this is how TestMapper deals with subfields
[tests.git] / system / Substrate.py
index 53b0c89..b776042 100644 (file)
@@ -59,6 +59,9 @@ import utils
 from TestSsh import TestSsh
 from TestMapper import TestMapper
 
+# too painful to propagate this cleanly
+verbose=None
+
 def header (message,banner=True):
     if not message: return
     if banner: print "===============",
@@ -230,7 +233,7 @@ class Pool:
         print 'Done'
         for (vname,bname) in self.load_starting():
             self.substrate.add_starting_dummy (bname, vname)
-        print 'After starting: IP pool'
+        print "After having loaded 'starting': IP pool"
         print self.line()
     # OS-dependent ping option (support for macos, for convenience)
     ping_timeout_option = None
@@ -261,13 +264,73 @@ class Box:
         self.test_ssh().run("shutdown -r now",message="Rebooting %s"%self.hostname,
                             dry_run=options.dry_run)
 
+    def hostname_fedora (self,virt=None):
+        result = "%s {"%self.hostname
+        if virt: result += "%s-"%virt
+        result += "%s"%self.fedora()
+        # too painful to propagate this cleanly
+        global verbose
+        if verbose:
+            result += "-%s" % self.uname()
+        result += "}"
+        return result
+
+    separator = "===composite==="
+
+    # probe the ssh link
+    # take this chance to gather useful stuff
+    def probe (self):
+        # try it only once
+        if self._probed is not None: return self._probed
+        composite_command = [ ]
+        composite_command += [ "hostname" ]
+        composite_command += [ ";" , "echo", Box.separator , ";" ]
+        composite_command += [ "uptime" ]
+        composite_command += [ ";" , "echo", Box.separator , ";" ]
+        composite_command += [ "uname", "-r"]
+        composite_command += [ ";" , "echo", Box.separator , ";" ]
+        composite_command += [ "cat" , "/etc/fedora-release" ]
+
+        # due to colons and all, this is going wrong on the local box (typically testmaster)
+        # I am reluctant to change TestSsh as it might break all over the place, so
+        if self.test_ssh().is_local():
+            probe_argv = [ "bash", "-c", " ".join (composite_command) ]
+        else:
+            probe_argv=self.test_ssh().actual_argv(composite_command)
+        composite=self.backquote ( probe_argv, trash_err=True )
+        self._hostname = self._uptime = self._uname = self._fedora = "** Unknown **"
+        if not composite: 
+            print "root@%s unreachable"%self.hostname
+            self._probed=''
+        else:
+            try:
+                pieces = composite.split(Box.separator)
+                pieces = [ x.strip() for x in pieces ]
+                [self._hostname, self._uptime, self._uname, self._fedora] = pieces
+                # customize
+                self._uptime = ', '.join([ x.strip() for x in self._uptime.split(',')[2:]])
+                self._fedora = self._fedora.replace("Fedora release ","f").split(" ")[0]
+            except:
+                import traceback
+                print 'BEG issue with pieces',pieces
+                traceback.print_exc()
+                print 'END issue with pieces',pieces
+            self._probed=self._hostname
+        return self._probed
+
+    # use argv=['bash','-c',"the command line"]
     def uptime(self):
+        self.probe()
         if hasattr(self,'_uptime') and self._uptime: return self._uptime
-        return '*undef* uptime'
-    def sense_uptime (self):
-        command=['uptime']
-        self._uptime=self.backquote_ssh(command,trash_err=True).strip()
-        if not self._uptime: self._uptime='unreachable'
+        return '*unprobed* uptime'
+    def uname(self):
+        self.probe()
+        if hasattr(self,'_uname') and self._uname: return self._uname
+        return '*unprobed* uname'
+    def fedora(self):
+        self.probe()
+        if hasattr(self,'_fedora') and self._fedora: return self._fedora
+        return '*unprobed* fedora'
 
     def run(self,argv,message=None,trash_err=False,dry_run=False):
         if dry_run:
@@ -296,15 +359,6 @@ class Box:
             result= subprocess.Popen(argv,stdout=subprocess.PIPE,stderr=file('/dev/null','w')).communicate()[0]
         return result
 
