new sfa_renew_slice step
[tests.git] / system / TestPlc.py
index cfd5a80..06da593 100644 (file)
@@ -1,4 +1,6 @@
-# $Id$
+# Thierry Parmentelat <thierry.parmentelat@inria.fr>
+# Copyright (C) 2010 INRIA 
+#
 import os, os.path
 import datetime
 import time
@@ -14,10 +16,10 @@ from TestUser import TestUser
 from TestKey import TestKey
 from TestSlice import TestSlice
 from TestSliver import TestSliver
-from TestBox import TestBox
+from TestBoxQemu import TestBoxQemu
 from TestSsh import TestSsh
 from TestApiserver import TestApiserver
-from Trackers import TrackerPlc, TrackerQemu
+from TestAuthSfa import TestAuthSfa
 
 # step methods must take (self) and return a boolean (options is a member of the class)
 
@@ -36,18 +38,17 @@ def standby_generic (func):
     return actual
 
 def node_mapper (method):
-    def actual(self):
+    def actual(self,*args, **kwds):
         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
+        for test_node in self.all_nodes():
+            if not node_method(test_node, *args, **kwds): overall=False
         return overall
+    # restore the doc text
+    actual.__doc__=TestNode.__dict__[method.__name__].__doc__
     return actual
 
-def slice_mapper_options (method):
+def slice_mapper (method):
     def actual(self):
         overall=True
         slice_method = TestSlice.__dict__[method.__name__]
@@ -57,62 +58,110 @@ def slice_mapper_options (method):
             test_slice=TestSlice(self,test_site,slice_spec)
             if not slice_method(test_slice,self.options): overall=False
         return overall
+    # restore the doc text
+    actual.__doc__=TestSlice.__dict__[method.__name__].__doc__
+    return actual
+
+def auth_sfa_mapper (method):
+    def actual(self):
+        overall=True
+        auth_method = TestAuthSfa.__dict__[method.__name__]
+        for auth_spec in self.plc_spec['sfa']['auth_sfa_specs']:
+            test_auth=TestAuthSfa(self,auth_spec)
+            if not auth_method(test_auth,self.options): overall=False
+        return overall
+    # restore the doc text
+    actual.__doc__=TestAuthSfa.__dict__[method.__name__].__doc__
     return actual
 
 SEP='<sep>'
+SEPSFA='<sep_sfa>'
 
 class TestPlc:
 
     default_steps = [
-        'display','trqemu_record','trqemu_free','uninstall','install','install_rpm', 
-        'configure', 'start', 'fetch_keys', 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,
-        # better use of time: do this now that the nodes are taking off
-        'plcsh_stress_test', SEP,
-        'nodes_ssh_debug', 'nodes_ssh_boot', 'check_slice', 'check_initscripts', SEP,
-        'check_tcp',  SEP,
-        'force_gather_logs', 'force_trplc_record','force_trplc_free',
+        'show', SEP,
+        'vs_delete','timestamp_vs','vs_create', SEP,
+        'plc_install', 'plc_configure', 'plc_start', SEP,
+        'keys_fetch', 'keys_store', 'keys_clear_known_hosts', 'speed_up_slices', SEP,
+        'initscripts', 'sites', 'nodes', 'slices', 'nodegroups', 'leases', SEP,
+        'nodestate_reinstall', 'qemu_local_init','bootcd', 'qemu_local_config', SEP,
+        'qemu_export', 'qemu_kill_mine', 'qemu_start', 'timestamp_qemu', SEP,
+        'sfa_install_all', 'sfa_configure', 'cross_sfa_configure', 'sfa_start', 'sfa_import', SEPSFA,
+        'sfi_configure@1', 'sfa_add_site@1','sfa_add_pi@1', SEPSFA,
+        'sfa_add_user@1', 'sfa_update_user@1', 'sfa_add_slice@1', 'sfa_renew_slice@1', SEPSFA,
+        'sfa_discover@1', 'sfa_create_slice@1', 'sfa_check_slice_plc@1', 'sfa_update_slice@1', SEPSFA,
+        'sfi_list@1', 'sfi_show@1', 'sfi_slices@1', 'sfa_utest@1', SEPSFA,
+        # we used to run plcsh_stress_test, and then ssh_node_debug and ssh_node_boot
+        # but as the stress test might take a while, we sometimes missed the debug mode..
+        'ssh_node_debug@1', 'plcsh_stress_test@1', SEP,
+        'ssh_node_boot@1', 'ssh_slice', 'check_initscripts', SEP,
+        'ssh_slice_sfa@1', 'sfa_delete_slice@1', 'sfa_delete_user@1', SEPSFA,
+        'cross_check_tcp@1', 'check_system_slice', SEP,
+        'empty_slices', 'ssh_slice_off', 'fill_slices', SEP,
+        'force_gather_logs', SEP,
         ]
     other_steps = [ 
-        'stop_all_vservers','fresh_install', 'cache_rpm', 'stop', 'vs_start', SEP,
-        'check_sanity',  SEP,
-        'clean_initscripts', 'clean_nodegroups','clean_all_sites', SEP,
-        'clean_sites', 'clean_nodes', 
-        'clean_slices', 'clean_keys', SEP,
-        'show_boxes', 'list_all_qemus', 'list_qemus', 'kill_qemus', SEP,
-        'db_dump' , 'db_restore', 'trplc_cleanup','trqemu_cleanup','trackers_cleanup', SEP,
-        'standby_1 through 20',
+        'export', 'show_boxes', SEP,
+        'check_hooks', 'plc_stop', 'vs_start', 'vs_stop', SEP,
+        'delete_initscripts', 'delete_nodegroups','delete_all_sites', SEP,
+        'delete_sites', 'delete_nodes', 'delete_slices', 'keys_clean', SEP,
+        'delete_leases', 'list_leases', SEP,
+        'populate', SEP,
+        'nodestate_show','nodestate_safeboot','nodestate_boot', SEP,
+        'qemu_list_all', 'qemu_list_mine', 'qemu_kill_all', SEP,
+       'sfa_install_core', 'sfa_install_sfatables', 'sfa_install_plc', 'sfa_install_client', SEPSFA,
+        'sfa_plcclean', 'sfa_dbclean', 'sfa_stop','sfa_uninstall', 'sfi_clean', SEPSFA,
+        'plc_db_dump' , 'plc_db_restore', SEP,
+        'check_netflow','check_drl', SEP,
+        'debug_nodemanager', SEP,
+        'standby_1_through_20',SEP,
         ]
 
     @staticmethod
     def printable_steps (list):
-        return " ".join(list).replace(" "+SEP+" "," \\\n")
+        single_line=" ".join(list)+" "
+        return single_line.replace(" "+SEP+" "," \\\n").replace(" "+SEPSFA+" "," \\\n")
     @staticmethod
     def valid_step (step):
-        return step != SEP
+        return step != SEP and step != SEPSFA
+
+    # turn off the sfa-related steps when build has skipped SFA
+    # this was originally for centos5 but is still valid
+    # for up to f12 as recent SFAs with sqlalchemy won't build before f14
+    @staticmethod
+    def check_whether_build_has_sfa (rpms_url):
+        utils.header ("Checking if build provides SFA package...")
+        # warning, we're now building 'sface' so let's be a bit more picky
+        retcod=os.system ("curl --silent %s/ | grep -q sfa-"%rpms_url)
+        # full builds are expected to return with 0 here
+        if retcod==0:
+            utils.header("build does provide SFA")
+        else:
+            # move all steps containing 'sfa' from default_steps to other_steps
+            utils.header("SFA package not found - removing steps with sfa or sfi")
+            sfa_steps= [ step for step in TestPlc.default_steps if step.find('sfa')>=0 or step.find("sfi")>=0 ]
+            TestPlc.other_steps += sfa_steps
+            for step in sfa_steps: TestPlc.default_steps.remove(step)
 
     def __init__ (self,plc_spec,options):
        self.plc_spec=plc_spec
         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:
-            raise Exception,'chroot-based myplc testing is deprecated'
+       self.test_ssh=TestSsh(self.plc_spec['host_box'],self.options.buildname)
+        self.vserverip=plc_spec['vserverip']
+        self.vservername=plc_spec['vservername']
+        self.url="https://%s:443/PLCAPI/"%plc_spec['vserverip']
        self.apiserver=TestApiserver(self.url,options.dry_run)
         
