branching
[tests.git] / system / TestPlc.py
index db196f3..095762a 100644 (file)
@@ -3,10 +3,9 @@ import os, os.path
 import datetime
 import time
 import sys
-import xmlrpclib
-import datetime
 import traceback
 from types import StringTypes
+import socket
 
 import utils
 from TestSite import TestSite
@@ -14,94 +13,123 @@ from TestNode import TestNode
 from TestUser import TestUser
 from TestKey import TestKey
 from TestSlice import TestSlice
+from TestSliver import TestSliver
 from TestBox import TestBox
+from TestSsh import TestSsh
+from TestApiserver import TestApiserver
 
-# step methods must take (self, options) and return a boolean
+# step methods must take (self) and return a boolean (options is a member of the class)
 
-def standby(minutes):
-        utils.header('Entering StandBy for %d mn'%minutes)
+def standby(minutes,dry_run):
+    utils.header('Entering StandBy for %d mn'%minutes)
+    if dry_run:
+        print 'dry_run'
+    else:
         time.sleep(60*minutes)
-        return True
+    return True
 
 def standby_generic (func):
-    def actual(self,options):
+    def actual(self):
         minutes=int(func.__name__.split("_")[1])
-        return standby(minutes)
+        return standby(minutes,self.options.dry_run)
+    return actual
+
+def node_mapper (method):
+    def actual(self):
+        overall=True
+        node_method = TestNode.__dict__[method.__name__]
+        for site_spec in self.plc_spec['sites']:
+            test_site = TestSite (self,site_spec)
+            for node_spec in site_spec['nodes']:
+                test_node = TestNode (self,test_site,node_spec)
+                if not node_method(test_node): overall=False
+        return overall
     return actual
 
+def slice_mapper_options (method):
+    def actual(self):
+        overall=True
+        slice_method = TestSlice.__dict__[method.__name__]
+        for slice_spec in self.plc_spec['slices']:
+            site_spec = self.locate_site (slice_spec['sitename'])
+            test_site = TestSite(self,site_spec)
+            test_slice=TestSlice(self,test_site,slice_spec)
+            if not slice_method(test_slice,self.options): overall=False
+        return overall
+    return actual
+
+SEP='<sep>'
+
 class TestPlc:
 
-    def __init__ (self,plc_spec):
+    default_steps = ['uninstall','install','install_rpm', 
+                     'configure', 'start', SEP,
+                     'store_keys', 'clear_known_hosts', 'initscripts', SEP,
+                     'sites', 'nodes', 'slices', 'nodegroups', SEP,
+                     'init_node','bootcd', 'configure_qemu', 'export_qemu',
+                     'kill_all_qemus', 'reinstall_node','start_node', SEP,
+                     'nodes_booted', 'nodes_ssh', 'check_slice',
+                     'check_initscripts', 'check_tcp',SEP,
+                     'force_gather_logs', 'force_kill_qemus', 'force_record_tracker','force_free_tracker' ]
+    other_steps = [ 'stop_all_vservers','fresh_install', 'cache_rpm', 'stop', SEP,
+                    'clean_initscripts', 'clean_sites', 'clean_nodes', 
+                    'clean_slices', 'clean_keys', SEP,
+                    'show_boxes', 'list_all_qemus', 'list_qemus', SEP,
+                    'db_dump' , 'db_restore', ' cleanup_tracker',
+                    'standby_1 through 20'
+                    ]
+
+    @staticmethod
+    def printable_steps (list):
+        return " ".join(list).replace(" "+SEP+" "," \\\n")
+    @staticmethod
+    def valid_step (step):
+        return step != SEP
+
+    def __init__ (self,plc_spec,options):
        self.plc_spec=plc_spec
-       self.path=os.path.dirname(sys.argv[0])
+        self.options=options
+       self.test_ssh=TestSsh(self.plc_spec['hostname'],self.options.buildname)
         try:
             self.vserverip=plc_spec['vserverip']
             self.vservername=plc_spec['vservername']
             self.url="https://%s:443/PLCAPI/"%plc_spec['vserverip']
             self.vserver=True
         except:
-            self.vserver=False
-            self.url="https://%s:443/PLCAPI/"%plc_spec['hostname']
-        utils.header('Using API url %s'%self.url)
-       self.server=xmlrpclib.Server(self.url,allow_none=True)
+            raise Exception,'chroot-based myplc testing is deprecated'
+       self.apiserver=TestApiserver(self.url,options.dry_run)
         
     def name(self):
         name=self.plc_spec['name']
-        if self.vserver:
-            return name+"[%s]"%self.vservername
-        else:
-            return name+"[chroot]"
+        return "%s.%s"%(name,self.vservername)
+
+    def hostname(self):
+        return self.plc_spec['hostname']
 
     def is_local (self):
-        return self.plc_spec['hostname'] == 'localhost'
+        return self.test_ssh.is_local()
 
     # define the API methods on this object through xmlrpc
     # would help, but not strictly necessary
     def connect (self):
        pass
-    
-    # command gets run in the chroot/vserver
-    def host_to_guest(self,command):
-        if self.vserver:
-            return "vserver %s exec %s"%(self.vservername,command)
-        else:
-            return "chroot /plc/root %s"%utils.backslash_shell_specials(command)
-
-    # command gets run on the right box
-    def to_host(self,command):
-        if self.is_local():
-            return command
-        else:
-            return "ssh %s %s"%(self.plc_spec['hostname'],utils.backslash_shell_specials(command))
-
-    def full_command(self,command):
-        return self.to_host(self.host_to_guest(command))
 
+    def actual_command_in_guest (self,command):
+        return self.test_ssh.actual_command(self.host_to_guest(command))
+    
     def run_in_guest (self,command):
-        return utils.system(self.full_command(command))
+        return utils.system(self.actual_command_in_guest(command))
+    
     def run_in_host (self,command):
-        return utils.system(self.to_host(command))
+        return self.test_ssh.run_in_buildname(command)
 
+    #command gets run in the vserver
+    def host_to_guest(self,command):
+        return "vserver %s exec %s"%(self.vservername,command)
+    
     # xxx quick n dirty
     def run_in_guest_piped (self,local,remote):
-        return utils.system(local+" | "+self.full_command(remote))
-
-    # copy a file to the myplc root image - pass in_data=True if the file must go in /plc/data
-    def copy_in_guest (self, localfile, remotefile, in_data=False):
-        if in_data:
-            chroot_dest="/plc/data"
-        else:
-            chroot_dest="/plc/root"
-        if self.is_local():
-            if not self.vserver:
-                utils.system("cp %s %s/%s"%(localfile,chroot_dest,remotefile))
-            else:
-                utils.system("cp %s /vservers/%s/%s"%(localfile,self.vservername,remotefile))
-        else:
-            if not self.vserver:
-                utils.system("scp %s %s:%s/%s"%(localfile,self.plc_spec['hostname'],chroot_dest,remotefile))
-            else:
-                utils.system("scp %s %s@/vservers/%s/%s"%(localfile,self.plc_spec['hostname'],self.vservername,remotefile))
+        return utils.system(local+" | "+self.test_ssh.actual_command(self.host_to_guest(remote),keep_stdin=True))
 
     def auth_root (self):
        return {'Username':self.plc_spec['PLC_ROOT_USER'],
@@ -120,16 +148,29 @@ class TestPlc:
     def locate_node (self,nodename):
         for site in self.plc_spec['sites']:
             for node in site['nodes']:
-                if node['node_fields']['hostname'] == nodename:
+                if node['name'] == nodename:
                     return (site,node)
         raise Exception,"Cannot locate node %s"%nodename
         
+    def locate_hostname (self,hostname):
+        for site in self.plc_spec['sites']:
+            for node in site['nodes']:
+                if node['node_fields']['hostname'] == hostname:
+                    return (site,node)
+        raise Exception,"Cannot locate hostname %s"%hostname
+        
     def locate_key (self,keyname):
         for key in self.plc_spec['keys']:
             if key['name'] == keyname:
                 return key
         raise Exception,"Cannot locate key %s"%keyname
 
+    def locate_slice (self, slicename):
+        for slice in self.plc_spec['slices']:
+            if slice['slice_fields']['name'] == slicename:
+                return slice
+        raise Exception,"Cannot locate slice %s"%slicename
+
     # all different hostboxes used in this plc
     def gather_hostBoxes(self):
         # maps on sites and nodes, return [ (host_box,test_node) ]
@@ -140,7 +181,7 @@ class TestPlc:
                 test_node = TestNode (self, test_site, node_spec)
                 if not test_node.is_real():
                     tuples.append( (test_node.host_box(),test_node) )
-        # transform into a dict { 'host_box' -> [ hostnames .. ] }
+        # transform into a dict { 'host_box' -> [ test_node .. ] }
         result = {}
         for (box,node) in tuples:
             if not result.has_key(box):
@@ -150,131 +191,133 @@ class TestPlc:
         return result
                     
     # a step for checking this stuff
-    def showboxes (self,options):
-        print 'showboxes'
+    def show_boxes (self):
         for (box,nodes) in self.gather_hostBoxes().iteritems():
             print box,":"," + ".join( [ node.name() for node in nodes ] )
         return True
 
     # make this a valid step
-    def kill_all_qemus(self,options):
+    def kill_all_qemus(self):
+        # this is the brute force version, kill all qemus on that host box
         for (box,nodes) in self.gather_hostBoxes().iteritems():
-            # this is the brute force version, kill all qemus on that host box
-            TestBox(box).kill_all_qemus()
+            # pass the first nodename, as we don't push template-qemu on testboxes
+            nodedir=nodes[0].nodedir()
+            TestBox(box,self.options.buildname).kill_all_qemus(nodedir)
         return True
 