-    def probe (self):
-        if self._probed is not None: return self._probed
-        # first probe the ssh link
-        probe_argv=self.test_ssh().actual_argv(['hostname'])
-        self._probed=self.backquote ( probe_argv, trash_err=True )
-        if not self._probed: print "root@%s unreachable"%self.hostname
-        return self._probed
-
-    # use argv=['bash','-c',"the command line"]
     # if you have any shell-expanded arguments like *
     # and if there's any chance the command is adressed to the local host
     def backquote_ssh (self, argv, trash_err=False):
@@ -338,9 +392,9 @@ class BuildBox (Box):
 
     def list(self, verbose=False):
         if not self.build_instances: 
-            header ('No build process on %s (%s)'%(self.hostname,self.uptime()))
+            header ('No build process on %s (%s)'%(self.hostname_fedora(),self.uptime()))
         else:
-            header ("Builds on %s (%s)"%(self.hostname,self.uptime()))
+            header ("Builds on %s (%s)"%(self.hostname_fedora(),self.uptime()))
             for b in self.build_instances: 
                 header (b.line(),banner=False)
 
@@ -348,35 +402,39 @@ class BuildBox (Box):
         if not options.soft:
             Box.reboot(self,options)
         else:
-            command=['pkill','vbuild']
+            self.soft_reboot (options)
+
+build_matcher=re.compile("\s*(?P<pid>[0-9]+).*-[bo]\s+(?P<buildname>[^\s]+)(\s|\Z)")
+build_matcher_initvm=re.compile("\s*(?P<pid>[0-9]+).*initvm.*\s+(?P<buildname>[^\s]+)\s*\Z")
+
+class BuildLxcBox (BuildBox):
+    def soft_reboot (self, options):
+            command=['pkill','lbuild']
             self.run_ssh(command,"Terminating vbuild processes",dry_run=options.dry_run)
 
     # inspect box and find currently running builds
-    matcher=re.compile("\s*(?P<pid>[0-9]+).*-[bo]\s+(?P<buildname>[^\s]+)(\s|\Z)")
-    matcher_building_vm=re.compile("\s*(?P<pid>[0-9]+).*init-vserver.*\s+(?P<buildname>[^\s]+)\s*\Z")
     def sense(self, options):
-        print 'bb',
-        self.sense_uptime()
-        pids=self.backquote_ssh(['pgrep','vbuild'],trash_err=True)
+        print 'xb',
+        pids=self.backquote_ssh(['pgrep','lbuild'],trash_err=True)
         if not pids: return
         command=['ps','-o','pid,command'] + [ pid for pid in pids.split("\n") if pid]
         ps_lines=self.backquote_ssh (command).split('\n')
         for line in ps_lines:
             if not line.strip() or line.find('PID')>=0: continue
-            m=BuildBox.matcher.match(line)
+            m=build_matcher.match(line)
             if m: 
                 date=time.strftime('%Y-%m-%d',time.localtime(time.time()))
                 buildname=m.group('buildname').replace('@DATE@',date)
                 self.add_build (buildname,m.group('pid'))
                 continue
-            m=BuildBox.matcher_building_vm.match(line)
+            m=build_matcher_initvm.match(line)
             if m: 
                 # buildname is expansed here
                 self.add_build (buildname,m.group('pid'))
                 continue
-            header('BuildBox.sense: command %r returned line that failed to match'%command)
+            header('BuildLxcBox.sense: command %r returned line that failed to match'%command)
             header(">>%s<<"%line)
-
+    
 ############################################################
 class PlcInstance:
     def __init__ (self, plcbox):
@@ -388,31 +446,6 @@ class PlcInstance:
     def set_now (self): self.timestamp=int(time.time())
     def pretty_timestamp (self): return time.strftime("%Y-%m-%d:%H-%M",time.localtime(self.timestamp))
 
-class PlcVsInstance (PlcInstance):
-    def __init__ (self, plcbox, vservername, ctxid):
-        PlcInstance.__init__(self,plcbox)
-        self.vservername=vservername
-        self.ctxid=ctxid
-
-    def vplcname (self):
-        return self.vservername.split('-')[-1]
-    def buildname (self):
-        return self.vservername.rsplit('-',2)[0]
-
-    def line (self):
-        msg="== %s =="%(self.vplcname())
-        msg += " [=%s]"%self.vservername
-        if self.ctxid==0:  msg+=" not (yet?) running"
-        else:              msg+=" (ctx=%s)"%self.ctxid     
-        if self.timestamp: msg += " @ %s"%self.pretty_timestamp()
-        else:              msg += " *unknown timestamp*"
-        return msg
-
-    def kill (self):
-        msg="vserver stopping %s on %s"%(self.vservername,self.plc_box.hostname)
-        self.plc_box.run_ssh(['vserver',self.vservername,'stop'],msg)
-        self.plc_box.forget(self)
-
 class PlcLxcInstance (PlcInstance):
     # does lxc have a context id of any kind ?
     def __init__ (self, plcbox, lxcname, pid):
@@ -453,7 +486,7 @@ class PlcBox (Box):
 
