autoconf: don't remove disconnected hosts from nagios

 Currently auto-config removes all the hosts with disconnected
status in 'peer status' from nagios configuration. But that
is not correct as a host can goto disconnected status for
many differnt reasons and we should not remove the configuration.

 Similarly disconnected hosts need not be added to nagios
configuration during auto discovery.

  Fixing this by ingnoring all the disconnected hosts in delta
calculation.

Note: Host name can't be used to identify disconnected hosts
as we may not be able to communicate the disconnected host
to get the host name. Hence Host UUID in peer status is
used to intentify the host.

Bug-Url: https://bugzilla.redhat.com/1091170

Change-Id: I3fe730d2545f8cbc8224c576712b09bb4c16e712
Signed-off-by: Ramesh Nachimuthu <rnachimu@redhat.com>
Reviewed-on: http://review.gluster.org/7677
Reviewed-by: Sahina Bose <sabose@redhat.com>

2 files changed, 50 insertions, 33 deletions

diff --git a/plugins/config_generator.py b/plugins/config_generator.py
index 465eb61..ed32912 100644
--- a/plugins/config_generator.py
+++ b/plugins/config_generator.py
@@ -21,6 +21,7 @@
 from pynag import Model
 
 import server_utils
+from glusternagios.glustercli import HostStatus
 
 """
 Change mode helps to identify the change in the defintion.
@@ -41,7 +42,7 @@ class GlusterNagiosConfManager:
 
     #Create nagios host configuration with the given attributes
     def createHost(self, hostName, alias, template,
-                   address, hostGroups, checkCommand, services):
+                   address, hostGroups, checkCommand, services, uuid):
         host = {}
         host['host_name'] = hostName
         host['alias'] = alias
@@ -55,6 +56,8 @@ class GlusterNagiosConfManager:
         #aggregate all the host services under the host
         if services:
             host['host_services'] = services
+        if uuid:
+            host['_HOST_UUID'] = uuid
         return host
 
     def __createVolumeUtilizationService(self, volume, clusterName):
@@ -230,18 +233,19 @@ class GlusterNagiosConfManager:
         #Create host config for Gluster cluster with volume related services
         clusterHostConfig = self.createHost(
             cluster['name'], cluster['name'], "gluster-cluster",
-            cluster['name'], None, None, clusterServices)
+            cluster['name'], None, None, clusterServices, None)
         hostsConfigs.append(clusterHostConfig)
         #Create host config for all hosts in the cluster with brick related
         #services
         for host in cluster['hosts']:
-            brickServices = self.createBrickServices(host)
-            hostGroups = "gluster_hosts,%s" % (cluster['name'])
-            hostConfig = self.createHost(
-                host['hostname'], host['hostname'], "gluster-host",
-                host['hostip'], hostGroups, None, brickServices)
-            hostsConfigs.append(hostConfig)
-
+            if host['status'] == HostStatus.CONNECTED:
+                brickServices = self.createBrickServices(host)
+                hostGroups = "gluster_hosts,%s" % (cluster['name'])
+                hostConfig = self.createHost(
+                    host['hostname'], host['hostname'], "gluster-host",
+                    host['hostip'], hostGroups, None, brickServices,
+                    host.get('uuid'))
+                hostsConfigs.append(hostConfig)
         hostGroup["_hosts"] = hostsConfigs
         return hostGroup
 
diff --git a/plugins/discovery.py b/plugins/discovery.py
index 2f1b9f6..3e3b0a5 100755
--- a/plugins/discovery.py
+++ b/plugins/discovery.py
@@ -24,6 +24,7 @@ import shutil
 import sys
 