     # make this a valid step
-    def list_all_qemus(self,options):
+    def list_all_qemus(self):
         for (box,nodes) in self.gather_hostBoxes().iteritems():
-           # push the script
-           TestBox(box).copy("qemu_kill.sh")   
             # this is the brute force version, kill all qemus on that host box
-            TestBox(box).run("./qemu_kill.sh -l")
+            TestBox(box,self.options.buildname).list_all_qemus()
         return True
 
     # kill only the right qemus
-    def kill_qemus(self,options):
+    def list_qemus(self):
         for (box,nodes) in self.gather_hostBoxes().iteritems():
-           # push the script
-           TestBox(box).copy("qemu_kill.sh")   
             # the fine-grain version
             for node in nodes:
-                node.kill_qemu()
+                node.list_qemu()
         return True
 
-    def clear_ssh_config (self,options):
-        # install local ssh_config file as root's .ssh/config - ssh should be quiet
-        # dir might need creation first
-        self.run_in_guest("mkdir /root/.ssh")
-        self.run_in_guest("chmod 700 /root/.ssh")
-        # this does not work - > redirection somehow makes it until an argument to cat
-        #self.run_in_guest_piped("cat ssh_config","cat > /root/.ssh/config")
-        self.copy_in_guest("ssh_config","/root/.ssh/config",True)
-        return True
-            
-    #################### step methods
-
-    ### uninstall
-    def uninstall_chroot(self,options):
-        self.run_in_host('service plc safestop')
-        #####detecting the last myplc version installed and remove it
-        self.run_in_host('rpm -e myplc')
-        ##### Clean up the /plc directory
-        self.run_in_host('rm -rf  /plc/data')
-        ##### stop any running vservers
-        self.run_in_host('for vserver in $(ls /vservers/* | sed -e s,/vservers/,,) ; do vserver $vserver stop ; done')
+    # kill only the right qemus
+    def kill_qemus(self):
+        for (box,nodes) in self.gather_hostBoxes().iteritems():
+            # the fine-grain version
+            for node in nodes:
+                node.kill_qemu()
         return True
 
-    def uninstall_vserver(self,options):
-        self.run_in_host("vserver --silent %s delete"%self.vservername)
+
+    ### utility methods for handling the pool of IP addresses allocated to plcs
+    # Logic
+    # (*) running plcs are recorded in the file named ~/running-test-plcs
+    # (*) this file contains a line for each running plc, older first
+    # (*) each line contains the vserver name + the hostname of the (vserver) testbox where it sits
+    # (*) the free_tracker method performs a vserver stop on the oldest entry
+    # (*) the record_tracker method adds an entry at the bottom of the file
+    # (*) the cleanup_tracker method stops all known vservers and removes the tracker file
+
+    TRACKER_FILE="~/running-test-plcs"
+
+    def record_tracker (self):
+        command="echo %s %s >> %s"%(self.vservername,self.test_ssh.hostname,TestPlc.TRACKER_FILE)
+        (code,output) = utils.output_of (self.test_ssh.actual_command(command))
+        if code != 0:
+            print "WARNING : COULD NOT record_tracker %s as a running plc on %s"%(self.vservername,self.test_ssh.hostname)
+            return False
+        print "Recorded %s in running plcs on host %s"%(self.vservername,self.test_ssh.hostname)
         return True
 
-    def uninstall(self,options):
-        # if there's a chroot-based myplc running, and then a native-based myplc is being deployed
-        # it sounds safer to have the former uninstalled too
-        # now the vserver method cannot be invoked for chroot instances as vservername is required
-        if self.vserver:
-            self.uninstall_vserver(options)
-            self.uninstall_chroot(options)
-        else:
-            self.uninstall_chroot(options)
+    def free_tracker (self):
+        command="head -1 %s"%TestPlc.TRACKER_FILE
+        (code,line) = utils.output_of(self.test_ssh.actual_command(command))
+        if code != 0:
+            print "No entry found in %s on %s"%(TestPlc.TRACKER_FILE,self.test_ssh.hostname)
+            return False
+        try:
+            [vserver_to_stop,hostname] = line.split()
+        except:
+            print "WARNING: free_tracker: Could not parse %s - skipped"%TestPlc.TRACKER_FILE
+            return False
+        stop_command = "vserver --silent %s stop"%vserver_to_stop
+        utils.system(self.test_ssh.actual_command(stop_command))
+        x=TestPlc.TRACKER_FILE
+        flush_command = "tail --lines=+2 %s > %s.tmp ; mv %s.tmp %s"%(x,x,x,x)
+        utils.system(self.test_ssh.actual_command(flush_command))
         return True
 