     # fill one slot even though this one is not started yet
     def add_dummy (self, plcname):
-        dummy=PlcVsInstance(self,'dummy_'+plcname,0)
+        dummy=PlcLxcInstance(self,'dummy_'+plcname,0)
         dummy.set_now()
         self.plc_instances.append(dummy)
 
@@ -475,88 +508,7 @@ class PlcBox (Box):
             for p in self.plc_instances: 
                 header (p.line(),banner=False)
 
-    def get_uname(self):
-        self._uname=self.backquote_ssh(['uname','-r']).strip()
-
-    # expecting sense () to have filled self._uname
-    def uname(self):
-        if hasattr(self,'_uname') and self._uname: return self._uname
-        return '*undef* uname'
-
-class PlcVsBox (PlcBox):
-
-    def add_vserver (self,vservername,ctxid):
-        for plc in self.plc_instances:
-            if plc.vservername==vservername: 
-                header("WARNING, duplicate myplc %s running on %s"%\
-                           (vservername,self.hostname),banner=False)
-                return
-        self.plc_instances.append(PlcVsInstance(self,vservername,ctxid))
-    
-    def line(self): 
-        msg="%s [max=%d,free=%d, VS-based] (%s)"%(self.hostname, self.max_plcs,self.free_slots(),self.uname())
-        return msg
-        
-    def plc_instance_by_vservername (self, vservername):
-        for p in self.plc_instances:
-            if p.vservername==vservername: return p
-        return None
-
-    def soft_reboot (self, options):
-        self.run_ssh(['service','util-vserver','stop'],"Stopping all running vservers on %s"%(self.hostname,),
-                     dry_run=options.dry_run)
-
-    def sense (self, options):
-        print 'vp',
-        self.get_uname()
-        # try to find fullname (vserver_stat truncates to a ridiculously short name)
-        # fetch the contexts for all vservers on that box
-        map_command=['grep','.','/etc/vservers/*/context','/dev/null',]
-        context_map=self.backquote_ssh (map_command)
-        # at this point we have a set of lines like
-        # /etc/vservers/2010.01.20--k27-f12-32-vplc03/context:40144
-        ctx_dict={}
-        for map_line in context_map.split("\n"):
-            if not map_line: continue
-            [path,xid] = map_line.split(':')
-            ctx_dict[xid]=os.path.basename(os.path.dirname(path))
-        # at this point ctx_id maps context id to vservername
-
-        command=['vserver-stat']
-        vserver_stat = self.backquote_ssh (command)
-        for vserver_line in vserver_stat.split("\n"):
-            if not vserver_line: continue
-            context=vserver_line.split()[0]
-            if context=="CTX": continue
-            try:
-                longname=ctx_dict[context]
-                self.add_vserver(longname,context)
-            except:
-                print 'WARNING: found ctx %s in vserver_stat but was unable to figure a corresp. vserver'%context
-
-        # scan timestamps 
-        running_vsnames = [ i.vservername for i in self.plc_instances ]
-        command=   ['grep','.']
-        command += ['/vservers/%s.timestamp'%vs for vs in running_vsnames]
-        command += ['/dev/null']
-        ts_lines=self.backquote_ssh(command,trash_err=True).split('\n')
-        for ts_line in ts_lines:
-            if not ts_line.strip(): continue
-            # expect /vservers/<vservername>.timestamp:<timestamp>
-            try:
-                (ts_file,timestamp)=ts_line.split(':')
-                ts_file=os.path.basename(ts_file)
-                (vservername,_)=os.path.splitext(ts_file)
-                timestamp=int(timestamp)
-                p=self.plc_instance_by_vservername(vservername)
-                if not p: 
-                    print 'WARNING zombie plc',self.hostname,ts_line
-                    print '... was expecting',vservername,'in',[i.vservername for i in self.plc_instances]
-                    continue
-                p.set_timestamp(timestamp)
-            except:  print 'WARNING, could not parse ts line',ts_line
-        
-
+## we do not this at INRIA any more
 class PlcLxcBox (PlcBox):
 
     def add_lxc (self,lxcname,pid):
@@ -570,8 +522,10 @@ class PlcLxcBox (PlcBox):
 