+    def has_addresses_api (self):
+        return self.apiserver.has_method('AddIpAddress')
+
     def name(self):
         name=self.plc_spec['name']
         return "%s.%s"%(name,self.vservername)
 
     def hostname(self):
-        return self.plc_spec['hostname']
+        return self.plc_spec['host_box']
 
     def is_local (self):
         return self.test_ssh.is_local()
@@ -128,24 +177,65 @@ class TestPlc:
     def start_guest (self):
       return utils.system(self.test_ssh.actual_command(self.start_guest_in_host()))
     
+    def stop_guest (self):
+      return utils.system(self.test_ssh.actual_command(self.stop_guest_in_host()))
+    
     def run_in_guest (self,command):
         return utils.system(self.actual_command_in_guest(command))
     
     def run_in_host (self,command):
         return self.test_ssh.run_in_buildname(command)
 
-    #command gets run in the vserver
+    #command gets run in the plc's vm
     def host_to_guest(self,command):
-        return "vserver %s exec %s"%(self.vservername,command)
+        if self.options.plcs_use_lxc:
+            return "ssh -o StrictHostKeyChecking=no %s %s"%(self.vserverip,command)
+        else:
+            return "vserver %s exec %s"%(self.vservername,command)
     
-    #command gets run in the vserver
+    def vm_root_in_host(self):
+        if self.options.plcs_use_lxc:
+            return "/var/lib/lxc/%s/rootfs/"%(self.vservername)
+        else:
+            return "/vservers/%s"%(self.vservername)
+
+    def vm_timestamp_path (self):
+        if self.options.plcs_use_lxc:
+            return "/var/lib/lxc/%s/%s.timestamp"%(self.vservername,self.vservername)
+        else:
+            return "/vservers/%s.timestamp"%(self.vservername)
+
+    #start/stop the vserver
     def start_guest_in_host(self):
-        return "vserver %s start"%(self.vservername)
+        if self.options.plcs_use_lxc:
+            return "lxc-start --daemon --name=%s"%(self.vservername)
+        else:
+            return "vserver %s start"%(self.vservername)
+    
+    def stop_guest_in_host(self):
+        if self.options.plcs_use_lxc:
+            return "lxc-stop --name=%s"%(self.vservername)
+        else:
+            return "vserver %s stop"%(self.vservername)
     
     # xxx quick n dirty
     def run_in_guest_piped (self,local,remote):
         return utils.system(local+" | "+self.test_ssh.actual_command(self.host_to_guest(remote),keep_stdin=True))
 
+    def yum_check_installed (self, rpms):
+        if isinstance (rpms, list): 
+            rpms=" ".join(rpms)
+        return self.run_in_guest("rpm -q %s"%rpms)==0
+        
+    # does a yum install in the vs, ignore yum retcod, check with rpm
+    def yum_install (self, rpms):
+        if isinstance (rpms, list): 
+            rpms=" ".join(rpms)
+        self.run_in_guest("yum -y install %s"%rpms)
+        # yum-complete-transaction comes with yum-utils, that is in vtest.pkgs
+        self.run_in_guest("yum-complete-transaction -y")
+        return self.yum_check_installed (rpms)
+
     def auth_root (self):
        return {'Username':self.plc_spec['PLC_ROOT_USER'],
                'AuthMethod':'password',
@@ -174,11 +264,24 @@ class TestPlc:
                     return (site,node)
         raise Exception,"Cannot locate hostname %s"%hostname
         
-    def locate_key (self,keyname):
+    def locate_key (self,key_name):
         for key in self.plc_spec['keys']:
-            if key['name'] == keyname:
+            if key['key_name'] == key_name:
                 return key
-        raise Exception,"Cannot locate key %s"%keyname
+        raise Exception,"Cannot locate key %s"%key_name
+
+    def locate_private_key_from_key_names (self, key_names):
+        # locate the first avail. key
+        found=False
+        for key_name in key_names:
+            key_spec=self.locate_key(key_name)
+            test_key=TestKey(self,key_spec)
+            publickey=test_key.publicpath()
+            privatekey=test_key.privatepath()
+            if os.path.isfile(publickey) and os.path.isfile(privatekey):
+                found=True
+        if found: return privatekey
+        else:     return None
 
     def locate_slice (self, slicename):
         for slice in self.plc_spec['slices']:
@@ -238,28 +341,32 @@ class TestPlc:
                     
     # a step for checking this stuff
     def show_boxes (self):
+        'print summary of nodes location'
         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):
+    def qemu_kill_all(self):
+        'kill all qemu instances on the qemu boxes involved by this setup'
         # this is the brute force version, kill all qemus on that host box
         for (box,nodes) in self.gather_hostBoxes().iteritems():
             # 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)
+            TestBoxQemu(box,self.options.buildname).qemu_kill_all(nodedir)
         return True
 
     # make this a valid step
-    def list_all_qemus(self):
+    def qemu_list_all(self):
+        'list all qemu instances on the qemu boxes involved by this setup'
         for (box,nodes) in self.gather_hostBoxes().iteritems():
             # this is the brute force version, kill all qemus on that host box
-            TestBox(box,self.options.buildname).list_all_qemus()
+            TestBoxQemu(box,self.options.buildname).qemu_list_all()
         return True
 
     # kill only the right qemus
-    def list_qemus(self):
+    def qemu_list_mine(self):
+        'list qemu instances for our nodes'
         for (box,nodes) in self.gather_hostBoxes().iteritems():
             # the fine-grain version
             for node in nodes:
@@ -267,7 +374,8 @@ class TestPlc:
         return True
 
     # kill only the right qemus
-    def kill_qemus(self):
+    def qemu_kill_mine(self):
+        'kill the qemu instances for our nodes'
         for (box,nodes) in self.gather_hostBoxes().iteritems():
             # the fine-grain version
             for node in nodes:
@@ -275,14 +383,36 @@ class TestPlc:
         return True
 
     #################### display config
-    def display (self):
-        self.display_pass (1)
-        self.display_pass (2)
+    def show (self):
+        "show test configuration after localization"
+        self.show_pass (1)
+        self.show_pass (2)
+        return True
+
+    def export (self):
+        "print cut'n paste-able stuff to export env variables to your shell"
+        # guess local domain from hostname
+        domain=socket.gethostname().split('.',1)[1]
+        fqdn="%s.%s"%(self.plc_spec['host_box'],domain)
+        print "export BUILD=%s"%self.options.buildname
+        if self.options.plcs_use_lxc:
+            print "export PLCHOSTLXC=%s"%fqdn
+        else:
+            print "export PLCHOSTVS=%s"%fqdn
+        print "export GUESTNAME=%s"%self.plc_spec['vservername']
+        vplcname=self.plc_spec['vservername'].split('-')[-1]
+        print "export GUESTHOSTNAME=%s.%s"%(vplcname,domain)
+        # find hostname of first node
+        (hostname,qemubox) = self.all_node_infos()[0]
+        print "export KVMHOST=%s.%s"%(qemubox,domain)
+        print "export NODE=%s"%(hostname)
         return True
 
