Merge branch 'develop' into librbd

Author: GowthamShanmugam

tendrl/ceph_integration/ceph.py

@@ -7,6 +7,7 @@
 import socket
 import struct
 import subprocess
+import sys
 import tempfile
 import time
 import rbd
@@ -29,7 +30,7 @@
 # always
 # present, although this is the case on a nicely ceph-deploy'd system
 RADOS_NAME = 'client.admin'
-
+SRC_DIR = '/etc/ceph'
 
 def fire_event(data, tag):
     return {tag: data}
@@ -53,6 +54,18 @@ class AdminSocketError(MonitoringError):
     pass
 
 
+def get_ceph_version():
+    result = ceph_command(None, ['--version'])
+    try:
+        version = result['out'].split(' ')[2]
+        return version
+    except (KeyError, AttributeError, IndexError) as ex:
+        sys.stdout.write("Error getting ceph --version")
+        sys.stdout.write(str(ex))
+        raise ex
+
+
+
 def rados_command(cluster_handle, prefix, args=None, decode=True):
     """Safer wrapper for ceph_argparse.json_command, which raises
 
@@ -280,8 +293,12 @@ def rados_commands(fsid, cluster_name, commands):
     import rados
 
     # Open a RADOS session
+    if cluster_name is None:
+        cluster_name = "ceph"
+
+    _conf_file = os.path.join(SRC_DIR, cluster_name  + ".conf")
     cluster_handle = rados.Rados(
-        name=RADOS_NAME, clustername=cluster_name, conffile=''
+        name=RADOS_NAME, clustername=cluster_name, conffile=_conf_file
     )
     cluster_handle.connect()
 
@@ -529,8 +546,12 @@ def get_cluster_object(cluster_name, sync_type):
     assert sync_type in SYNC_TYPES
 
     # Open a RADOS session
+    if cluster_name is None:
+        cluster_name = "ceph"
+
+    _conf_file = os.path.join(SRC_DIR, cluster_name  + ".conf")
     cluster_handle = rados.Rados(
-        name=RADOS_NAME, clustername=cluster_name, conffile=''
+        name=RADOS_NAME, clustername=cluster_name, conffile=_conf_file
     )
     cluster_handle.connect()
 
@@ -698,10 +719,12 @@ def get_heartbeats():
             if "client" in filename:
                 continue
             service_data = service_status(filename)
-        except (rados.Error, MonitoringError):
+        except (rados.Error, MonitoringError) as ex:
             # Failed to get info for this service, stale socket or
             # unresponsive, exclude it from report
-            pass
+            sys.stdout.write("Error getting ceph service status from admin "
+                             "socket %s" % filename)
+            sys.stdout.write(str(ex))
         else:
             if not service_data:
                 continue
@@ -713,51 +736,46 @@ def get_heartbeats():
                 # A mon in quorum is elegible to emit a cluster heartbeat
                 mon_sockets[service_data['fsid']] = filename
 
-    # Installed Ceph version (as oppose to per-service running ceph version)
-    try:
-        ceph_version_str = subprocess.check_output(
-            "rpm -qa | grep ceph-[0-1]", shell=True
-        )
-        ceph_version_str = ceph_version_str.split("-")[1]
-    except subprocess.CalledProcessError:
-        ceph_version_str = None
-    if ceph_version_str:
-        ceph_version = ceph_version_str
-    else:
-        ceph_version = None
+    ceph_version = get_ceph_version()
 
     # For each ceph cluster with an in-quorum mon on this node, interrogate
     # the cluster
     cluster_heartbeat = {}
     for fsid, socket_path in mon_sockets.items():
+        cluster_handle = None
         try:
+            _conf_file = os.path.join(SRC_DIR, fsid_names[fsid] + ".conf")
             cluster_handle = rados.Rados(
-                name=RADOS_NAME, clustername=fsid_names[fsid], conffile=''
+                name=RADOS_NAME, clustername=fsid_names[fsid],
+                conffile=_conf_file
             )
             cluster_handle.connect()
             cluster_heartbeat[fsid] = cluster_status(
                 cluster_handle, fsid_names[fsid]
             )
-        except (rados.Error, MonitoringError):
-            # Something went wrong getting data for this cluster, exclude it
-            # from our report
-            pass
+        except (rados.Error, MonitoringError) as ex:
+            # Something went wrong getting data for this cluster
+            sys.stdout.write("Error fetching ceph (fsid: %s) cluster maps "
+                             "from "
+                             "admin socket %s" % (fsid_names[fsid],
+                                                  socket_path))
+            sys.stdout.write(str(ex))
+            raise ex
+        finally:
+            if cluster_handle:
+                cluster_handle.shutdown()
 
-    cluster_handle.shutdown()
     return ceph_version, cluster_heartbeat
 
 
 def service_status(socket_path):
     """Given an admin socket path, learn all we can about that service
 
     """
-    try:
-        cluster_name, service_type, service_id = \
-            re.match(
-                "^(.+?)-(.+?)\.(.+)\.asok$",
-                os.path.basename(socket_path)).groups()
-    except AttributeError:
-        return None
+    cluster_name, service_type, service_id = \
+        re.match(
+            "^(.+?)-(.+?)\.(.+)\.asok$",
+            os.path.basename(socket_path)).groups()
 
     status = None
     # Interrogate the service for its FSID
@@ -890,11 +908,20 @@ def _heartbeat(fsid):
 
 
 def heartbeat(fsid=None):
+    try:
+        import rados
+    except ImportError:
+        # Ceph isn't installed, report no services or clusters
+        return None, {}
+
     try:
         return _heartbeat(fsid)
-    except Exception:
-        # TODO(Rohan): Tackle this later
-        pass
+    except Exception as ex:
+        sys.stdout.write("Error getting heartbeat for ceph cluster fsid %s"
+                         % fsid)
+        sys.stdout.write(str(ex))
+        if type(ex) in [rados.Error, MonitoringError, AdminSocketError]:
+            raise ex
 
 
 def json_load_byteified(file_handle):

tendrl/ceph_integration/sds_sync/__init__.py

@@ -671,10 +671,13 @@ def _get_rbds(self, pool_name):
     def _get_utilization_data(self):
         from ceph_argparse import json_command
         import rados
+        _conf_file = os.path.join("/etc/ceph",
+                                  NS.tendrl_context.cluster_name + ".conf")
+        # TODO (shtripat) use ceph.ceph_command instead of rados/json_command
         cluster_handle = rados.Rados(
             name=ceph.RADOS_NAME,
-            clustername=self.name,
-            conffile=''
+            clustername=NS.tendrl_context.cluster_name,
+            conffile=_conf_file
         )
         cluster_handle.connect()
         prefix = 'df'
@@ -685,6 +688,7 @@ def _get_utilization_data(self):
             timeout=ceph.RADOS_TIMEOUT
         )
         if ret != 0:
+            cluster_handle.shutdown()
             raise rados.Error(outs)
         else:
             outbuf = outbuf.replace('RAW USED', 'RAW_USED')
@@ -695,6 +699,8 @@ def _get_utilization_data(self):
             cluster_stat = {}
             pool_stat = []
             pool_stat_available = False
+            cluster_handle.shutdown()
+
             while index < len(lines):
                 line = lines[index]
                 if line == "" or line == '\n':
@@ -823,6 +829,7 @@ def _get_utilization_data(self):
                     dict['pcnt_used'] = pool_fields[pool_pcnt_used_idx]
                     pool_stat.append(dict)
                 index += 1
+            
             return {'cluster': cluster_stat, 'pools': pool_stat}
 
     def _idx_in_list(self, list, str):