     # a line describing the box
     def line(self): 
-        msg="%s [max=%d,free=%d, LXC-based] (%s)"%(self.hostname, self.max_plcs,self.free_slots(),self.uname())
-        return msg
+        return "%s [max=%d,free=%d] (%s)"%(self.hostname_fedora(virt="lxc"), 
+                                           self.max_plcs,self.free_slots(),
+                                           self.uptime(),
+                                           )
     
     def plc_instance_by_lxcname (self, lxcname):
         for p in self.plc_instances:
@@ -588,10 +542,8 @@ class PlcLxcBox (PlcBox):
 
     # sense is expected to fill self.plc_instances with PlcLxcInstance's 
     # to describe the currently running VM's
-    # as well as to call  self.get_uname() once
     def sense (self, options):
         print "xp",
-        self.get_uname()
        command="rsync lxc-driver.sh  %s:/root"%self.hostname
         commands.getstatusoutput(command)
        command=['/root/lxc-driver.sh','-c','sense_all']
@@ -602,7 +554,8 @@ class PlcLxcBox (PlcBox):
            pid=lxc_line.split(";")[1]
            timestamp=lxc_line.split(";")[2]
             self.add_lxc(lxcname,pid)
-            timestamp=int(timestamp)
+            try: timestamp=int(timestamp)
+            except: timestamp=0
             p=self.plc_instance_by_lxcname(lxcname)
             if not p:
                 print 'WARNING zombie plc',self.hostname,lxcname
@@ -657,6 +610,9 @@ class QemuBox (Box):
                 return
         self.qemu_instances.append(QemuInstance(nodename,pid,self))
 
+    def node_names (self):
+        return [ qi.nodename for qi in self.qemu_instances ]
+
     def forget (self, qemu_instance):
         self.qemu_instances.remove(qemu_instance)
 
@@ -667,14 +623,15 @@ class QemuBox (Box):
         self.qemu_instances.append(dummy)
 
     def line (self):
-        msg="%s [max=%d,free=%d] (%s)"%(self.hostname, self.max_qemus,self.free_slots(),self.driver())
-        return msg
+        return "%s [max=%d,free=%d] (%s) %s"%(
+            self.hostname_fedora(virt="qemu"), self.max_qemus,self.free_slots(),
+            self.uptime(),self.driver())
 
     def list(self, verbose=False):
         if not self.qemu_instances: 
-            header ('No qemu process on %s'%(self.line()))
+            header ('No qemu on %s'%(self.line()))
         else:
-            header ("Active qemu processes on %s"%(self.line()))
+            header ("Qemus on %s"%(self.line()))
             self.qemu_instances.sort(timestamp_sort)
             for q in self.qemu_instances: 
                 header (q.line(),banner=False)
@@ -714,7 +671,7 @@ class QemuBox (Box):
                 self._driver='kqemu module loaded'
             # kvm might be loaded without kvm_intel (we dont have AMD)
             elif module.find('kvm_intel')==0:
-                self._driver='kvm_intel module loaded'
+                self._driver='kvm_intel OK'
         ########## find out running pids
         pids=self.backquote_ssh(['pgrep','qemu'])
         if not pids: return
@@ -789,17 +746,30 @@ class TestInstance:
     def set_broken (self, plcindex, step): 
         self.broken_steps.append ( (plcindex, step,) )
 