 from glusternagios import utils
+from glusternagios.glustercli import HostStatus
 from config_generator import GlusterNagiosConfManager
 import server_utils
 import submit_external_command
@@ -111,16 +112,17 @@ def discoverCluster(hostip, cluster):
     #Add the ip address of the root node given by the user to the peer list
     hostlist[0]['hostip'] = hostip
     for host in hostlist:
-        #Get host names
-        hostDetails = execNRPECommand(host['hostip'], "discoverhostparams")
-        host.update(hostDetails)
-        #Get the list of bricks for this host and add to dictionary
-        host['bricks'] = []
-        for volume in componentlist['volumes']:
-            for brick in volume['bricks']:
-                if brick['hostUuid'] == host['uuid']:
-                    brick['volumeName'] = volume['name']
-                    host['bricks'].append(brick)
+        #Get host names for all the connected hosts
+        if host['status'] == HostStatus.CONNECTED:
+            hostDetails = execNRPECommand(host['hostip'], "discoverhostparams")
+            host.update(hostDetails)
+            #Get the list of bricks for this host and add to dictionary
+            host['bricks'] = []
+            for volume in componentlist['volumes']:
+                for brick in volume['bricks']:
+                    if brick['hostUuid'] == host['uuid']:
+                        brick['volumeName'] = volume['name']
+                        host['bricks'].append(brick)
     clusterdata['hosts'] = hostlist
     clusterdata['volumes'] = componentlist['volumes']
     clusterdata['name'] = cluster
@@ -143,10 +145,11 @@ def setHostNameWithIP(clusterdata):
 def isHostsNamesUnique(clusterdata):
     hostnames = {}
     for host in clusterdata['hosts']:
-        if hostnames.get(host['hostname']) is None:
-            hostnames[host['hostname']] = host['hostip']
-        else:
-            return False
+        if host.get('status') == HostStatus.CONNECTED:
+            if hostnames.get(host.get('hostname')) is None:
+                hostnames[host.get('hostname')] = host['hostip']
+            else:
+                return False
     return True
 
 
@@ -244,29 +247,30 @@ def findAddUpdateHosts(hosts):
 
 
 #Find deleted hosts in the given cluster.
-def findDeletedHosts(hostgroup, hosts):
+def findDeletedHosts(hostgroup, hosts, ignoredHosts):
     deletedHosts = []
     hostConfigs = server_utils.getHostConfigsForCluster(hostgroup)
     for hostConfig in hostConfigs:
-        host = findHostInList(hosts, hostConfig['host_name'])
-        if host is None:
-            deletedHosts.append({'host_name': hostConfig['host_name'],
-                                 'changeMode': CHANGE_MODE_REMOVE})
+        if hostConfig.get('_HOST_UUID') not in ignoredHosts:
+            host = findHostInList(hosts, hostConfig['host_name'])
+            if host is None:
+                deletedHosts.append({'host_name': hostConfig['host_name'],
+                                     'changeMode': CHANGE_MODE_REMOVE})
     return deletedHosts
 
 
 #Find Added/Deleted/Updated hosts in cluster
-def findHostDelta(clusterConfig):
+def findHostDelta(clusterConfig, ignoredHosts):
     hostDelta = []
     updated = findAddUpdateHosts(clusterConfig['_hosts'])
     hostDelta.extend(updated)
     hostDelta.extend(findDeletedHosts(clusterConfig['hostgroup_name'],
-                                      clusterConfig['_hosts']))
+                                      clusterConfig['_hosts'], ignoredHosts))
     return hostDelta
 
 
 #Find changes to the cluster
-def findDelta(clusterConfig):
+def findDelta(clusterConfig, ignoredHosts):
     delta = {}
     delta['hostgroup_name'] = clusterConfig['hostgroup_name']
     delta['alias'] = clusterConfig['alias']
@@ -277,7 +281,7 @@ def findDelta(clusterConfig):
         delta['_hosts'] = clusterConfig['_hosts']
         return delta
 
-    hostDelta = findHostDelta(clusterConfig)
+    hostDelta = findHostDelta(clusterConfig, ignoredHosts)
     delta['_hosts'] = hostDelta
     if hostDelta:
         delta['changeMode'] = CHANGE_MODE_UPDATE
@@ -449,6 +453,14 @@ def sendCustomNotification(cluster, summary):
     submit_external_command.submitExternalCommand(cmdStr)
 
 
+def getAllNonConnectedHosts(hostList):
+    nonConnectedHosts = []
+    for host in hostList:
+        if host.get('status') != HostStatus.CONNECTED:
+            nonConnectedHosts.append(host.get('uuid'))
+    return nonConnectedHosts
+
+
 if __name__ == '__main__':
     args = parse_input()
     clusterdata = discoverCluster(args.hostip, args.cluster)
@@ -457,7 +469,8 @@ if __name__ == '__main__':
     if args.force:
         clusterDelta['changeMode'] = CHANGE_MODE_ADD
     else:
-        clusterDelta = findDelta(clusterDelta)
+        nonConnectedHosts = getAllNonConnectedHosts(clusterdata['hosts'])
+        clusterDelta = findDelta(clusterDelta, nonConnectedHosts)
 
     if clusterDelta.get('changeMode') is None:
         print "Cluster configurations are in sync"