fix(vmm): Emit startup metrics after server start

api_server.process_startup_time_us and process_startup_time_cpu_us
should be reported right after the server starts.

Signed-off-by: Takahiro Itazuri <[email protected]>

Author: zulinx86

CHANGELOG.md

@@ -48,6 +48,11 @@ and this project adheres to
   the UFFD Unix domain socket open to prevent the race condition between the
   guest memory mappings message and the shutdown event that was sometimes
   causing arrival of an empty message on the UFFD handler side.
+- [#5143](https://github.com/firecracker-microvm/firecracker/pull/5143): Fixed
+  to report `process_startup_time_us` and `process_startup_time_cpu_us` metrics
+  for `api_server` right after the API server starts, while previously reported
+  before applying seccomp filter and starting the API server. Users may observe
+  a bit longer startup time metrics.
 
 ## [1.11.0]
 

src/firecracker/src/api_server/mod.rs

@@ -70,11 +70,6 @@ impl ApiServer {
         // Set the api payload size limit.
         server.set_payload_max_size(api_payload_limit);
 
-        // Store process start time metric.
-        process_time_reporter.report_start_time();
-        // Store process CPU start time metric.
-        process_time_reporter.report_cpu_start_time();
-
         // Load seccomp filters on the API thread.
         // Execution panics if filters cannot be loaded, use --no-seccomp if skipping filters
         // altogether is the desired behaviour.
@@ -88,6 +83,11 @@ impl ApiServer {
         server.start_server().expect("Cannot start HTTP server");
         info!("API server started.");
 
+        // Store process start time metric.
+        process_time_reporter.report_start_time();
+        // Store process CPU start time metric.
+        process_time_reporter.report_cpu_start_time();
+
         loop {
             let request_vec = match server.requests() {
                 Ok(vec) => vec,