+    def second_letter (self):
+        if not self.broken_steps: return '='
+        else:
+            really_broken = [ step for (i,step) in self.broken_steps if '_ignore' not in step ]
+            # W is for warning like what's in the build mail
+            if len(really_broken)==0: return 'W'
+            else: return 'B'
+
     def line (self):
-        double='=='
-        if self.pids: double='*'+double[1]
-        if self.broken_steps: double=double[0]+'B'
+        # make up a 2-letter sign
+        # first letter : '=', unless build is running : '*'
+        double = '*' if self.pids else '='
+        # second letter : '=' if fine, 'W' for warnings (only ignored steps) 'B' for broken
+        letter2 = self.second_letter()
+        double += letter2
         msg = " %s %s =="%(double,self.buildname)
         if not self.pids:       pass
         elif len(self.pids)==1: msg += " (pid=%s)"%self.pids[0]
         else:                   msg += " !!!pids=%s!!!"%self.pids
         msg += " @%s"%self.pretty_timestamp()
-        if self.broken_steps:
-            msg += " [BROKEN=" + " ".join( [ "%s@%s"%(s,i) for (i,s) in self.broken_steps ] ) + "]"
+        if letter2 != '=':
+            msg2 = ( ' BROKEN' if letter2 == 'B' else ' WARNING' )
+            # sometimes we have an empty plcindex
+            msg += " [%s="%msg2 + " ".join( [ "%s@%s"%(s,i) if i else s for (i,s) in self.broken_steps ] ) + "]"
         return msg
 
 class TestBox (Box):
@@ -847,9 +817,9 @@ class TestBox (Box):
 
     matcher_proc=re.compile (".*/proc/(?P<pid>[0-9]+)/cwd.*/root/(?P<buildname>[^/]+)$")
     matcher_grep=re.compile ("/root/(?P<buildname>[^/]+)/logs/trace.*:TRACE:\s*(?P<plcindex>[0-9]+).*step=(?P<step>\S+).*")
+    matcher_grep_missing=re.compile ("grep: /root/(?P<buildname>[^/]+)/logs/trace: No such file or directory")
     def sense (self, options):
         print 'tm',
-        self.sense_uptime()
         self.starting_ips=[x for x in self.backquote_ssh(['cat',Starting.location], trash_err=True).strip().split('\n') if x]
 
         # scan timestamps on all tests
@@ -868,10 +838,19 @@ class TestBox (Box):
                 t=self.add_timestamp(buildname,timestamp)
             except:  print 'WARNING, could not parse ts line',ts_line
 
-        command=['bash','-c',"grep KO /root/*/logs/trace-* /dev/null" ]
+        # let's try to be robust here -- tests that fail very early like e.g.
+        # "Cannot make space for a PLC instance: vplc IP pool exhausted", that occurs as part of provision
+        # will result in a 'trace' symlink to an inexisting 'trace-<>.txt' because no step has gone through
+        # simple 'trace' should exist though as it is created by run_log
+        command=['bash','-c',"grep KO /root/*/logs/trace /dev/null 2>&1" ]
         trace_lines=self.backquote_ssh (command).split('\n')
         for line in trace_lines:
             if not line.strip(): continue
+            m=TestBox.matcher_grep_missing.match(line)
+            if m:
+                buildname=m.group('buildname')
+                self.add_broken(buildname,'','NO STEP DONE')
+                continue
             m=TestBox.matcher_grep.match(line)
             if m: 
                 buildname=m.group('buildname')
@@ -899,16 +878,16 @@ class TestBox (Box):
         
         
     def line (self):
-        return "%s (%s)"%(self.hostname,self.uptime())
+        return self.hostname_fedora()
 
     def list (self, verbose=False):
         # verbose shows all tests
         if verbose:
             instances = self.test_instances
-            msg="knwown tests"
+            msg="tests"
         else:
             instances = [ i for i in self.test_instances if i.is_running() ]
-            msg="known running tests"
+            msg="running tests"
 
         if not instances:
             header ("No %s on %s"%(msg,self.line()))
@@ -929,40 +908,28 @@ class Options: pass
 
 class Substrate:
 
-    def __init__ (self, plcs_on_vs=True, plcs_on_lxc=False):
+    def __init__ (self):
         self.options=Options()
         self.options.dry_run=False
         self.options.verbose=False
         self.options.reboot=False
         self.options.soft=False
         self.test_box = TestBox (self.test_box_spec())
-        self.build_boxes = [ BuildBox(h) for h in self.build_boxes_spec() ]
-        # for compat with older LocalSubstrate
-        try:
-            self.plc_vs_boxes = [ PlcVsBox (h,m) for (h,m) in self.plc_vs_boxes_spec ()]
-            self.plc_lxc_boxes = [ PlcLxcBox (h,m) for (h,m) in self.plc_lxc_boxes_spec ()]
-        except:
-            self.plc_vs_boxes = [ PlcVsBox (h,m) for (h,m) in self.plc_boxes_spec ()]
-            self.plc_lxc_boxes = [ ]
+        self.build_lxc_boxes = [ BuildLxcBox(h) for h in self.build_lxc_boxes_spec() ]
+        self.plc_lxc_boxes = [ PlcLxcBox (h,m) for (h,m) in self.plc_lxc_boxes_spec ()]
         self.qemu_boxes = [ QemuBox (h,m) for (h,m) in self.qemu_boxes_spec ()]
         self._sensed=False
 
