheus.
   Optional, defaults to 16.
+- `cache` - size of the query cache, defaults to 10000.
 - `required` - decides how pint will report errors if it's unable to get a valid response
   from this Prometheus server. If `required` is `true` and all API calls to this Prometheus
   fail pint will report those as `bug` level problem. If it's set to `false` pint will

internal/checks/base_test.go

@@ -62,7 +62,7 @@ func simpleProm(name, uri string, timeout time.Duration, required bool) *promapi
 	return promapi.NewFailoverGroup(
 		name,
 		[]*promapi.Prometheus{
-			promapi.NewPrometheus(name, uri, timeout, 16),
+			promapi.NewPrometheus(name, uri, timeout, 16, 1000),
 		},
 		required,
 	)