Set nodename as hostname prior to post for el class distros

Initial collective join combined
with the orderly collective startup
hit a chicken and egg problem.

Disable initting on first enrollment
to let enrollment drive
that specific initialization to
restore behavior.

confluent_client/bin/nodebmcreset

@@ -47,14 +47,14 @@ exitcode = 0
 errorNodes = set([])
 session.stop_if_noderange_over(noderange, options.maxnodes)
 success = session.simple_noderange_command(noderange, 'configuration/management_controller/reset', 'reset', key='state', errnodes=errorNodes)   # = 0 if successful
-if success != 0:
-    sys.exit(success)
 
 # Determine which nodes were successful and print them
 
 allNodes = set([])
 
 for node in session.read('/noderange/{0}/nodes/'.format(noderange)):
+    if 'error' in node and success != 0:
+        sys.exit(success)
     allNodes.add(node['item']['href'].replace("/", ""))
 
 goodNodes = allNodes - errorNodes

confluent_osdeploy/el7/profiles/default/scripts/prechroot.sh

@@ -12,6 +12,8 @@ export confluent_mgr confluent_profile nodename
 cp -a /etc/confluent /mnt/sysimage/etc
 chmod -R og-rwx /mnt/sysimage/etc/confluent
 cp /tmp/functions /mnt/sysimage/etc/confluent/
+hostnamectl set-hostname $nodename
+cp /etc/hostname /mnt/sysimage/etc/hostname
 . /tmp/functions
 if [ -f /tmp/cryptboot ]; then
     cp /tmp/cryptboot /mnt/sysimage/tmp/

confluent_osdeploy/el8/profiles/default/scripts/prechroot.sh

@@ -12,6 +12,8 @@ export confluent_mgr confluent_profile nodename
 cp -a /etc/confluent /mnt/sysimage/etc
 chmod -R og-rwx /mnt/sysimage/etc/confluent
 cp /tmp/functions /mnt/sysimage/etc/confluent/
+hostnamectl set-hostname $nodename
+cp /etc/hostname /mnt/sysimage/etc/hostname
 . /tmp/functions
 if [ -f /tmp/cryptboot ]; then
     cp /tmp/cryptboot /mnt/sysimage/tmp/

confluent_server/confluent/collective/manager.py

@@ -244,6 +244,7 @@ def get_myname():
 def handle_connection(connection, cert, request, local=False):
     global currentleader
     global retrythread
+    global initting
     connection.settimeout(5)
     operation = request['operation']
     if cert:
@@ -395,6 +396,11 @@ def handle_connection(connection, cert, request, local=False):
             tlvdata.send(connection, {'error': 'Invalid token'})
             connection.close()
             return
+        if not list(cfm.list_collective()):
+            # First enrollment of a collective, since the collective doesn't
+            # quite exist, then set initting false to let the enrollment action
+            # drive this particular initialization
+            initting = False
         myrsp = base64.b64encode(myrsp)
         fprint = util.get_fingerprint(cert)
         myfprint = util.get_fingerprint(mycert)

confluent_server/confluent/noderange.py

@@ -106,7 +106,7 @@ class NodeRange(object):
         self.endpage = None
         self.cfm = config
         try:
-            elements = _parser.parseString("(" + noderange + ")").asList()[0]
+            elements = _parser.parseString("(" + noderange + ")", parseAll=True).asList()[0]
         except pp.ParseException as pe:
             raise Exception("Invalid syntax")
         if noderange[0] in ('<', '>'):

confluent_server/confluent/sshutil.py

@@ -34,7 +34,7 @@ def normalize_uid():
 
 def assure_agent():
     if sshver() <= 7.6:
-        return
+        return False
     global agent_pid
     if agent_pid is None:
         sai = subprocess.check_output(['ssh-agent'])
@@ -51,6 +51,7 @@ def assure_agent():
             if k == 'SSH_AGENT_PID':
                 agent_pid = v
             os.environ[k] = v
+    return True
 
 def get_passphrase():
     if sshver() <= 7.6:
@@ -90,9 +91,11 @@ def initialize_ca():
 
 
 def prep_ssh_key(keyname):
-    assure_agent()
     if keyname in ready_keys:
         return
+    if not assure_agent():
+        ready_keys[keyname] = 1
+        return
     tmpdir = tempfile.mkdtemp()
     try:
         askpass = os.path.join(tmpdir, 'askpass.sh')
@@ -121,8 +124,9 @@ def sign_host_key(pubkey, nodename, principals=()):
         principals.add(nodename)
         principals = ','.join(sorted(principals))
         flags = '-Us' if sshver() > 7.6 else '-s'
+        keyname = '/etc/confluent/ssh/ca.pub' if flags == '-Us' else '/etc/confluent/ssh/ca'
         subprocess.check_call(
-            ['ssh-keygen', flags, '/etc/confluent/ssh/ca.pub', '-I', nodename,
+            ['ssh-keygen', flags, keyname, '-I', nodename,
              '-n', principals, '-h', pkeyname])
         certname = pkeyname.replace('.pub', '-cert.pub')
         with open(certname) as cert: