X-Git-Url: https://git.proxmox.com/?a=blobdiff_plain;f=ceph%2Fqa%2Ftasks%2Fceph_manager.py;h=2516043217b7d63cdb0085c239e10257f652c571;hb=f6b5b4d738b87d88d2de35127b6b0e41eae2a272;hp=c058735aa8a27d9aee0b0a5869e62ee0f37f1e30;hpb=12732ca2e80d168d344a265acffc1fbd1fa1f1b5;p=ceph.git

diff --git a/ceph/qa/tasks/ceph_manager.py b/ceph/qa/tasks/ceph_manager.py
index c058735aa..251604321 100644
--- a/ceph/qa/tasks/ceph_manager.py
+++ b/ceph/qa/tasks/ceph_manager.py
@@ -2467,6 +2467,36 @@ class CephManager:
         pgs = self.get_pg_stats()
         return self._get_num_active_down(pgs) == len(pgs)
 
+    def dump_pgs_not_active_clean(self):
+        """
+        Dumps all pgs that are not active+clean
+        """
+        pgs = self.get_pg_stats()
+        for pg in pgs:
+           if pg['state'] != 'active+clean':
+             self.log('PG %s is not active+clean' % pg['pgid'])
+             self.log(pg)
+
+    def dump_pgs_not_active_down(self):
+        """
+        Dumps all pgs that are not active or down
+        """
+        pgs = self.get_pg_stats()
+        for pg in pgs:
+           if 'active' not in pg['state'] and 'down' not in pg['state']:
+             self.log('PG %s is not active or down' % pg['pgid'])
+             self.log(pg)
+
+    def dump_pgs_not_active(self):
+        """
+        Dumps all pgs that are not active
+        """
+        pgs = self.get_pg_stats()
+        for pg in pgs:
+           if 'active' not in pg['state']:
+             self.log('PG %s is not active' % pg['pgid'])
+             self.log(pg)
+
     def wait_for_clean(self, timeout=1200):
         """
         Returns true when all pgs are clean.
@@ -2482,11 +2512,10 @@ class CephManager:
                 else:
                     self.log("no progress seen, keeping timeout for now")
                     if time.time() - start >= timeout:
-                        self.log('dumping pgs')
-                        out = self.raw_cluster_cmd('pg', 'dump')
-                        self.log(out)
+                        self.log('dumping pgs not clean')
+                        self.dump_pgs_not_active_clean()
                         assert time.time() - start < timeout, \
-                            'failed to become clean before timeout expired'
+                            'wait_for_clean: failed before timeout expired'
             cur_active_clean = self.get_num_active_clean()
             if cur_active_clean != num_active_clean:
                 start = time.time()
@@ -2568,11 +2597,10 @@ class CephManager:
                     if now - start >= timeout:
                         if self.is_recovered():
                             break
-                        self.log('dumping pgs')
-                        out = self.raw_cluster_cmd('pg', 'dump')
-                        self.log(out)
+                        self.log('dumping pgs not recovered yet')
+                        self.dump_pgs_not_active_clean()
                         assert now - start < timeout, \
-                            'failed to recover before timeout expired'
+                            'wait_for_recovery: failed before timeout expired'
             cur_active_recovered = self.get_num_active_recovered()
             if cur_active_recovered != num_active_recovered:
                 start = time.time()
@@ -2590,11 +2618,10 @@ class CephManager:
         while not self.is_active():
             if timeout is not None:
                 if time.time() - start >= timeout:
-                    self.log('dumping pgs')
-                    out = self.raw_cluster_cmd('pg', 'dump')
-                    self.log(out)
+                    self.log('dumping pgs not active')
+                    self.dump_pgs_not_active()
                     assert time.time() - start < timeout, \
-                        'failed to recover before timeout expired'
+                        'wait_for_active: failed before timeout expired'
             cur_active = self.get_num_active()
             if cur_active != num_active:
                 start = time.time()
@@ -2613,11 +2640,10 @@ class CephManager:
         while not self.is_active_or_down():
             if timeout is not None:
                 if time.time() - start >= timeout:
-                    self.log('dumping pgs')
-                    out = self.raw_cluster_cmd('pg', 'dump')
-                    self.log(out)
+                    self.log('dumping pgs not active or down')
+                    self.dump_pgs_not_active_down()
                     assert time.time() - start < timeout, \
-                        'failed to recover before timeout expired'
+                        'wait_for_active_or_down: failed before timeout expired'
             cur_active_down = self.get_num_active_down()
             if cur_active_down != num_active_down:
                 start = time.time()
@@ -2667,11 +2693,10 @@ class CephManager:
         while not self.is_active():
             if timeout is not None:
                 if time.time() - start >= timeout:
-                    self.log('dumping pgs')
-                    out = self.raw_cluster_cmd('pg', 'dump')
-                    self.log(out)
+                    self.log('dumping pgs not active')
+                    self.dump_pgs_not_active()
                     assert time.time() - start < timeout, \
-                        'failed to become active before timeout expired'
+                        'wait_till_active: failed before timeout expired'
             time.sleep(3)
         self.log("active!")