test: fix check for major page faults in test_balloon.py::test_stats

Major page fault occur when a page fault can only be satisfied via disk
I/O [1]. In `test_stats`, we asserted that the number of major page
faults increased during execution of `make_guest_dirty_memory`, but this
cannot happen, as it uses MAP_ANONYMOUS, and thus will not trigger any
major page faults. The reason the test used to work in the past is that
it was not waiting for the VM to boot, so the first reading of the
balloon statistics happened before the boot process was finished.
However, during boot we _do_ expect major faults to happen, as things
need to be read from the rootfs.

Fix the test to instead just assert a non-zero number of major faults
happened during boot.

[1]: https://en.wikipedia.org/wiki/Page_fault#Major

Signed-off-by: Patrick Roy <[email protected]>

Author: roypat

tests/integration_tests/functional/test_balloon.py

@@ -341,17 +341,28 @@ def test_stats(uvm_plain_any):
 
     # Add a memory balloon with stats enabled.
     test_microvm.api.balloon.put(
-        amount_mib=0, deflate_on_oom=True, stats_polling_interval_s=1
+        amount_mib=0,
+        deflate_on_oom=True,
+        stats_polling_interval_s=STATS_POLLING_INTERVAL_S,
     )
 
     # Start the microvm.
     test_microvm.start()
     test_microvm.wait_for_up()
     firecracker_pid = test_microvm.firecracker_pid
 
+    # Give Firecracker enough time to poll the stats at least once post-boot
+    time.sleep(STATS_POLLING_INTERVAL_S * 2)
+
     # Get an initial reading of the stats.
     initial_stats = test_microvm.api.balloon_stats.get().json()
 
+    # Major faults happen when a page fault has to be satisfied from disk. They are not
+    # triggered by our `make_guest_dirty_memory` workload, as it uses MAP_ANONYMOUS, which
+    # only triggers minor faults. However, during the boot process, things are read from the
+    # rootfs, so we should at least see a non-zero number of major faults.
+    assert initial_stats["major_faults"] > 0
+
     # Dirty 10MB of pages.
     make_guest_dirty_memory(test_microvm.ssh, amount_mib=10)
     time.sleep(1)
@@ -361,7 +372,6 @@ def test_stats(uvm_plain_any):
     # Make sure that the stats catch the page faults.
     after_workload_stats = test_microvm.api.balloon_stats.get().json()
     assert initial_stats.get("minor_faults", 0) < after_workload_stats["minor_faults"]
-    assert initial_stats.get("major_faults", 0) < after_workload_stats["major_faults"]
 
     # Now inflate the balloon with 10MB of pages.
     test_microvm.api.balloon.patch(amount_mib=10)