Agent CDN - Added Warning in Initialize Phase when new Agent CDN is not reachable - AB#2241315 (#5202)

* Agent CDN - Added Warning in Initialize Phase when new Agent CDN is not reachable

* Added AgentCDNAccessStatus Telemetry Event

Author: AdityaMankal-MS

src/Agent.Sdk/Knob/AgentKnobs.cs

@@ -786,13 +786,21 @@ public class AgentKnobs
             "Timeout for channel communication between agent listener and worker processes.",
             new EnvironmentKnobSource("PIPELINE_ARTIFACT_ASSOCIATE_TIMEOUT"),
             new BuiltInDefaultKnobSource("900")); // 15 * 60 - Setting the timeout to 15 minutes to account for slowness from azure storage and retries.
-        
+
         public static readonly Knob RollbackToDefaultTfExe = new Knob(
             nameof(RollbackToDefaultTfExe),
             "If true, the agent will install the default versions of TF, vstsom",
             new RuntimeKnobSource("ROLLBACK_TO_DEFAULT_TF_EXE"),
             new EnvironmentKnobSource("ROLLBACK_TO_DEFAULT_TF_EXE"),
             new PipelineFeatureSource("RollbackToDefaultTfExe"),
             new BuiltInDefaultKnobSource("false"));
+
+        public static readonly Knob AgentCDNConnectivityFailWarning = new Knob(
+            nameof(AgentCDNConnectivityFailWarning),
+            "Show warning message when the Agent CDN Endpoint (download.agent.dev.azure.com) is not reachable. ",
+            new RuntimeKnobSource("AGENT_CDN_CONNECTIVITY_FAIL_WARNING"),
+            new EnvironmentKnobSource("AGENT_CDN_CONNECTIVITY_FAIL_WARNING"),
+            new PipelineFeatureSource("AgentCDNConnectivityFailWarning"),
+            new BuiltInDefaultKnobSource("false"));
     }
 }

src/Agent.Sdk/Util/PlatformUtil.cs

@@ -18,6 +18,8 @@
 using Newtonsoft.Json;
 using System.ServiceProcess;
 using Agent.Sdk.Util;
+using System.Net.Http;
+using System.Net;
 
 namespace Agent.Sdk
 {
@@ -438,6 +440,47 @@ public static bool DetectAzureVM()
             }
             return isAzureVM;
         }
+
+        // The URL of the agent package hosted on Azure CDN
+        private const string _agentPackageUri = "https://download.agent.dev.azure.com/agent/4.252.0/vsts-agent-win-x64-4.252.0.zip";
+
+#nullable enable
+        /// <summary>
+        /// Checks if the agent CDN endpoint is accessible by sending an HTTP HEAD request.
+        /// </summary>
+        /// <param name="webProxy">
+        /// Optional <see cref="IWebProxy"/> to route the request through a proxy. If null, the system default proxy settings are used.
+        /// </param>
+        /// <remarks>
+        /// - Returns <c>true</c> if the endpoint responds with a successful (2xx) status code.
+        /// - Returns <c>false</c> if the endpoint responds with a non-success status code (4xx, 5xx).
+        /// - Throws exceptions (e.g., timeout, DNS failure) if the request cannot be completed.
+        /// - Uses a 5-second timeout to avoid hanging.
+        /// - All HTTP resources are properly disposed after the request completes.
+        /// </remarks>
+        /// <returns><c>true</c> if the endpoint is reachable and returns success; otherwise, <c>false</c>.</returns>
+        public static async Task<bool> IsAgentCdnAccessibleAsync(IWebProxy? webProxy = null)
+        {
+            // Configure the HttpClientHandler with the proxy if provided
+            using HttpClientHandler handler = new()
+            {
+                Proxy = webProxy,
+                UseProxy = webProxy is not null
+            };
+            handler.CheckCertificateRevocationList = true; // Check for certificate revocation
+            using HttpClient httpClient = new(handler);
+
+            // Construct a HEAD request to avoid downloading the full file
+            using HttpRequestMessage request = new(HttpMethod.Head, _agentPackageUri);
+
+            // Apply a 5-second timeout to prevent hanging
+            using CancellationTokenSource cts = new(TimeSpan.FromSeconds(5));
+
+            // Send the request and return whether the response status indicates success
+            HttpResponseMessage response = await httpClient.SendAsync(request, cts.Token);
+            return response.IsSuccessStatusCode;
+        }
+#nullable disable
     }
 
