Fix: Update Prometheus gauge handling in HTTP client plugins

Add increment calls at the beginning and change misplaced increment to decrement after HTTP requests in various plugins. This ensures proper tracking of active requests.

Signed-off-by: Christian Roessner <[email protected]>

Author: Christian Roessner

server/lua-plugins.d/actions/haveibeenpwnd.lua

@@ -90,6 +90,8 @@ function nauthilus_call_action(request)
         nauthilus_prometheus.create_gauge_vec(HCCR, "Measure the number of total concurrent HTTP client requests", { "service" })
         nauthilus_prometheus.create_histogram_vec(N .. "_duration_seconds", "HTTP request to the haveibeenpwnd network", { "http" })
 
+        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+
         local timer = nauthilus_prometheus.start_histogram_timer(N .. "_duration_seconds", { http = "get" })
         local result, err = http.get("https://api.pwnedpasswords.com/range/" .. hash:sub(1, 5), {
             timeout = "10s",
@@ -100,7 +102,7 @@ function nauthilus_call_action(request)
         })
         nauthilus_prometheus.stop_timer(timer)
         nauthilus_util.if_error_raise(err)
-        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+        nauthilus_prometheus.decrement_gauge(HCCR, { service = N })
 
         if result.status_code ~= 200 then
             nauthilus_util.if_error_raise(N .. "_status_code=" .. tostring(result.status_code))

server/lua-plugins.d/actions/telegram.lua

@@ -156,14 +156,16 @@ function nauthilus_call_action(request)
         nauthilus_prometheus.create_gauge_vec(HCCR, "Measure the number of total concurrent HTTP client requests", { "service" })
         nauthilus_prometheus.create_histogram_vec(N .. "_duration_seconds", "HTTP request to the telegram network", { "bot" })
 
+        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+
         local timer = nauthilus_prometheus.start_histogram_timer(N .. "_duration_seconds", { bot = "send" })
         local _, err_bat = bot:sendMessage({
             chat_id = tonumber(os.getenv("TELEGRAM_CHAT_ID")),
             text = headline .. mustache:render(":\n\nSESSION {{session}}\nTS {{ts}}\nIP {{client_ip}}\nHOSTNAME {{hostname}}\nPROTOCOL {{proto}}\nDISPLAY_NAME {{display_name}}\nACCOUNT {{account}}\nUNIQUE ID {{unique_user_id}}\nUSERNAME {{username}}\nPWND INFO {{pwnd_info}}", values)
         })
         nauthilus_prometheus.stop_timer(timer)
         nauthilus_util.if_error_raise(err_bat)
-        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+        nauthilus_prometheus.decrement_gauge(HCCR, { service = N })
     end
 
     rt.post_telegram = true

server/lua-plugins.d/features/blocklist.lua

@@ -52,6 +52,8 @@ function nauthilus_call_feature(request)
     nauthilus_prometheus.create_gauge_vec(HCCR, "Measure the number of total concurrent HTTP client requests", { "service" })
     nauthilus_prometheus.create_histogram_vec(N .. "_duration_seconds", "HTTP request to the blocklist service", { "http" })
 
+    nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+
     local timer = nauthilus_prometheus.start_histogram_timer(N .. "_duration_seconds", { http = "post" })
     local result, request_err = http.post(os.getenv("BLOCKLIST_URL"), {
         timeout = "10s",
@@ -64,7 +66,7 @@ function nauthilus_call_feature(request)
     })
     nauthilus_prometheus.stop_timer(timer)
     nauthilus_util.if_error_raise(request_err)
-    nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+    nauthilus_prometheus.decrement_gauge(HCCR, { service = N })
 
     if result.status_code ~= 200 then
         nauthilus_util.if_error_raise(N .. "_status_code=" .. tostring(result.status_code))

server/lua-plugins.d/filters/geoip.lua

@@ -85,6 +85,8 @@ function nauthilus_call_filter(request)
         nauthilus_prometheus.create_histogram_vec(N .. "_duration_seconds", "HTTP request to the geoip-policyd service", { "http" })
         nauthilus_prometheus.create_counter_vec(N .. "_count", "Count GeoIP countries", { "country", "status" })
 
+        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+
         local timer = nauthilus_prometheus.start_histogram_timer(N .. "_duration_seconds", { http = "post" })
         local  result, request_err = http.post(os.getenv("GEOIP_POLICY_URL"), {
             timeout = "10s",
@@ -97,7 +99,7 @@ function nauthilus_call_filter(request)
         })
         nauthilus_prometheus.stop_timer(timer)
         nauthilus_util.if_error_raise(request_err)
-        nauthilus_prometheus.increment_gauge(HCCR, { service = N })
+        nauthilus_prometheus.decrement_gauge(HCCR, { service = N })
 
         if result.status_code ~= 202 then
             nauthilus_util.if_error_raise(N .. "_status_code=" .. tostring(result.code))