     # entry point
-    def display_pass (self,passno):
+    always_display_keys=['PLC_WWW_HOST','nodes','sites',]
+    def show_pass (self,passno):
         for (key,val) in self.plc_spec.iteritems():
+            if not self.options.verbose and key not in TestPlc.always_display_keys: continue
             if passno == 2:
                 if key == 'sites':
                     for site in val:
@@ -299,65 +429,67 @@ class TestPlc:
                     for key in val:
                         self.display_key_spec (key)
             elif passno == 1:
-                if key not in ['sites','initscripts','slices','keys']:
-                    print '*   ',key,':',val
+                if key not in ['sites','initscripts','slices','keys', 'sfa']:
+                    print '+   ',key,':',val
 
     def display_site_spec (self,site):
-        print '* ======== site',site['site_fields']['name']
+        print '+ ======== site',site['site_fields']['name']
         for (k,v) in site.iteritems():
+            if not self.options.verbose and k not in TestPlc.always_display_keys: continue
             if k=='nodes':
                 if v: 
-                    print '*       ','nodes : ',
+                    print '+       ','nodes : ',
                     for node in v:  
                         print node['node_fields']['hostname'],'',
                     print ''
             elif k=='users':
                 if v: 
-                    print '*       users : ',
+                    print '+       users : ',
                     for user in v:  
                         print user['name'],'',
                     print ''
             elif k == 'site_fields':
-                print '*       login_base',':',v['login_base']
+                print '+       login_base',':',v['login_base']
             elif k == 'address_fields':
                 pass
             else:
-                print '*       ',k,
-                PrettyPrinter(indent=8,depth=2).pprint(v)
+                print '+       ',
+                utils.pprint(k,v)
         
     def display_initscript_spec (self,initscript):
-        print '* ======== initscript',initscript['initscript_fields']['name']
+        print '+ ======== initscript',initscript['initscript_fields']['name']
 
     def display_key_spec (self,key):
-        print '* ======== key',key['name']
+        print '+ ======== key',key['key_name']
 
     def display_slice_spec (self,slice):
-        print '* ======== slice',slice['slice_fields']['name']
+        print '+ ======== slice',slice['slice_fields']['name']
         for (k,v) in slice.iteritems():
             if k=='nodenames':
                 if v: 
-                    print '*       nodes : ',
+                    print '+       nodes : ',
                     for nodename in v:  
                         print nodename,'',
                     print ''
             elif k=='usernames':
                 if v: 
-                    print '*       users : ',
+                    print '+       users : ',
                     for username in v:  
                         print username,'',
                     print ''
             elif k=='slice_fields':
-                print '*       fields',':',
+                print '+       fields',':',
                 print 'max_nodes=',v['max_nodes'],
                 print ''
             else:
-                print '*       ',k,v
+                print '+       ',k,v
 
     def display_node_spec (self,node):
-        print "*           node",node['name'],"host_box=",node['host_box'],
+        print "+           node=%s host_box=%s"%(node['name'],node['host_box']),
         print "hostname=",node['node_fields']['hostname'],
         print "ip=",node['interface_fields']['ip']
-    
+        if self.options.verbose:
+            utils.pprint("node details",node,depth=3)
 
     # another entry point for just showing the boxes involved
     def display_mapping (self):
@@ -366,73 +498,53 @@ class TestPlc:
 
     @staticmethod
     def display_mapping_plc (plc_spec):
-        print '* MyPLC',plc_spec['name']
-        print '*\tvserver address = root@%s:/vservers/%s'%(plc_spec['hostname'],plc_spec['vservername'])
-        print '*\tIP = %s/%s'%(plc_spec['PLC_API_HOST'],plc_spec['vserverip'])
+        print '+ MyPLC',plc_spec['name']
+        # WARNING this would not be right for lxc-based PLC's - should be harmless though
+        print '+\tvserver address = root@%s:/vservers/%s'%(plc_spec['host_box'],plc_spec['vservername'])
+        print '+\tIP = %s/%s'%(plc_spec['PLC_API_HOST'],plc_spec['vserverip'])
         for site_spec in plc_spec['sites']:
             for node_spec in site_spec['nodes']:
                 TestPlc.display_mapping_node(node_spec)
 
     @staticmethod
     def display_mapping_node (node_spec):
-        print '*   NODE %s'%(node_spec['name'])
-        print '*\tqemu box %s'%node_spec['host_box']
-        print '*\thostname=%s'%node_spec['node_fields']['hostname']
-
-    ### tracking
-    def trplc_record (self):
-        tracker = TrackerPlc(self.options)
-        tracker.record(self.test_ssh.hostname,self.vservername)
-        tracker.store()
-        return True
-
-    def trplc_free (self):
-        tracker = TrackerPlc(self.options)
-        tracker.free()
-        tracker.store()
-        return True
-
-    def trplc_cleanup (self):
-        tracker = TrackerPlc(self.options)
-        tracker.cleanup()
-        tracker.store()
-        return True
-
-    def trqemu_record (self):
-        tracker=TrackerQemu(self.options)
-        for site_spec in self.plc_spec['sites']:
-            for node_spec in site_spec['nodes']:
-                tracker.record(node_spec['host_box'],self.options.buildname,node_spec['node_fields']['hostname'])
-        tracker.store()
-        return True
-
-    def trqemu_free (self):
-        tracker=TrackerQemu(self.options)
-        for site_spec in self.plc_spec['sites']:
-            for node_spec in site_spec['nodes']:
-                tracker.free()
-        tracker.store()
-        return True
-
-    def trqemu_cleanup (self):
-        tracker=TrackerQemu(self.options)
-        for site_spec in self.plc_spec['sites']:
-            for node_spec in site_spec['nodes']:
-                tracker.cleanup()
-        tracker.store()
-        return True
-
-    def trackers_cleanup (self):
-        self.trqemu_cleanup()
-        self.trplc_cleanup()
-        return True
-
-    def uninstall(self):
-        self.run_in_host("vserver --silent %s delete"%self.vservername)
-        return True
+        print '+   NODE %s'%(node_spec['name'])
+        print '+\tqemu box %s'%node_spec['host_box']
+        print '+\thostname=%s'%node_spec['node_fields']['hostname']
+
+    # write a timestamp in /vservers/<>.timestamp
+    # cannot be inside the vserver, that causes vserver .. build to cough
+    def timestamp_vs (self):
+        "Create a timestamp to remember creation date for this plc"
+        now=int(time.time())
+        # TODO-lxc check this one
+        # a first approx. is to store the timestamp close to the VM root like vs does
+        stamp_path=self.vm_timestamp_path ()
+        stamp_dir = os.path.dirname (stamp_path)
+        utils.system(self.test_ssh.actual_command("mkdir -p %s"%stamp_dir))
+        return utils.system(self.test_ssh.actual_command("echo %d > %s"%(now,stamp_path)))==0
+        
+    # this is called inconditionnally at the beginning of the test sequence 
+    # just in case this is a rerun, so if the vm is not running it's fine
+    def vs_delete(self):
+        "vserver delete the test myplc"
+        stamp_path=self.vm_timestamp_path()
+        self.run_in_host("rm -f %s"%stamp_path)
+        if self.options.plcs_use_lxc:
+            self.run_in_host("lxc-stop --name %s"%self.vservername)
+            self.run_in_host("lxc-destroy --name %s"%self.vservername)
+            return True
+        else:
+            self.run_in_host("vserver --silent %s delete"%self.vservername)
+            return True
 
     ### install
-    def install(self):
+    # historically the build was being fetched by the tests
+    # now the build pushes itself as a subdir of the tests workdir
+    # so that the tests do not have to worry about extracting the build (svn, git, or whatever)
+    def vs_create (self):
+        "vserver creation (no install done)"
+        # push the local build/ dir to the testplc box 
         if self.is_local():
             # a full path for the local calls
             build_dir=os.path.dirname(sys.argv[0])
@@ -442,10 +554,9 @@ class TestPlc:
         else:
             # 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:
-            return False
+            # remove for safety; do *not* mkdir first, otherwise we end up with build/build/
+            self.test_ssh.rmdir(build_dir)
+            self.test_ssh.copy(build_dir,recursive=True)
         # the repo url is taken from arch-rpms-url 
         # with the last step (i386) removed
         repo_url = self.options.arch_rpms_url
@@ -456,45 +567,70 @@ class TestPlc:
         test_env_options += " -p %s"%self.options.personality
         test_env_options += " -d %s"%self.options.pldistro
         test_env_options += " -f %s"%self.options.fcdistro
-        script="vtest-init-vserver.sh"
+        if self.options.plcs_use_lxc:
+            script="vtest-init-lxc.sh"
+        else:
+            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
+            print "Cannot reverse lookup %s"%self.vserverip
+            print "This is considered fatal, as this might pollute the test results"
+            return False
         create_vserver="%(build_dir)s/%(script)s %(test_env_options)s %(vserver_name)s %(repo_url)s -- %(vserver_options)s"%locals()
         return self.run_in_host(create_vserver) == 0
 