-    ### install
-    def install_chroot(self,options):
-        # nothing to do
+    # this should/could stop only the ones in TRACKER_FILE if that turns out to be reliable
+    def cleanup_tracker (self):
+        stop_all = "cd /vservers ; for i in * ; do vserver --silent $i stop ; done"
+        utils.system(self.test_ssh.actual_command(stop_all))
+        clean_tracker = "rm -f %s"%TestPlc.TRACKER_FILE
+        utils.system(self.test_ssh.actual_command(clean_tracker))
+
+    def uninstall(self):
+        self.run_in_host("vserver --silent %s delete"%self.vservername)
         return True
 
-    # xxx this would not work with hostname != localhost as mylc-init-vserver was extracted locally
-    def install_vserver(self,options):
+    ### install
+    def install(self):
         # we need build dir for vtest-init-vserver
         if self.is_local():
             # a full path for the local calls
-            build_dir=self.path+"/build"
+            build_dir=os.path.dirname(sys.argv[0])+"/build"
         else:
-            # use a standard name - will be relative to HOME 
-            build_dir="tests-system-build"
-        build_checkout = "svn checkout %s %s"%(options.build_url,build_dir)
+            # use a standard name - will be relative to remote buildname
+            build_dir="build"
+       # run checkout in any case - would do an update if already exists
+        build_checkout = "svn checkout %s %s"%(self.options.build_url,build_dir)
         if self.run_in_host(build_checkout) != 0:
-            raise Exception,"Cannot checkout build dir"
-        # the repo url is taken from myplc-url 
-        # with the last two steps (i386/myplc...) removed
-        repo_url = options.myplc_url
-        repo_url = os.path.dirname(repo_url)
-        repo_url = os.path.dirname(repo_url)
-        create_vserver="%s/vtest-init-vserver.sh %s %s -- --interface eth0:%s"%\
-            (build_dir,self.vservername,repo_url,self.vserverip)
-        if self.run_in_host(create_vserver) != 0:
-            raise Exception,"Could not create vserver for %s"%self.vservername
-        return True
-
-    def install(self,options):
-        if self.vserver:
-            return self.install_vserver(options)
+            return False
+        # the repo url is taken from arch-rpms-url 
+        # with the last step (i386.) removed
+        repo_url = self.options.arch_rpms_url
+        for level in [ 'arch' ]:
+           repo_url = os.path.dirname(repo_url)
+        if self.options.arch == "i386":
+            personality_option="-p linux32"
         else:
-            return self.install_chroot(options)
-
-    ### install_rpm
-    def install_rpm_chroot(self,options):
-        utils.header('Installing from %s'%options.myplc_url)
-        url=options.myplc_url
-        self.run_in_host('rpm -Uvh '+url)
-        self.run_in_host('service plc mount')
-        return True
-
-    def install_rpm_vserver(self,options):
-        self.run_in_guest("yum -y install myplc-native")
-        return True
+            personality_option="-p linux64"
+        script="vtest-init-vserver.sh"
+        vserver_name = self.vservername
+        vserver_options="--netdev eth0 --interface %s"%self.vserverip
+        try:
+            vserver_hostname=socket.gethostbyaddr(self.vserverip)[0]
+            vserver_options += " --hostname %s"%vserver_hostname
+        except:
+            pass
+        create_vserver="%(build_dir)s/%(script)s %(personality_option)s %(vserver_name)s %(repo_url)s -- %(vserver_options)s"%locals()
+        return self.run_in_host(create_vserver) == 0
 
-    def install_rpm(self,options):
-        if self.vserver:
-            return self.install_rpm_vserver(options)
-        else:
-            return self.install_rpm_chroot(options)
+    ### install_rpm 
+    def install_rpm(self):
+        return self.run_in_guest("yum -y install myplc-native")==0
 
     ### 
-    def configure(self,options):
-        tmpname='%s/%s.plc-config-tty'%(options.path,self.name())
+    def configure(self):
+        tmpname='%s.plc-config-tty'%(self.name())
         fileconf=open(tmpname,'w')
         for var in [ 'PLC_NAME',
                      'PLC_ROOT_PASSWORD',
@@ -296,37 +339,30 @@ class TestPlc:
         utils.system('rm %s'%tmpname)
         return True
 
-    # the chroot install is slightly different to this respect
-    def start(self, options):
-        if self.vserver:
-            self.run_in_guest('service plc start')
-        else:
-            self.run_in_host('service plc start')
+    def start(self):
+        self.run_in_guest('service plc start')
         return True
         
-    def stop(self, options):
-        if self.vserver:
-            self.run_in_guest('service plc stop')
-        else:
-            self.run_in_host('service plc stop')
+    def stop(self):
+        self.run_in_guest('service plc stop')
         return True
         
