qa: add filesystem sync stuck test support

This will test the sync of the filesystem, which maybe stuck for
at most 5 seconds. This was because the related code will wait
for all the unsafe requests to get safe reply from MDSes, but the
MDSes just think that it's unnecessary to flush the mdlog immediately
after early reply, and the mdlog will be flushed every 5 seconds
in the tick thread.

This should have been fixed in kclient and libcephfs by triggering
mdlog flush before waiting requests' safe reply.

Fixes: https://tracker.ceph.com/issues/55283
Signed-off-by: Xiubo Li <xiubli@redhat.com>
(cherry picked from commit b6fc5480f6ba6352fa72062e1376d0dd6b9074cd)

diff --git a/qa/tasks/cephfs/test_misc.py b/qa/tasks/cephfs/test_misc.py
index a6cb3da4efce9a47b2eaf48836e8ba6c74ed8e5f..33c1142ff46a73eebd1fea6692da95c8f8304cfc 100644 (file)
--- a/qa/tasks/cephfs/test_misc.py
+++ b/qa/tasks/cephfs/test_misc.py
@@ -7,6 +7,7 @@ import errno
 import time
 import json
 import logging
+import os
 
 log = logging.getLogger(__name__)
 
@@ -200,6 +201,36 @@ class TestMisc(CephFSTestCase):
         assert info['path'] == "/foo"
 
 
+    def test_filesystem_sync_stuck_for_around_5s(self):
+        """
+        To check whether the fsync will be stuck to wait for the mdlog to be
+        flushed for at most 5 seconds.
+        """
+
+        dir_path = "fsync_do_not_wait_mdlog_testdir"
+        self.mount_a.run_shell(["mkdir", dir_path])
+
+        # run create/delete directories and test the sync time duration
+        for i in range(300):
+            for j in range(5):
+                self.mount_a.run_shell(["mkdir", os.path.join(dir_path, f"{i}_{j}")])
+            start = time.time()
+            self.mount_a.run_shell(["sync"])
+            duration = time.time() - start
+            log.info(f"mkdir i = {i}, duration = {duration}")
+            self.assertLess(duration, 4)
+
+            for j in range(5):
+                self.mount_a.run_shell(["rm", "-rf", os.path.join(dir_path, f"{i}_{j}")])
+            start = time.time()
+            self.mount_a.run_shell(["sync"])
+            duration = time.time() - start
+            log.info(f"rmdir i = {i}, duration = {duration}")
+            self.assertLess(duration, 4)
+
+        self.mount_a.run_shell(["rm", "-rf", dir_path])
+
+
 class TestCacheDrop(CephFSTestCase):
     CLIENTS_REQUIRED = 1