         self.vplc_pool = Pool (self.vplc_ips(),"for vplcs",self)
         self.vnode_pool = Pool (self.vnode_ips(),"for vnodes",self)
         
-        self.rescope (plcs_on_vs=plcs_on_vs, plcs_on_lxc=plcs_on_lxc)
-
-    # which plc boxes are we interested in ?
-    def rescope (self, plcs_on_vs, plcs_on_lxc):
-        self.plc_boxes=[]
-        if plcs_on_vs: self.plc_boxes += self.plc_vs_boxes
-        if plcs_on_lxc: self.plc_boxes += self.plc_lxc_boxes
+        self.build_boxes = self.build_lxc_boxes
+        self.plc_boxes = self.plc_lxc_boxes
         self.default_boxes = self.plc_boxes + self.qemu_boxes
         self.all_boxes = self.build_boxes + [ self.test_box ] + self.plc_boxes + self.qemu_boxes
 
     def summary_line (self):
         msg  = "["
-        msg += " %d vp"%len(self.plc_vs_boxes)
         msg += " %d xp"%len(self.plc_lxc_boxes)
         msg += " %d tried plc boxes"%len(self.plc_boxes)
         msg += "]"
@@ -1105,12 +1072,14 @@ class Substrate:
                                    'name': plc_name,
                                    'vservername':vservername,
                                    'vserverip':vplc_ip,
-                                   'PLC_DB_HOST':vplc_hostname,
-                                   'PLC_API_HOST':vplc_hostname,
-                                   'PLC_BOOT_HOST':vplc_hostname,
-                                   'PLC_WWW_HOST':vplc_hostname,
-                                   'PLC_NET_DNS1' : self.network_settings() [ 'interface_fields:dns1' ],
-                                   'PLC_NET_DNS2' : self.network_settings() [ 'interface_fields:dns2' ],
+#                                   'settings': {
+                                   'settings:PLC_DB_HOST':vplc_hostname,
+                                   'settings:PLC_API_HOST':vplc_hostname,
+                                   'settings:PLC_BOOT_HOST':vplc_hostname,
+                                   'settings:PLC_WWW_HOST':vplc_hostname,
+                                   'settings:PLC_NET_DNS1' : self.network_settings() [ 'interface_fields:dns1' ],
+                                   'settings:PLC_NET_DNS2' : self.network_settings() [ 'interface_fields:dns2' ],
+#                                      }
                                    } ) ]
                   }
 
@@ -1199,11 +1168,11 @@ class Substrate:
 
     def localize_sfa_rspec (self,plc,options):
        
-        plc['sfa']['SFA_REGISTRY_HOST'] = plc['PLC_DB_HOST']
-        plc['sfa']['SFA_AGGREGATE_HOST'] = plc['PLC_DB_HOST']
-        plc['sfa']['SFA_SM_HOST'] = plc['PLC_DB_HOST']
-        plc['sfa']['SFA_DB_HOST'] = plc['PLC_DB_HOST']
-        plc['sfa']['SFA_PLC_URL'] = 'https://' + plc['PLC_API_HOST'] + ':443/PLCAPI/' 
+        plc['sfa']['settings']['SFA_REGISTRY_HOST'] = plc['settings']['PLC_DB_HOST']
+        plc['sfa']['settings']['SFA_AGGREGATE_HOST'] = plc['settings']['PLC_DB_HOST']
+        plc['sfa']['settings']['SFA_SM_HOST'] = plc['settings']['PLC_DB_HOST']
+        plc['sfa']['settings']['SFA_DB_HOST'] = plc['settings']['PLC_DB_HOST']
+        plc['sfa']['settings']['SFA_PLC_URL'] = 'https://%s:443/PLCAPI/' % plc['settings']['PLC_API_HOST']
        return plc
 