     # could use a TestKey class
-    def store_keys(self, options):
+    def store_keys(self):
         for key_spec in self.plc_spec['keys']:
-            TestKey(self,key_spec).store_key()
+               TestKey(self,key_spec).store_key()
         return True
 
-    def clean_keys(self, options):
-        utils.system("rm -rf %s/keys/"%self.path)
+    def clean_keys(self):
+        utils.system("rm -rf %s/keys/"%os.path(sys.argv[0]))
 
-    def sites (self,options):
-        return self.do_sites(options)
+    def sites (self):
+        return self.do_sites()
     
-    def clean_sites (self,options):
-        return self.do_sites(options,action="delete")
+    def clean_sites (self):
+        return self.do_sites(action="delete")
     
-    def do_sites (self,options,action="add"):
+    def do_sites (self,action="add"):
         for site_spec in self.plc_spec['sites']:
             test_site = TestSite (self,site_spec)
             if (action != "add"):
@@ -341,12 +377,12 @@ class TestPlc:
                 test_site.create_users()
         return True
 
-    def nodes (self, options):
-        return self.do_nodes(options)
-    def clean_nodes (self, options):
-        return self.do_nodes(options,action="delete")
+    def nodes (self):
+        return self.do_nodes()
+    def clean_nodes (self):
+        return self.do_nodes(action="delete")
 
-    def do_nodes (self, options,action="add"):
+    def do_nodes (self,action="add"):
         for site_spec in self.plc_spec['sites']:
             test_site = TestSite (self,site_spec)
             if action != "add":
@@ -358,14 +394,14 @@ class TestPlc:
             else:
                 utils.header("Creating nodes for site %s in %s"%(test_site.name(),self.name()))
                 for node_spec in site_spec['nodes']:
-                    utils.show_spec('Creating node %s'%node_spec,node_spec)
+                    utils.pprint('Creating node %s'%node_spec,node_spec)
                     test_node = TestNode (self,test_site,node_spec)
                     test_node.create_node ()
         return True
 
     # create nodegroups if needed, and populate
     # no need for a clean_nodegroups if we are careful enough
-    def nodegroups (self, options):
+    def nodegroups (self):
         # 1st pass to scan contents
         groups_dict = {}
         for site_spec in self.plc_spec['sites']:
@@ -382,12 +418,40 @@ class TestPlc:
                         groups_dict[nodegroupname].append(test_node.name())
         auth=self.auth_root()
         for (nodegroupname,group_nodes) in groups_dict.iteritems():
-            try:
-                self.server.GetNodeGroups(auth,{'name':nodegroupname})[0]
-            except:
-                self.server.AddNodeGroup(auth,{'name':nodegroupname})
-            for node in group_nodes:
-                self.server.AddNodeToNodeGroup(auth,node,nodegroupname)
+            print 'nodegroups:','dealing with nodegroup',nodegroupname,'on nodes',group_nodes
+            # first, check if the nodetagtype is here
+            tag_types = self.apiserver.GetNodeTagTypes(auth,{'name':nodegroupname})
+            if tag_types:
+                tag_type_id = tag_types[0]['node_tag_type_id']
+                print 'node-tag-type',nodegroupname,'already exists'
+            else:
+                tag_type_id = self.apiserver.AddNodeTagType(auth,
+                                                            {'name':nodegroupname,
+                                                             'description':
+                                                                 'for nodegroup %s'%nodegroupname,
+                                                             'category':'test',
+                                                             'min_role_id':10})
+            # create nodegroup
+            nodegroups = self.apiserver.GetNodeGroups (auth, {'groupname':nodegroupname})
+            if nodegroups:
+                print 'nodegroup',nodegroupname,'already exists'
+            else:
+                self.apiserver.AddNodeGroup(auth,
+                                            {'groupname': nodegroupname,
+                                             'node_tag_type_id': tag_type_id,
+                                             'value': 'yes'})
+            # set node tag on all nodes, value='yes'
+            for nodename in group_nodes:
+                # check if already set
+                # xxx need node_id - need improvement in the API
+                node_id = self.apiserver.GetNodes(auth,nodename)[0]['node_id']
+                node_tag = self.apiserver.GetNodeTags(auth,
+                                                      {'node_id':node_id,
+                                                       'node_tag_type_id':tag_type_id})
+                if node_tag:
+                    print 'node',nodename,'already has tag',nodegroupname
+                else:
+                    self.apiserver.AddNodeTag(auth, node_id, nodegroupname,"yes")
         return True
 
     def all_hostnames (self) :
@@ -398,7 +462,10 @@ class TestPlc:
         return hostnames
 