     ### install_rpm 
-    def install_rpm(self):
+    def plc_install(self):
+        "yum install myplc, noderepo, and the plain bootstrapfs"
+
+        # workaround for getting pgsql8.2 on centos5
+        if self.options.fcdistro == "centos5":
+            self.run_in_guest("rpm -Uvh http://download.fedora.redhat.com/pub/epel/5/i386/epel-release-5-3.noarch.rpm")
+
+        # compute nodefamily
         if self.options.personality == "linux32":
             arch = "i386"
         elif self.options.personality == "linux64":
             arch = "x86_64"
         else:
             raise Exception, "Unsupported personality %r"%self.options.personality
-        return \
-            self.run_in_guest("yum -y install myplc-native")==0 and \
-            self.run_in_guest("yum -y install noderepo-%s-%s"%(self.options.pldistro,arch))==0 and \
-            self.run_in_guest("yum -y install bootstrapfs-%s-%s-plain"%(self.options.pldistro,arch))==0 
+        nodefamily="%s-%s-%s"%(self.options.pldistro,self.options.fcdistro,arch)
+
+        pkgs_list=[]
+        pkgs_list.append ("slicerepo-%s"%nodefamily)
+        pkgs_list.append ("myplc")
+        pkgs_list.append ("noderepo-%s"%nodefamily)
+        pkgs_list.append ("nodeimage-%s-plain"%nodefamily)
+        pkgs_string=" ".join(pkgs_list)
+        return self.yum_install (pkgs_list)
 
     ### 
-    def configure(self):
+    def plc_configure(self):
+        "run plc-config-tty"
         tmpname='%s.plc-config-tty'%(self.name())
         fileconf=open(tmpname,'w')
         for var in [ 'PLC_NAME',
-                     'PLC_ROOT_PASSWORD',
                      'PLC_ROOT_USER',
+                     'PLC_ROOT_PASSWORD',
+                     'PLC_SLICE_PREFIX',
                      'PLC_MAIL_ENABLED',
                      'PLC_MAIL_SUPPORT_ADDRESS',
                      'PLC_DB_HOST',
+#                     'PLC_DB_PASSWORD',
+                    # Above line was added for integrating SFA Testing
                      'PLC_API_HOST',
                      'PLC_WWW_HOST',
                      'PLC_BOOT_HOST',
                      'PLC_NET_DNS1',
-                     'PLC_NET_DNS2']:
+                     'PLC_NET_DNS2',
+                     'PLC_RESERVATION_GRANULARITY',
+                     'PLC_OMF_ENABLED',
+                     'PLC_OMF_XMPP_SERVER',
+                     ]:
             fileconf.write ('e %s\n%s\n'%(var,self.plc_spec[var]))
         fileconf.write('w\n')
         fileconf.write('q\n')
@@ -504,51 +640,61 @@ class TestPlc:
         utils.system('rm %s'%tmpname)
         return True
 
-    def start(self):
+    def plc_start(self):
+        "service plc start"
         self.run_in_guest('service plc start')
         return True
 
-    def stop(self):
+    def plc_stop(self):
+        "service plc stop"
         self.run_in_guest('service plc stop')
         return True
         
     def vs_start (self):
+        "start the PLC vserver"
         self.start_guest()
         return True
 
+    def vs_stop (self):
+        "stop the PLC vserver"
+        self.stop_guest()
+        return True
+
     # stores the keys from the config for further use
-    def store_keys(self):
+    def keys_store(self):
+        "stores test users ssh keys in keys/"
         for key_spec in self.plc_spec['keys']:
                TestKey(self,key_spec).store_key()
         return True
 
-    def clean_keys(self):
-        utils.system("rm -rf %s/keys/"%os.path(sys.argv[0]))
+    def keys_clean(self):
+        "removes keys cached in keys/"
+        utils.system("rm -rf ./keys")
+        return True
 
     # fetches the ssh keys in the plc's /etc/planetlab and stores them in keys/
     # for later direct access to the nodes
-    def fetch_keys(self):
+    def keys_fetch(self):
+        "gets ssh keys in /etc/planetlab/ and stores them locally in keys/"
         dir="./keys"
         if not os.path.isdir(dir):
             os.mkdir(dir)
         vservername=self.vservername
+        vm_root=self.vm_root_in_host()
         overall=True
-        prefix = 'root_ssh_key'
-        for ext in [ 'pub', 'rsa' ] :
-            src="/vservers/%(vservername)s/etc/planetlab/%(prefix)s.%(ext)s"%locals()
-            dst="keys/%(vservername)s.%(ext)s"%locals()
-            if self.test_ssh.fetch(src,dst) != 0: overall=False
         prefix = 'debug_ssh_key'
         for ext in [ 'pub', 'rsa' ] :
-            src="/vservers/%(vservername)s/etc/planetlab/%(prefix)s.%(ext)s"%locals()
+            src="%(vm_root)s/etc/planetlab/%(prefix)s.%(ext)s"%locals()
             dst="keys/%(vservername)s-debug.%(ext)s"%locals()
             if self.test_ssh.fetch(src,dst) != 0: overall=False
         return overall
 
     def sites (self):
+        "create sites with PLCAPI"
         return self.do_sites()
     
-    def clean_sites (self):
+    def delete_sites (self):
+        "delete sites with PLCAPI"
         return self.do_sites(action="delete")
     
     def do_sites (self,action="add"):
@@ -566,16 +712,23 @@ class TestPlc:
                 test_site.create_users()
         return True
 
-    def clean_all_sites (self):
+    def delete_all_sites (self):
+        "Delete all sites in PLC, and related objects"
         print 'auth_root',self.auth_root()
-        site_ids = [s['site_id'] for s in self.apiserver.GetSites(self.auth_root(), {}, ['site_id'])]
-        for site_id in site_ids:
+        sites = self.apiserver.GetSites(self.auth_root(), {}, ['site_id'])
+        for site in sites:
+            # keep automatic site - otherwise we shoot in our own foot, root_auth is not valid anymore
+            if site['login_base']==self.plc_spec['PLC_SLICE_PREFIX']: continue
+            site_id=site['site_id']
             print 'Deleting site_id',site_id
             self.apiserver.DeleteSite(self.auth_root(),site_id)
+        return True
 
     def nodes (self):
+        "create nodes with PLCAPI"
         return self.do_nodes()
-    def clean_nodes (self):
+    def delete_nodes (self):
+        "delete nodes with PLCAPI"
         return self.do_nodes(action="delete")
 
     def do_nodes (self,action="add"):
@@ -596,10 +749,74 @@ class TestPlc:
         return True
 
     def nodegroups (self):
+        "create nodegroups with PLCAPI"
         return self.do_nodegroups("add")
-    def clean_nodegroups (self):
+    def delete_nodegroups (self):
+        "delete nodegroups with PLCAPI"
         return self.do_nodegroups("delete")
 