     #################### release:
@@ -1215,29 +1184,59 @@ class Substrate:
     #################### show results for interactive mode
     def get_box (self,boxname):
         for b in self.build_boxes + self.plc_boxes + self.qemu_boxes + [self.test_box] :
-            if b.shortname()==boxname:
-                return b
+            if b.shortname()==boxname:                          return b
+            try:
+                if b.shortname()==boxname.split('.')[0]:        return b
+            except: pass
         print "Could not find box %s"%boxname
         return None
 
-    def list_boxes(self,box_or_names):
-        print 'Sensing',
+    # deal with the mix of boxes and names and stores the current focus 
+    # as a list of Box instances in self.focus_all
+    def normalize (self, box_or_names):
+        self.focus_all=[]
         for box in box_or_names:
             if not isinstance(box,Box): box=self.get_box(box)
-            if not box: continue
+            if not box: 
+                print 'Warning - could not handle box',box
+            self.focus_all.append(box)
+        # elaborate by type
+        self.focus_build = [ x for x in self.focus_all if isinstance(x,BuildBox) ]
+        self.focus_plc = [ x for x in self.focus_all if isinstance(x,PlcBox) ]
+        self.focus_qemu = [ x for x in self.focus_all if isinstance(x,QemuBox) ]
+                             
+    def list_boxes(self):
+        print 'Sensing',
+        for box in self.focus_all:
             box.sense(self.options)
         print 'Done'
-        for box in box_or_names:
-            if not isinstance(box,Box): box=self.get_box(box)
-            if not box: continue
+        for box in self.focus_all:
             box.list(self.options.verbose)
 
-    def reboot_boxes(self,box_or_names):
-        for box in box_or_names:
-            if not isinstance(box,Box): box=self.get_box(box)
-            if not box: continue
+    def reboot_boxes(self):
+        for box in self.focus_all:
             box.reboot(self.options)
 
+    def sanity_check (self):
+        print 'Sanity check'
+        self.sanity_check_plc()
+        self.sanity_check_qemu()
+
+    def sanity_check_plc (self):
+        pass
+
+    def sanity_check_qemu (self):
+        all_nodes=[]
+        for box in self.focus_qemu:
+            all_nodes += box.node_names()
+        hash={}
+        for node in all_nodes:
+            if node not in hash: hash[node]=0
+            hash[node]+=1
+        for (node,count) in hash.items():
+            if count!=1: print 'WARNING - duplicate node',node
+        
+
     ####################
     # can be run as a utility to probe/display/manage the local infrastructure
     def main (self):
@@ -1245,7 +1244,7 @@ class Substrate:
         parser.add_option ('-r',"--reboot",action='store_true',dest='reboot',default=False,
                            help='reboot mode (use shutdown -r)')
         parser.add_option ('-s',"--soft",action='store_true',dest='soft',default=False,
-                           help='soft mode for reboot (vserver stop or kill qemus)')
+                           help='soft mode for reboot (terminates processes)')
         parser.add_option ('-t',"--testbox",action='store_true',dest='testbox',default=False,
                            help='add test box') 
         parser.add_option ('-b',"--build",action='store_true',dest='builds',default=False,
@@ -1262,8 +1261,6 @@ class Substrate:
                            help='dry run mode')
         (self.options,args)=parser.parse_args()
 
-        self.rescope (plcs_on_vs=True, plcs_on_lxc=True)
-
         boxes=args
         if self.options.testbox: boxes += [self.test_box]
         if self.options.builds: boxes += self.build_boxes
@@ -1271,9 +1268,16 @@ class Substrate:
         if self.options.qemus: boxes += self.qemu_boxes
         if self.options.all: boxes += self.all_boxes
         
+        global verbose
+        verbose=self.options.verbose
         # default scope is -b -p -q -t
         if not boxes:
             boxes = self.build_boxes + self.plc_boxes + self.qemu_boxes + [self.test_box]
 
-        if self.options.reboot: self.reboot_boxes (boxes)
-        else:                   self.list_boxes (boxes)
+        self.normalize (boxes)
+
+        if self.options.reboot:
+            self.reboot_boxes ()
+        else:
+            self.list_boxes ()
+            self.sanity_check ()