     # gracetime : during the first <gracetime> minutes nothing gets printed
-    def do_nodes_booted (self, minutes, gracetime=2):
+    def do_nodes_booted (self, minutes, gracetime,period=30):
+        if self.options.dry_run:
+            print 'dry_run'
+            return True
         # compute timeout
         timeout = datetime.datetime.now()+datetime.timedelta(minutes=minutes)
         graceout = datetime.datetime.now()+datetime.timedelta(minutes=gracetime)
@@ -409,7 +476,7 @@ class TestPlc:
         status = dict ( [ (hostname,'undef') for hostname in tocheck ] )
         while tocheck:
             # get their status
-            tocheck_status=self.server.GetNodes(self.auth_root(), tocheck, ['hostname','boot_state' ] )
+            tocheck_status=self.apiserver.GetNodes(self.auth_root(), tocheck, ['hostname','boot_state' ] )
             # update status
             for array in tocheck_status:
                 hostname=array['hostname']
@@ -418,12 +485,12 @@ class TestPlc:
                     utils.header ("%s has reached the 'boot' state"%hostname)
                 else:
                     # if it's a real node, never mind
-                    (site_spec,node_spec)=self.locate_node(hostname)
+                    (site_spec,node_spec)=self.locate_hostname(hostname)
                     if TestNode.is_real_model(node_spec['node_fields']['model']):
                         utils.header("WARNING - Real node %s in %s - ignored"%(hostname,boot_state))
                         # let's cheat
                         boot_state = 'boot'
-                    if datetime.datetime.now() > graceout:
+                    elif datetime.datetime.now() > graceout:
                         utils.header ("%s still in '%s' state"%(hostname,boot_state))
                         graceout=datetime.datetime.now()+datetime.timedelta(1)
                 status[hostname] = boot_state
@@ -436,94 +503,103 @@ class TestPlc:
                     utils.header("FAILURE due to %s in '%s' state"%(hostname,status[hostname]))
                 return False
             # otherwise, sleep for a while
-            time.sleep(15)
+            time.sleep(period)
         # only useful in empty plcs
         return True
 
-    def nodes_booted(self,options):
-        return self.do_nodes_booted(minutes=5)
-    
-    #to scan and store the nodes's public keys and avoid to ask for confirmation when  ssh 
-    def scan_publicKeys(self,hostnames):
-        try:
-            temp_knownhosts="/root/known_hosts"
-            remote_knownhosts="/root/.ssh/known_hosts"
-            self.run_in_host("touch %s"%temp_knownhosts )
-            for hostname in hostnames:
-                utils.header("Scan Public %s key and store it in the known_host file(under the root image) "%hostname)
-                scan=self.run_in_host('ssh-keyscan -t rsa %s >> %s '%(hostname,temp_knownhosts))
-            #Store the public keys in the right root image
-            self.copy_in_guest(temp_knownhosts,remote_knownhosts,True)
-            #clean the temp keys file used
-            self.run_in_host('rm -f  %s '%temp_knownhosts )
-        except Exception, err:
-            print err
-            
-    def do_check_nodesSsh(self,minutes):
+    def nodes_booted(self):
+        return self.do_nodes_booted(minutes=20,gracetime=15)
+
+    def do_nodes_ssh(self,minutes,gracetime,period=30):
         # compute timeout
         timeout = datetime.datetime.now()+datetime.timedelta(minutes=minutes)
-        #graceout = datetime.datetime.now()+datetime.timedelta(minutes=gracetime)
+        graceout = datetime.datetime.now()+datetime.timedelta(minutes=gracetime)
         tocheck = self.all_hostnames()
-        self.scan_publicKeys(tocheck)
+#        self.scan_publicKeys(tocheck)
         utils.header("checking Connectivity on nodes %r"%tocheck)
         while tocheck:
             for hostname in tocheck:
                 # try to ssh in nodes
-                access=self.run_in_guest('ssh -i /etc/planetlab/root_ssh_key.rsa root@%s date'%hostname )
-                if (not access):
+                node_test_ssh = TestSsh (hostname,key="/etc/planetlab/root_ssh_key.rsa")
+                success=self.run_in_guest(node_test_ssh.actual_command("hostname"))==0
+                if success:
                     utils.header('The node %s is sshable -->'%hostname)
                     # refresh tocheck
                     tocheck.remove(hostname)
                 else:
-                    (site_spec,node_spec)=self.locate_node(hostname)
+                    # we will have tried real nodes once, in case they're up - but if not, just skip
+                    (site_spec,node_spec)=self.locate_hostname(hostname)
                     if TestNode.is_real_model(node_spec['node_fields']['model']):
                         utils.header ("WARNING : check ssh access into real node %s - skipped"%hostname)
-                    tocheck.remove(hostname)
-            if not tocheck:
+                       tocheck.remove(hostname)
+                    elif datetime.datetime.now() > graceout:
+                        utils.header("Could not ssh-enter root context on %s"%hostname)
+            if  not tocheck:
                 return True
             if datetime.datetime.now() > timeout:
                 for hostname in tocheck:
                     utils.header("FAILURE to ssh into %s"%hostname)
                 return False
             # otherwise, sleep for a while
-            time.sleep(15)
+            time.sleep(period)
         # only useful in empty plcs
         return True
         