+    YEAR = 365*24*3600
+    @staticmethod
+    def translate_timestamp (start,grain,timestamp):
+        if timestamp < TestPlc.YEAR:    return start+timestamp*grain
+        else:                           return timestamp
+
+    @staticmethod
+    def timestamp_printable (timestamp):
+        return time.strftime('%m-%d %H:%M:%S UTC',time.gmtime(timestamp))
+
+    def leases(self):
+        "create leases (on reservable nodes only, use e.g. run -c default -c resa)"
+        now=int(time.time())
+        grain=self.apiserver.GetLeaseGranularity(self.auth_root())
+        print 'API answered grain=',grain
+        start=(now/grain)*grain
+        start += grain
+        # find out all nodes that are reservable
+        nodes=self.all_reservable_nodenames()
+        if not nodes: 
+            utils.header ("No reservable node found - proceeding without leases")
+            return True
+        ok=True
+        # attach them to the leases as specified in plc_specs
+        # this is where the 'leases' field gets interpreted as relative of absolute
+        for lease_spec in self.plc_spec['leases']:
+            # skip the ones that come with a null slice id
+            if not lease_spec['slice']: continue
+            lease_spec['t_from']=TestPlc.translate_timestamp(start,grain,lease_spec['t_from'])
+            lease_spec['t_until']=TestPlc.translate_timestamp(start,grain,lease_spec['t_until'])
+            lease_addition=self.apiserver.AddLeases(self.auth_root(),nodes,
+                                                    lease_spec['slice'],lease_spec['t_from'],lease_spec['t_until'])
+            if lease_addition['errors']:
+                utils.header("Cannot create leases, %s"%lease_addition['errors'])
+                ok=False
+            else:
+                utils.header('Leases on nodes %r for %s from %d (%s) until %d (%s)'%\
+                              (nodes,lease_spec['slice'],
+                               lease_spec['t_from'],TestPlc.timestamp_printable(lease_spec['t_from']),
+                               lease_spec['t_until'],TestPlc.timestamp_printable(lease_spec['t_until'])))
+                
+        return ok
+
+    def delete_leases (self):
+        "remove all leases in the myplc side"
+        lease_ids= [ l['lease_id'] for l in self.apiserver.GetLeases(self.auth_root())]
+        utils.header("Cleaning leases %r"%lease_ids)
+        self.apiserver.DeleteLeases(self.auth_root(),lease_ids)
+        return True
+
+    def list_leases (self):
+        "list all leases known to the myplc"
+        leases = self.apiserver.GetLeases(self.auth_root())
+        now=int(time.time())
+        for l in leases:
+            current=l['t_until']>=now
+            if self.options.verbose or current:
+                utils.header("%s %s from %s until %s"%(l['hostname'],l['name'],
+                                                       TestPlc.timestamp_printable(l['t_from']), 
+                                                       TestPlc.timestamp_printable(l['t_until'])))
+        return True
+
     # create nodegroups if needed, and populate
     def do_nodegroups (self, action="add"):
         # 1st pass to scan contents
@@ -629,8 +846,7 @@ class TestPlc:
                     tag_type_id = self.apiserver.AddTagType(auth,
                                                             {'tagname':nodegroupname,
                                                              'description': 'for nodegroup %s'%nodegroupname,
-                                                             'category':'test',
-                                                             'min_role_id':10})
+                                                             'category':'test'})
                 print 'located tag (type)',nodegroupname,'as',tag_type_id
                 # create nodegroup
                 nodegroups = self.apiserver.GetNodeGroups (auth, {'groupname':nodegroupname})
@@ -649,7 +865,7 @@ class TestPlc:
                         expect_yes = self.apiserver.GetNodeTags(auth,
                                                                 {'hostname':nodename,
                                                                  'tagname':nodegroupname},
-                                                                ['tagvalue'])[0]['tagvalue']
+                                                                ['value'])[0]['value']
                         if expect_yes != "yes":
                             print 'Mismatch node tag on node',nodename,'got',expect_yes
                             overall=False
@@ -666,12 +882,32 @@ class TestPlc:
                     overall=False
         return overall
 
-    def all_hostnames (self) :
-        hostnames = []
+    # a list of TestNode objs
+    def all_nodes (self):
+        nodes=[]
+        for site_spec in self.plc_spec['sites']:
+            test_site = TestSite (self,site_spec)
+            for node_spec in site_spec['nodes']:
+                nodes.append(TestNode (self,test_site,node_spec))
+        return nodes
+
+    # return a list of tuples (nodename,qemuname)
+    def all_node_infos (self) :
+        node_infos = []
         for site_spec in self.plc_spec['sites']:
-            hostnames += [ node_spec['node_fields']['hostname'] \
-                           for node_spec in site_spec['nodes'] ]
-        return hostnames
+            node_infos += [ (node_spec['node_fields']['hostname'],node_spec['host_box']) \
+                                for node_spec in site_spec['nodes'] ]
+        return node_infos
+    
+    def all_nodenames (self): return [ x[0] for x in self.all_node_infos() ]
+    def all_reservable_nodenames (self): 
+        res=[]
+        for site_spec in self.plc_spec['sites']:
+            for node_spec in site_spec['nodes']:
+                node_fields=node_spec['node_fields']
+                if 'node_type' in node_fields and node_fields['node_type']=='reservable':
+                    res.append(node_fields['hostname'])
+        return res
 
     # silent_minutes : during the first <silent_minutes> minutes nothing gets printed
     def nodes_check_boot_state (self, target_boot_state, timeout_minutes, silent_minutes,period=15):
@@ -720,9 +956,9 @@ class TestPlc:
         return True
 
     def nodes_booted(self):
-        return self.nodes_check_boot_state('boot',timeout_minutes=20,silent_minutes=15)
+        return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=28)
 
-    def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=20):
+    def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=15):
         # compute timeout
         timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes)
         graceout = datetime.datetime.now()+datetime.timedelta(minutes=silent_minutes)
@@ -732,82 +968,83 @@ class TestPlc:
             local_key = "keys/%(vservername)s-debug.rsa"%locals()
         else: 
             message="boot"
-            local_key = "keys/%(vservername)s.rsa"%locals()
-        tocheck = self.all_hostnames()
-        utils.header("checking ssh access (expected in %s mode) to nodes %r"%(message,tocheck))
+           local_key = "keys/key_admin.rsa"
+        node_infos = self.all_node_infos()
+        utils.header("checking ssh access (expected in %s mode) to nodes:"%message)
+        for (nodename,qemuname) in node_infos:
+            utils.header("hostname=%s -- qemubox=%s"%(nodename,qemuname))
         utils.header("max timeout is %d minutes, silent for %d minutes (period is %s)"%\
                          (timeout_minutes,silent_minutes,period))
-        while tocheck:
-            for hostname in tocheck:
+        while node_infos:
+            for node_info in node_infos:
+                (hostname,qemuname) = node_info
                 # try to run 'hostname' in the node
                 command = TestSsh (hostname,key=local_key).actual_command("hostname;uname -a")
                 # don't spam logs - show the command only after the grace period 
-                if datetime.datetime.now() > graceout:
-                    success=utils.system(command)
-                else:
-                    # truly silent, just print out a dot to show we're alive
-                    print '.',
-                    sys.stdout.flush()
-                    command += " 2>/dev/null"
-                    if self.options.dry_run:
-                        print 'dry_run',command
-                        success=0
-                    else:
-                        success=os.system(command)
+                success = utils.system ( command, silent=datetime.datetime.now() < graceout)
                 if success==0:
                     utils.header('Successfully entered root@%s (%s)'%(hostname,message))
-                    # refresh tocheck
-                    tocheck.remove(hostname)
+                    # refresh node_infos
+                    node_infos.remove(node_info)
                 else:
                     # 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:
+                       node_infos.remove(node_info)
+            if  not node_infos:
                 return True
             if datetime.datetime.now() > timeout:
-                for hostname in tocheck:
-                    utils.header("FAILURE to ssh into %s"%hostname)
+                for (hostname,qemuname) in node_infos:
+                    utils.header("FAILURE to ssh into %s (on %s)"%(hostname,qemuname))
                 return False
             # otherwise, sleep for a while
             time.sleep(period)
         # only useful in empty plcs
         return True
         
-    def nodes_ssh_debug(self):
-        return self.check_nodes_ssh(debug=True,timeout_minutes=30,silent_minutes=10)
+    def ssh_node_debug(self):
+        "Tries to ssh into nodes in debug mode with the debug ssh key"
+        return self.check_nodes_ssh(debug=True,timeout_minutes=10,silent_minutes=8)
     
-    def nodes_ssh_boot(self):
-        return self.check_nodes_ssh(debug=False,timeout_minutes=30,silent_minutes=10)
+    def ssh_node_boot(self):
+        "Tries to ssh into nodes in production mode with the root ssh key"
+        return self.check_nodes_ssh(debug=False,timeout_minutes=40,silent_minutes=38)
     
     @node_mapper
-    def init_node (self): pass
+    def qemu_local_init (self): pass
     @node_mapper
     def bootcd (self): pass
     @node_mapper
-    def configure_qemu (self): pass
+    def qemu_local_config (self): pass
+    @node_mapper
+    def nodestate_reinstall (self): pass
+    @node_mapper
+    def nodestate_safeboot (self): pass
     @node_mapper
-    def reinstall_node (self): pass
+    def nodestate_boot (self): pass
     @node_mapper
-    def export_qemu (self): pass
+    def nodestate_show (self): pass
+    @node_mapper
+    def qemu_export (self): pass
         