 #pragma warning disable CS0659 // Type overrides Object.Equals(object o) but does not override Object.GetHashCode()

src/Agent.Worker/JobExtension.cs

@@ -16,6 +16,7 @@
 using Agent.Sdk.Knob;
 using Newtonsoft.Json;
 using Microsoft.VisualStudio.Services.Agent.Worker.Telemetry;
+using Microsoft.Identity.Client.TelemetryCore.TelemetryClient;
 
 namespace Microsoft.VisualStudio.Services.Agent.Worker
 {
@@ -230,6 +231,32 @@ public async Task<List<IStep>> InitializeJob(IExecutionContext jobContext, Pipel
                         }
                     }
 
+                    // Check if the Agent CDN is accessible
+                    if (AgentKnobs.AgentCDNConnectivityFailWarning.GetValue(context).AsBoolean())
+                    {
+                        try
+                        {
+                            Trace.Verbose("Checking if the Agent CDN Endpoint (download.agent.dev.azure.com) is reachable");
+                            bool isAgentCDNAccessible = await PlatformUtil.IsAgentCdnAccessibleAsync(agentWebProxy.WebProxy);
+                            if (isAgentCDNAccessible)
+                            {
+                                context.Output("Agent CDN is accessible.");
+                            }
+                            else
+                            {
+                                context.Warning(StringUtil.Loc("AgentCdnAccessFailWarning"));
+                            }
+                            PublishAgentCDNAccessStatusTelemetry(context, isAgentCDNAccessible);
+                        }
+                        catch (Exception ex)
+                        {
+                            // Handles network-level or unexpected exceptions (DNS failure, timeout, etc.)
+                            context.Warning(StringUtil.Loc("AgentCdnAccessFailWarning"));
+                            PublishAgentCDNAccessStatusTelemetry(context, false);
+                            Trace.Error($"Exception when attempting a HEAD request to Agent CDN: {ex}");
+                        }
+                    }
+
                     if (PlatformUtil.RunningOnWindows)
                     {
                         // This is for internal testing and is not publicly supported. This will be removed from the agent at a later time.
@@ -755,6 +782,31 @@ private void PublishKnobsInfo(IExecutionContext jobContext)
             PublishTelemetry(jobContext, telemetryData, "KnobsStatus");
         }
 
+        private void PublishAgentCDNAccessStatusTelemetry(IExecutionContext context, bool isAgentCDNAccessible)
+        {
+            try
+            {
+                var telemetryData = new Dictionary<string, string>
+                {
+                    ["JobId"] = context?.Variables?.System_JobId?.ToString() ?? string.Empty,
+                    ["isAgentCDNAccessible"] = isAgentCDNAccessible.ToString()
+                };
+
+                var cmd = new Command("telemetry", "publish")
+                {
+                    Data = JsonConvert.SerializeObject(telemetryData)
+                };
+                cmd.Properties["area"] = "PipelinesTasks";
+                cmd.Properties["feature"] = "CDNConnectivityCheck";
+
+                PublishTelemetry(context, telemetryData, "AgentCDNAccessStatus");
+            }
+            catch (Exception ex)
+            {
+                Trace.Verbose($"Ignoring exception during 'AgentCDNAccessStatus' telemetry publish: '{ex.Message}'");
+            }
+        }
+
         private void PublishTelemetry(IExecutionContext context, Dictionary<string, string> telemetryData, string feature)
         {
             try

src/Misc/layoutbin/en-US/strings.json

@@ -10,6 +10,7 @@
   "AddEnvironmentVMResourceTags": "Environment Virtual Machine resource tags? (Y/N)",
   "AgentAddedSuccessfully": "Successfully added the agent",
   "AgentAlreadyInsideContainer": "Container feature is not supported when agent is already running inside container. Please reference documentation (https://go.microsoft.com/fwlink/?linkid=875268)",
+  "AgentCdnAccessFailWarning": "Action Required: Azure Pipelines Agent cannot reach the new CDN URL. Allowlist 'download.agent.dev.azure.com' now to prevent pipeline failures. Details: https://devblogs.microsoft.com/devops/cdn-domain-url-change-for-agents-in-pipelines/",
   "AgentDoesNotSupportContainerFeatureRhel6": "Agent does not support the container feature on Red Hat Enterprise Linux 6 or CentOS 6.",
   "AgentDowngrade": "Downgrading agent to a lower version. This is usually due to a rollback of the currently published agent for a bug fix. To disable this behavior, set environment variable AZP_AGENT_DOWNGRADE_DISABLED=true before launching your agent.",
   "AgentExit": "Agent will exit shortly for update, should back online within 10 seconds.",