-    def nodes_ssh(self, options):
-        return  self.do_check_nodesSsh(minutes=2)
+    def nodes_ssh(self):
+        return self.do_nodes_ssh(minutes=6,gracetime=4)
     
-    def bootcd (self, options):
-        for site_spec in self.plc_spec['sites']:
-            test_site = TestSite (self,site_spec)
-            for node_spec in site_spec['nodes']:
-                test_node=TestNode (self,test_site,node_spec)
-                test_node.create_boot_cd(options.path)
-        return True
-
-    def do_check_intiscripts(self):
-           for site_spec in self.plc_spec['sites']:
-                   test_site = TestSite (self,site_spec)
-                   for slice_spec in self.plc_spec['slices']:
-                           test_slice=TestSlice (self,test_site,slice_spec)
-                           init_status=test_slice.get_initscript()
-                           if (not init_status):
-                                   return False
-                   return init_status
+    @node_mapper
+    def init_node (self): pass
+    @node_mapper
+    def bootcd (self): pass
+    @node_mapper
+    def configure_qemu (self): pass
+    @node_mapper
+    def reinstall_node (self): pass
+    @node_mapper
+    def export_qemu (self): pass
+        
+    def do_check_initscripts(self):
+        overall = True
+        for slice_spec in self.plc_spec['slices']:
+            if not slice_spec.has_key('initscriptname'):
+                continue
+            initscript=slice_spec['initscriptname']
+            for nodename in slice_spec['nodenames']:
+                (site,node) = self.locate_node (nodename)
+                # xxx - passing the wrong site - probably harmless
+                test_site = TestSite (self,site)
+                test_slice = TestSlice (self,test_site,slice_spec)
+                test_node = TestNode (self,test_site,node)
+                test_sliver = TestSliver (self, test_node, test_slice)
+                if not test_sliver.check_initscript(initscript):
+                    overall = False
+        return overall
            
-    def check_initscripts(self, options):
-           return self.do_check_intiscripts()
+    def check_initscripts(self):
+           return self.do_check_initscripts()
                    
-    def initscripts (self, options):
+    def initscripts (self):
         for initscript in self.plc_spec['initscripts']:
-            utils.show_spec('Adding Initscript in plc %s'%self.plc_spec['name'],initscript)
-            self.server.AddInitScript(self.auth_root(),initscript['initscript_fields'])
+            utils.pprint('Adding Initscript in plc %s'%self.plc_spec['name'],initscript)
+            self.apiserver.AddInitScript(self.auth_root(),initscript['initscript_fields'])
+        return True
+
+    def clean_initscripts (self):
+        for initscript in self.plc_spec['initscripts']:
+            initscript_name = initscript['initscript_fields']['name']
+            print('Attempting to delete %s in plc %s'%(initscript_name,self.plc_spec['name']))
+            try:
+                self.apiserver.DeleteInitScript(self.auth_root(),initscript_name)
+                print initscript_name,'deleted'
+            except:
+                print 'deletion went wrong - probably did not exist'
         return True
 
-    def slices (self, options):
+    def slices (self):
         return self.do_slices()
 
-    def clean_slices (self, options):
+    def clean_slices (self):
         return self.do_slices("delete")
 
     def do_slices (self,  action="add"):
@@ -535,40 +611,110 @@ class TestPlc:
                 utils.header("Deleting slices in site %s"%test_site.name())
                 test_slice.delete_slice()
             else:    
-                utils.show_spec("Creating slice",slice)
+                utils.pprint("Creating slice",slice)
                 test_slice.create_slice()
                 utils.header('Created Slice %s'%slice['slice_fields']['name'])
         return True
         
-    def check_slices(self, options):
+    @slice_mapper_options
+    def check_slice(self): pass
+
+    @node_mapper
+    def clear_known_hosts (self): pass
+    
+    @node_mapper
+    def start_node (self) : pass
+
+    def all_sliver_objs (self):
+        result=[]
         for slice_spec in self.plc_spec['slices']:
-            site_spec = self.locate_site (slice_spec['sitename'])
-            test_site = TestSite(self,site_spec)
-            test_slice=TestSlice(self,test_site,slice_spec)
-            status=test_slice.do_check_slice(options)
-            if (not status):
-                return False
-        return status
+            slicename = slice_spec['slice_fields']['name']
+            for nodename in slice_spec['nodenames']:
+                result.append(self.locate_sliver_obj (nodename,slicename))
+        return result
+
+    def locate_sliver_obj (self,nodename,slicename):
+        (site,node) = self.locate_node(nodename)
+        slice = self.locate_slice (slicename)
+        # build objects
+        test_site = TestSite (self, site)
+        test_node = TestNode (self, test_site,node)
+        # xxx the slice site is assumed to be the node site - mhh - probably harmless
+        test_slice = TestSlice (self, test_site, slice)
+        return TestSliver (self, test_node, test_slice)
+
+    def check_tcp (self):
+        specs = self.plc_spec['tcp_test']
+        overall=True
+        for spec in specs:
+            port = spec['port']
+            # server side
+            s_test_sliver = self.locate_sliver_obj (spec['server_node'],spec['server_slice'])
+            if not s_test_sliver.run_tcp_server(port,timeout=10):
+                overall=False
+                break
+
+            # idem for the client side
+            c_test_sliver = self.locate_sliver_obj(spec['server_node'],spec['server_slice'])
+            if not c_test_sliver.run_tcp_client(s_test_sliver.test_node.name(),port):
+                overall=False
+        return overall
     