-    ### check sanity : invoke scripts from qaapi/qa/tests/{node,slice}
-    def check_sanity_node (self): 
-        return self.locate_first_node().check_sanity()
-    def check_sanity_sliver (self) : 
-        return self.locate_first_sliver().check_sanity()
+    ### check hooks : invoke scripts from hooks/{node,slice}
+    def check_hooks_node (self): 
+        return self.locate_first_node().check_hooks()
+    def check_hooks_sliver (self) : 
+        return self.locate_first_sliver().check_hooks()
     
-    def check_sanity (self):
-        return self.check_sanity_node() and self.check_sanity_sliver()
+    def check_hooks (self):
+        "runs unit tests in the node and slice contexts - see hooks/{node,slice}"
+        return self.check_hooks_node() and self.check_hooks_sliver()
 
     ### initscripts
     def do_check_initscripts(self):
         overall = True
         for slice_spec in self.plc_spec['slices']:
-            if not slice_spec.has_key('initscriptname'):
+            if not slice_spec.has_key('initscriptstamp'):
                 continue
-            initscript=slice_spec['initscriptname']
+            stamp=slice_spec['initscriptstamp']
             for nodename in slice_spec['nodenames']:
                 (site,node) = self.locate_node (nodename)
                 # xxx - passing the wrong site - probably harmless
@@ -815,20 +1052,23 @@ class TestPlc:
                 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):
+                if not test_sliver.check_initscript_stamp(stamp):
                     overall = False
         return overall
            
     def check_initscripts(self):
-           return self.do_check_initscripts()
-                   
+        "check that the initscripts have triggered"
+        return self.do_check_initscripts()
+    
     def initscripts (self):
+        "create initscripts with PLCAPI"
         for initscript in self.plc_spec['initscripts']:
             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):
+    def delete_initscripts (self):
+        "delete initscripts with PLCAPI"
         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']))
@@ -841,65 +1081,445 @@ class TestPlc:
 
     ### manage slices
     def slices (self):
-        return self.do_slices()
+        "create slices with PLCAPI"
+        return self.do_slices(action="add")
+
+    def delete_slices (self):
+        "delete slices with PLCAPI"
+        return self.do_slices(action="delete")
 
-    def clean_slices (self):
-        return self.do_slices("delete")
+    def fill_slices (self):
+        "add nodes in slices with PLCAPI"
+        return self.do_slices(action="fill")
+
+    def empty_slices (self):
+        "remove nodes from slices with PLCAPI"
+        return self.do_slices(action="empty")
 
     def do_slices (self,  action="add"):
         for slice in self.plc_spec['slices']:
             site_spec = self.locate_site (slice['sitename'])
             test_site = TestSite(self,site_spec)
             test_slice=TestSlice(self,test_site,slice)
-            if action != "add":
-                utils.header("Deleting slices in site %s"%test_site.name())
+            if action == "delete":
                 test_slice.delete_slice()
-            else:    
-                utils.pprint("Creating slice",slice)
+            elif action=="fill":
+                test_slice.add_nodes()
+            elif action=="empty":
+                test_slice.delete_nodes()
+            else:
                 test_slice.create_slice()
-                utils.header('Created Slice %s'%slice['slice_fields']['name'])
         return True
         
-    @slice_mapper_options
-    def check_slice(self): pass
+    @slice_mapper
+    def ssh_slice(self): pass
+    @slice_mapper
+    def ssh_slice_off (self): pass
 
     @node_mapper
-    def clear_known_hosts (self): pass
+    def keys_clear_known_hosts (self): pass
     
+    def speed_up_slices (self):
+        "tweak nodemanager settings on all nodes using a conf file"
+        # create the template on the server-side 
+        template="%s.nodemanager"%self.name()
+        template_file = open (template,"w")
+        template_file.write('OPTIONS="-p 30 -r 11 -d"\n')
+        template_file.close()
+        in_vm="/var/www/html/PlanetLabConf/nodemanager"
+        remote="%s/%s"%(self.vm_root_in_host(),in_vm)
+        self.test_ssh.copy_abs(template,remote)
+        # Add a conf file
+        self.apiserver.AddConfFile (self.auth_root(),
+                                    {'dest':'/etc/sysconfig/nodemanager',
+                                     'source':'PlanetLabConf/nodemanager',
+                                     'postinstall_cmd':'service nm restart',})
+        return True
+
+    def debug_nodemanager (self):
+        "sets verbose mode for nodemanager, and speeds up cycle even more (needs speed_up_slices first)"
+        template="%s.nodemanager"%self.name()
+        template_file = open (template,"w")
+        template_file.write('OPTIONS="-p 10 -r 6 -v -d"\n')
+        template_file.close()
+        in_vm="/var/www/html/PlanetLabConf/nodemanager"
+        remote="%s/%s"%(self.vm_root_in_host(),in_vm)
+        self.test_ssh.copy_abs(template,remote)
+        return True
+
     @node_mapper
-    def start_node (self) : pass
+    def qemu_start (self) : pass
 
-    def check_tcp (self):
-        specs = self.plc_spec['tcp_test']
+    @node_mapper
+    def timestamp_qemu (self) : pass
+
+    # when a spec refers to a node possibly on another plc
+    def locate_sliver_obj_cross (self, nodename, slicename, other_plcs):
+        for plc in [ self ] + other_plcs:
+            try:
+                return plc.locate_sliver_obj (nodename, slicename)
+            except:
+                pass
+        raise Exception, "Cannot locate sliver %s@%s among all PLCs"%(nodename,slicename)
+
+    # implement this one as a cross step so that we can take advantage of different nodes
+    # in multi-plcs mode
+    def cross_check_tcp (self, other_plcs):
+        "check TCP connectivity between 2 slices (or in loopback if only one is defined)"
+        if 'tcp_specs' not in self.plc_spec or not self.plc_spec['tcp_specs']: 
+            utils.header ("check_tcp: no/empty config found")
+            return True
+        specs = self.plc_spec['tcp_specs']
         overall=True
         for spec in specs:
             port = spec['port']
             # server side
-            s_test_sliver = self.locate_sliver_obj (spec['server_node'],spec['server_slice'])
+            s_test_sliver = self.locate_sliver_obj_cross (spec['server_node'],spec['server_slice'],other_plcs)
             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'])
+            c_test_sliver = self.locate_sliver_obj_cross (spec['server_node'],spec['client_slice'],other_plcs)
             if not c_test_sliver.run_tcp_client(s_test_sliver.test_node.name(),port):
                 overall=False
         return overall
 
+    # painfully enough, we need to allow for some time as netflow might show up last
+    def check_system_slice (self): 
+        "all nodes: check that a system slice is alive"
+        # netflow currently not working in the lxc distro
+        # drl not built at all in the wtx distro
+        # if we find either of them we're happy
+        return self.check_netflow() or self.check_drl()
+    
+    # expose these
+    def check_netflow (self): return self._check_system_slice ('netflow')
+    def check_drl (self): return self._check_system_slice ('drl')
+
+    # we have the slices up already here, so it should not take too long
+    def _check_system_slice (self, slicename, timeout_minutes=5, period=15):
+        timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes)
+        test_nodes=self.all_nodes()
+        while test_nodes:
+            for test_node in test_nodes:
+                if test_node._check_system_slice (slicename,dry_run=self.options.dry_run):
+                    utils.header ("ok")
+                    test_nodes.remove(test_node)
+                else:
+                    print '.',
+            if not test_nodes:
+                return True
+            if datetime.datetime.now () > timeout:
+                for test_node in test_nodes:
+                    utils.header ("can't find system slice %s in %s"%(slicename,test_node.name()))
+                return False
+            time.sleep(period)
+        return True
+
     def plcsh_stress_test (self):
+        "runs PLCAPI stress test, that checks Add/Update/Delete on all types - preserves contents"
         # install the stress-test in the plc image
-        location = "/usr/share/plc_api/plcsh-stress-test.py"
-        remote="/vservers/%s/%s"%(self.vservername,location)
-        self.test_ssh.copy_abs("plcsh-stress-test.py",remote)
+        location = "/usr/share/plc_api/plcsh_stress_test.py"
+        remote="%s/%s"%(self.vm_root_in_host(),location)
+        self.test_ssh.copy_abs("plcsh_stress_test.py",remote)
         command = location
         command += " -- --check"
         if self.options.size == 1:
             command +=  " --tiny"
         return ( self.run_in_guest(command) == 0)
 
