fix: use microsecond precision for ICMP latency measurements

- Changed ping module to use microsecond timer resolution instead of millisecond
- Prevents 0ms readings for very fast pings (localhost, local network)
- Updated response_time_ms to float type to store decimal precision
- Migration converts existing integer values to float

lib/towerops/monitoring/check.ex

@@ -12,7 +12,7 @@ defmodule Towerops.Monitoring.Check do
   @foreign_key_type :binary_id
   schema "monitoring_checks" do
     field :status, Ecto.Enum, values: [:success, :failure]
-    field :response_time_ms, :integer
+    field :response_time_ms, :float
     field :checked_at, :utc_datetime
 
     belongs_to :device, Towerops.Devices.Device

lib/towerops/monitoring/ping.ex

@@ -28,16 +28,19 @@ defmodule Towerops.Monitoring.Ping do
   Pings an IP address using raw ICMP echo request/reply.
 
   Returns {:ok, response_time_ms} on success, {:error, reason} on failure.
+  Response time is returned as a float with microsecond precision.
   """
   @impl true
   def ping(ip_address, timeout_ms \\ 5000) do
-    start_time = System.monotonic_time(:millisecond)
+    start_time = System.monotonic_time(:microsecond)
 
     case send_icmp_echo(ip_address, timeout_ms) do
       :ok ->
-        end_time = System.monotonic_time(:millisecond)
-        response_time = end_time - start_time
-        {:ok, response_time}
+        end_time = System.monotonic_time(:microsecond)
+        response_time_us = end_time - start_time
+        # Convert to milliseconds with decimal precision
+        response_time_ms = response_time_us / 1000.0
+        {:ok, response_time_ms}
 
       {:error, reason} ->
         {:error, reason}

priv/repo/migrations/20260118185149_change_response_time_to_float.exs

@@ -0,0 +1,9 @@
+defmodule Towerops.Repo.Migrations.ChangeResponseTimeToFloat do
+  use Ecto.Migration
+
+  def change do
+    alter table(:monitoring_checks) do
+      modify :response_time_ms, :float
+    end
+  end
+end