-    def start_nodes (self, options):
-        utils.header("Starting  nodes")
+
+    def gather_logs (self):
+        # (1) get the plc's /var/log and store it locally in logs/myplc.var-log.<plcname>/*
+        # (2) get all the nodes qemu log and store it as logs/node.qemu.<node>.log
+        # (3) get the nodes /var/log and store is as logs/node.var-log.<node>/*
+        # (4) as far as possible get the slice's /var/log as logs/sliver.var-log.<sliver>/*
+        # (1)
+        print "-------------------- TestPlc.gather_logs : PLC's /var/log"
+        self.gather_var_logs ()
+        # (2) 
+        print "-------------------- TestPlc.gather_logs : nodes's QEMU logs"
         for site_spec in self.plc_spec['sites']:
-            TestSite(self,site_spec).start_nodes (options)
+            test_site = TestSite (self,site_spec)
+            for node_spec in site_spec['nodes']:
+                test_node=TestNode(self,test_site,node_spec)
+                test_node.gather_qemu_logs()
+        # (3)
+        print "-------------------- TestPlc.gather_logs : nodes's /var/log"
+        self.gather_nodes_var_logs()
+        # (4)
+        print "-------------------- TestPlc.gather_logs : sample sliver's /var/log"
+        self.gather_slivers_var_logs()
         return True
 
-    def stop_nodes (self, options):
-        self.kill_all_qemus()
+    def gather_slivers_var_logs(self):
+        for test_sliver in self.all_sliver_objs():
+            remote = test_sliver.tar_var_logs()
+            utils.system("mkdir -p logs/sliver.var-log.%s"%test_sliver.name())
+            command = remote + " | tar -C logs/sliver.var-log.%s -xf -"%test_sliver.name()
+            utils.system(command)
         return True
 
-    def check_tcp (self, options):
-        print 'check_tcp not yet implemented'
-        return True
+    def gather_var_logs (self):
+        to_plc = self.actual_command_in_guest("tar -C /var/log/ -cf - .")        
+        command = to_plc + "| tar -C logs/myplc.var-log.%s -xf -"%self.name()
+        utils.system("mkdir -p logs/myplc.var-log.%s"%self.name())
+        utils.system(command)
+
+    def gather_nodes_var_logs (self):
+        for site_spec in self.plc_spec['sites']:
+            test_site = TestSite (self,site_spec)
+            for node_spec in site_spec['nodes']:
+                test_node=TestNode(self,test_site,node_spec)
+                test_ssh = TestSsh (test_node.name(),key="/etc/planetlab/root_ssh_key.rsa")
+                to_plc = self.actual_command_in_guest ( test_ssh.actual_command("tar -C /var/log -cf - ."))
+                command = to_plc + "| tar -C logs/node.var-log.%s -xf -"%test_node.name()
+                utils.system("mkdir -p logs/node.var-log.%s"%test_node.name())
+                utils.system(command)
+
 
     # returns the filename to use for sql dump/restore, using options.dbname if set
-    def dbfile (self, database, options):
+    def dbfile (self, database):
         # uses options.dbname if it is found
         try:
-            name=options.dbname
+            name=self.options.dbname
             if not isinstance(name,StringTypes):
                 raise Exception
         except:
@@ -577,15 +723,14 @@ class TestPlc:
             name=str(d)
         return "/root/%s-%s.sql"%(database,name)
 
-    def db_dump(self, options):
-        
-        dump=self.dbfile("planetab4",options)
+    def db_dump(self):
+        dump=self.dbfile("planetab4")
         self.run_in_guest('pg_dump -U pgsqluser planetlab4 -f '+ dump)
         utils.header('Dumped planetlab4 database in %s'%dump)
         return True
 
-    def db_restore(self, options):
-        dump=self.dbfile("planetab4",options)
+    def db_restore(self):
+        dump=self.dbfile("planetab4")
         ##stop httpd service
         self.run_in_guest('service httpd stop')
         # xxx - need another wrapper