+    # populate runs the same utility without slightly different options
+    # in particular runs with --preserve (dont cleanup) and without --check
+    # also it gets run twice, once with the --foreign option for creating fake foreign entries
+
+    def sfa_install_all (self):
+        "yum install sfa sfa-plc sfa-sfatables sfa-client"
+        return self.yum_install ("sfa sfa-plc sfa-sfatables sfa-client")
+
+    def sfa_install_core(self):
+        "yum install sfa"
+        return self.yum_install ("sfa")
+        
+    def sfa_install_plc(self):
+        "yum install sfa-plc"
+        return self.yum_install("sfa-plc")
+        
+    def sfa_install_sfatables(self):
+        "yum install sfa-sfatables"
+        return self.yum_install ("sfa-sfatables")
+
+    # for some very odd reason, this sometimes fails with the following symptom
+    # # yum install sfa-client
+    # Setting up Install Process
+    # ...
+    # Downloading Packages:
+    # Running rpm_check_debug
+    # Running Transaction Test
+    # Transaction Test Succeeded
+    # Running Transaction
+    # Transaction couldn't start:
+    # installing package sfa-client-2.1-7.onelab.2012.05.23.i686 needs 68KB on the / filesystem
+    # [('installing package sfa-client-2.1-7.onelab.2012.05.23.i686 needs 68KB on the / filesystem', (9, '/', 69632L))]
+    # even though in the same context I have
+    # [2012.05.23--f14-32-sfastd1-1-vplc07] / # df -h 
+    # Filesystem            Size  Used Avail Use% Mounted on
+    # /dev/hdv1             806G  264G  501G  35% /
+    # none                   16M   36K   16M   1% /tmp
+    #
+    # so as a workaround, we first try yum install, and then invoke rpm on the cached rpm...
+    def sfa_install_client(self):
+        "yum install sfa-client"
+        first_try=self.yum_install("sfa-client")
+        if first_try: return True
+        utils.header ("********** Regular yum failed - special workaround in place, 2nd chance")
+        (code,cached_rpm_path)=utils.output_of(self.actual_command_in_guest('find /var/cache/yum -name sfa-client\*.rpm'))
+        utils.header("rpm_path=<<%s>>"%rpm_path)
+        # just for checking 
+        self.run_in_guest("rpm -i %s"%cached_rpm_path)
+        return self.yum_check_installed ("sfa-client")
+
+    def sfa_dbclean(self):
+        "thoroughly wipes off the SFA database"
+        return self.run_in_guest("sfaadmin.py registry nuke")==0 or \
+            self.run_in_guest("sfa-nuke.py")==0 or \
+            self.run_in_guest("sfa-nuke-plc.py")==0
+
+    def sfa_plcclean(self):
+        "cleans the PLC entries that were created as a side effect of running the script"
+        # ignore result 
+        sfa_spec=self.plc_spec['sfa']
+
+        for auth_sfa_spec in sfa_spec['auth_sfa_specs']:
+            login_base=auth_sfa_spec['login_base']
+            try: self.apiserver.DeleteSite (self.auth_root(),login_base)
+            except: print "Site %s already absent from PLC db"%login_base
+
+            for spec_name in ['pi_spec','user_spec']:
+                user_spec=auth_sfa_spec[spec_name]
+                username=user_spec['email']
+                try: self.apiserver.DeletePerson(self.auth_root(),username)
+                except: 
+                    # this in fact is expected as sites delete their members
+                    #print "User %s already absent from PLC db"%username
+                    pass
+
+        print "REMEMBER TO RUN sfa_import AGAIN"
+        return True
+
+    def sfa_uninstall(self):
+        "uses rpm to uninstall sfa - ignore result"
+        self.run_in_guest("rpm -e sfa sfa-sfatables sfa-client sfa-plc")
+        self.run_in_guest("rm -rf /var/lib/sfa")
+        self.run_in_guest("rm -rf /etc/sfa")
+        self.run_in_guest("rm -rf /var/log/sfa_access.log /var/log/sfa_import_plc.log /var/log/sfa.daemon")
+        # xxx tmp 
+        self.run_in_guest("rpm -e --noscripts sfa-plc")
+        return True
+
+    ### run unit tests for SFA
+    # NOTE: for some reason on f14/i386, yum install sfa-tests fails for no reason
+    # Running Transaction
+    # Transaction couldn't start:
+    # installing package sfa-tests-1.0-21.onelab.i686 needs 204KB on the / filesystem
+    # [('installing package sfa-tests-1.0-21.onelab.i686 needs 204KB on the / filesystem', (9, '/', 208896L))]
+    # no matter how many Gbs are available on the testplc
+    # could not figure out what's wrong, so...
+    # if the yum install phase fails, consider the test is successful
+    # other combinations will eventually run it hopefully
+    def sfa_utest(self):
+        "yum install sfa-tests and run SFA unittests"
+        self.run_in_guest("yum -y install sfa-tests")
+        # failed to install - forget it
+        if self.run_in_guest("rpm -q sfa-tests")!=0: 
+            utils.header("WARNING: SFA unit tests failed to install, ignoring")
+            return True
+        return self.run_in_guest("/usr/share/sfa/tests/testAll.py")==0
+
+    ###
+    def confdir(self):
+        dirname="conf.%s"%self.plc_spec['name']
+        if not os.path.isdir(dirname):
+            utils.system("mkdir -p %s"%dirname)
+        if not os.path.isdir(dirname):
+            raise "Cannot create config dir for plc %s"%self.name()
+        return dirname
+
+    def conffile(self,filename):
+        return "%s/%s"%(self.confdir(),filename)
+    def confsubdir(self,dirname,clean,dry_run=False):
+        subdirname="%s/%s"%(self.confdir(),dirname)
+        if clean:
+            utils.system("rm -rf %s"%subdirname)
+        if not os.path.isdir(subdirname): 
+            utils.system("mkdir -p %s"%subdirname)
+        if not dry_run and not os.path.isdir(subdirname):
+            raise "Cannot create config subdir %s for plc %s"%(dirname,self.name())
+        return subdirname
+        
+    def conffile_clean (self,filename):
+        filename=self.conffile(filename)
+        return utils.system("rm -rf %s"%filename)==0
+    
+    ###
+    def sfa_configure(self):
+        "run sfa-config-tty"
+        tmpname=self.conffile("sfa-config-tty")
+        fileconf=open(tmpname,'w')
+        for var in [ 'SFA_REGISTRY_ROOT_AUTH',
+                     'SFA_INTERFACE_HRN',
+                     'SFA_REGISTRY_LEVEL1_AUTH',
+                    'SFA_REGISTRY_HOST',
+                    'SFA_AGGREGATE_HOST',
+                     'SFA_SM_HOST',
+                    'SFA_PLC_URL',
+                     'SFA_PLC_USER',
+                     'SFA_PLC_PASSWORD',
+                     'SFA_DB_HOST',
+                     'SFA_DB_USER',
+                     'SFA_DB_PASSWORD',
+                     'SFA_DB_NAME',
+                     'SFA_API_LOGLEVEL',
+                     'SFA_GENERIC_FLAVOUR',
+                     'SFA_AGGREGATE_ENABLED',
+                     ]:
+            if self.plc_spec['sfa'].has_key(var):
+                fileconf.write ('e %s\n%s\n'%(var,self.plc_spec['sfa'][var]))
+        # the way plc_config handles booleans just sucks..
+        for var in []:
+            val='false'
+            if self.plc_spec['sfa'][var]: val='true'
+            fileconf.write ('e %s\n%s\n'%(var,val))
+        fileconf.write('w\n')
+        fileconf.write('R\n')
+        fileconf.write('q\n')
+        fileconf.close()
+        utils.system('cat %s'%tmpname)
+        self.run_in_guest_piped('cat %s'%tmpname,'sfa-config-tty')
+        return True
+
+    def aggregate_xml_line(self):
+        port=self.plc_spec['sfa']['neighbours-port']
+        return '<aggregate addr="%s" hrn="%s" port="%r"/>' % \
+            (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'],port)
+
+    def registry_xml_line(self):
+        return '<registry addr="%s" hrn="%s" port="12345"/>' % \
+            (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'])
+
+
+    # a cross step that takes all other plcs in argument
+    def cross_sfa_configure(self, other_plcs):
+        "writes aggregates.xml and registries.xml that point to all other PLCs in the test"
+        # of course with a single plc, other_plcs is an empty list
+        if not other_plcs:
+            return True
+        agg_fname=self.conffile("agg.xml")
+        file(agg_fname,"w").write("<aggregates>%s</aggregates>\n" % \
+                                     " ".join([ plc.aggregate_xml_line() for plc in other_plcs ]))
+        utils.header ("(Over)wrote %s"%agg_fname)
+        reg_fname=self.conffile("reg.xml")
+        file(reg_fname,"w").write("<registries>%s</registries>\n" % \
+                                     " ".join([ plc.registry_xml_line() for plc in other_plcs ]))
+        utils.header ("(Over)wrote %s"%reg_fname)
+        return self.test_ssh.copy_abs(agg_fname,'/%s/etc/sfa/aggregates.xml'%self.vm_root_in_host())==0 \
+            and  self.test_ssh.copy_abs(reg_fname,'/%s/etc/sfa/registries.xml'%self.vm_root_in_host())==0
+
+    def sfa_import(self):
+        "sfa-import-plc"
+        auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH']
+        return \
+            self.run_in_guest('sfaadmin.py reg import_registry')==0 
+# not needed anymore
+#        self.run_in_guest('cp /etc/sfa/authorities/%s/%s.pkey /etc/sfa/authorities/server.key'%(auth,auth))
+
+    def sfa_start(self):
+        "service sfa start"
+        return self.run_in_guest('service sfa start')==0
+
+    def sfi_configure(self):
+        "Create /root/sfi on the plc side for sfi client configuration"
+        if self.options.dry_run: 
+            utils.header("DRY RUN - skipping step")
+            return True
+        sfa_spec=self.plc_spec['sfa']
+        # cannot use auth_sfa_mapper to pass dir_name
+        for slice_spec in self.plc_spec['sfa']['auth_sfa_specs']:
+            test_slice=TestAuthSfa(self,slice_spec)
+            dir_basename=os.path.basename(test_slice.sfi_path())
+            dir_name=self.confsubdir("dot-sfi/%s"%dir_basename,clean=True,dry_run=self.options.dry_run)
+            test_slice.sfi_configure(dir_name)
+            # push into the remote /root/sfi area
+            location = test_slice.sfi_path()
+            remote="%s/%s"%(self.vm_root_in_host(),location)
+            self.test_ssh.mkdir(remote,abs=True)
+            # need to strip last level or remote otherwise we get an extra dir level
+            self.test_ssh.copy_abs(dir_name, os.path.dirname(remote), recursive=True)
+
+        return True
+
+    def sfi_clean (self):
+        "clean up /root/sfi on the plc side"
+        self.run_in_guest("rm -rf /root/sfi")
+        return True
+
+    @auth_sfa_mapper
+    def sfa_add_site (self): pass
+    @auth_sfa_mapper
+    def sfa_add_pi (self): pass
+    @auth_sfa_mapper
+    def sfa_add_user(self): pass
+    @auth_sfa_mapper
+    def sfa_update_user(self): pass
+    @auth_sfa_mapper
+    def sfa_add_slice(self): pass
+    @auth_sfa_mapper
+    def sfa_renew_slice(self): pass
+    @auth_sfa_mapper
+    def sfa_discover(self): pass
+    @auth_sfa_mapper
+    def sfa_create_slice(self): pass
+    @auth_sfa_mapper
+    def sfa_check_slice_plc(self): pass
+    @auth_sfa_mapper
+    def sfa_update_slice(self): pass
+    @auth_sfa_mapper
+    def sfi_list(self): pass
+    @auth_sfa_mapper
+    def sfi_show(self): pass
+    @auth_sfa_mapper
+    def sfi_slices(self): pass
+    @auth_sfa_mapper
+    def ssh_slice_sfa(self): pass
+    @auth_sfa_mapper
+    def sfa_delete_user(self): pass
+    @auth_sfa_mapper
+    def sfa_delete_slice(self): pass
+
+    def sfa_stop(self):
+        "service sfa stop"
+        self.run_in_guest('service sfa stop')==0
+        return True
+
+    def populate (self):
+        "creates random entries in the PLCAPI"
+        # install the stress-test in the plc image
+        location = "/usr/share/plc_api/plcsh_stress_test.py"
+        remote="%s/%s"%(self.vm_root_in_host(),location)
+        self.test_ssh.copy_abs("plcsh_stress_test.py",remote)
+        command = location
+        command += " -- --preserve --short-names"
+        local = (self.run_in_guest(command) == 0);
+        # second run with --foreign
+        command += ' --foreign'
+        remote = (self.run_in_guest(command) == 0);
+        return ( local and remote)
+
     def gather_logs (self):
+        "gets all possible logs from plc's/qemu node's/slice's for future reference"
         # (1.a) get the plc's /var/log/ and store it locally in logs/myplc.var-log.<plcname>/*
         # (1.b) get the plc's  /var/lib/pgsql/data/pg_log/ -> logs/myplc.pgsql-log.<plcname>/*
+        # (1.c) get the plc's /root/sfi -> logs/sfi.<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>/*
@@ -909,6 +1529,9 @@ class TestPlc:
         # (1.b)
         print "-------------------- TestPlc.gather_logs : PLC's /var/lib/psql/data/pg_log/"
         self.gather_pgsql_logs ()
+        # (1.c)
+        print "-------------------- TestPlc.gather_logs : PLC's /root/sfi/"
+        self.gather_root_sfi ()
         # (2) 
         print "-------------------- TestPlc.gather_logs : nodes's QEMU logs"
         for site_spec in self.plc_spec['sites']:
@@ -946,14 +1569,20 @@ class TestPlc:
         command = to_plc + "| tar -C logs/myplc.pgsql-log.%s -xf -"%self.name()
         utils.system(command)
 
+    def gather_root_sfi (self):
+        utils.system("mkdir -p logs/sfi.%s"%self.name())
+        to_plc = self.actual_command_in_guest("tar -C /root/sfi/ -cf - .")        
+        command = to_plc + "| tar -C logs/sfi.%s -xf -"%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()
+                test_ssh = TestSsh (test_node.name(),key="keys/key_admin.rsa")
+                command = test_ssh.actual_command("tar -C /var/log -cf - .")
+                command = command + "| 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)
 
@@ -971,25 +1600,30 @@ class TestPlc:
             name=str(d)
         return "/root/%s-%s.sql"%(database,name)
 
-    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)
+    def plc_db_dump(self):
+        'dump the planetlab5 DB in /root in the PLC - filename has time'
+        dump=self.dbfile("planetab5")
+        self.run_in_guest('pg_dump -U pgsqluser planetlab5 -f '+ dump)
+        utils.header('Dumped planetlab5 database in %s'%dump)
         return True
 
-    def db_restore(self):
-        dump=self.dbfile("planetab4")
+    def plc_db_restore(self):
+        'restore the planetlab5 DB - looks broken, but run -n might help'
+        dump=self.dbfile("planetab5")
         ##stop httpd service
         self.run_in_guest('service httpd stop')
         # xxx - need another wrapper
-        self.run_in_guest_piped('echo drop database planetlab4','psql --user=pgsqluser template1')
-        self.run_in_guest('createdb -U postgres --encoding=UNICODE --owner=pgsqluser planetlab4')
-        self.run_in_guest('psql -U pgsqluser planetlab4 -f '+dump)
+        self.run_in_guest_piped('echo drop database planetlab5','psql --user=pgsqluser template1')
+        self.run_in_guest('createdb -U postgres --encoding=UNICODE --owner=pgsqluser planetlab5')
+        self.run_in_guest('psql -U pgsqluser planetlab5 -f '+dump)
         ##starting httpd service
         self.run_in_guest('service httpd start')
 
         utils.header('Database restored from ' + dump)
 
+    def standby_1_through_20(self):
+        """convenience function to wait for a specified number of minutes"""
+        pass
     @standby_generic 
     def standby_1(): pass
     @standby_generic