X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=6b910a7e917082ed19f87e82b32193de5b4a1b49;hb=f395e8d3a322aae3b624a441e114931588f09126;hp=d6aed7cb01d2f2f5db71503547b49dcb58f9bf03;hpb=daa6101bf7c504d89de2eb6bbd06821e2f550539;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index d6aed7c..6b910a7 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -107,7 +107,7 @@ class slice_mapper__tasks (object): test_site = TestSite(self,site_spec) test_slice=TestSlice(self,test_site,slice_spec) tasks += slice_method (test_slice, self.options) - return Completer (tasks).run (decorator_self.timeout, decorator_self.silent, decorator_self.period) + return Completer (tasks, message=method.__name__).run (decorator_self.timeout, decorator_self.silent, decorator_self.period) # restore the doc text from the TestSlice method even if a bit odd wrappee.__name__ = method.__name__ wrappee.__doc__ = slice_method.__doc__ @@ -153,23 +153,28 @@ class TestPlc: 'sfa_register_user@1', 'sfa_update_user@1', 'sfa_register_slice@1', 'sfa_renew_slice@1', SEPSFA, 'sfa_remove_user_from_slice@1','sfi_show_slice_researchers@1', 'sfa_insert_user_in_slice@1','sfi_show_slice_researchers@1', SEPSFA, - 'sfa_discover@1', 'sfa_create_slice@1', 'sfa_check_slice_plc@1', 'sfa_update_slice@1', SEPSFA, + 'sfa_discover@1', 'sfa_rspec@1', 'sfa_allocate@1', 'sfa_provision@1', SEPSFA, + 'sfa_check_slice_plc@1', 'sfa_update_slice@1', SEPSFA, 'sfi_list@1', 'sfi_show_site@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.. 'probe_kvm_iptables', 'ping_node', 'ssh_node_debug', 'plcsh_stress_test@1', SEP, 'ssh_node_boot', 'node_bmlogs', 'ssh_slice', 'ssh_slice_basics', 'check_initscripts_ignore', SEP, - 'ssh_slice_sfa@1', 'sfa_delete_slice@1', 'sfa_delete_user@1', SEPSFA, + 'ssh_slice_sfa@1', SEPSFA, + 'sfa_rspec_empty@1', 'sfa_allocate_empty@1', 'sfa_provision_empty@1','sfa_check_slice_plc_empty@1', SEPSFA, + 'sfa_delete_slice@1', 'sfa_delete_user@1', SEPSFA, 'cross_check_tcp@1', 'check_system_slice', SEP, + # for inspecting the slice while it runs the first time + #'fail', # check slices are turned off properly 'empty_slices', 'ssh_slice_off', 'slice_fs_deleted_ignore', SEP, # check they are properly re-created with the same name - 'fill_slices', 'ssh_slice_again_ignore', SEP, + 'fill_slices', 'ssh_slice_again', SEP, 'gather_logs_force', SEP, ] other_steps = [ - 'export', 'show_boxes', SEP, + 'export', 'show_boxes', 'super_speed_up_slices', SEP, 'check_hooks', 'plc_stop', 'plcvm_start', 'plcvm_stop', SEP, 'delete_initscripts', 'delete_nodegroups','delete_all_sites', SEP, 'delete_sites', 'delete_nodes', 'delete_slices', 'keys_clean', SEP, @@ -179,6 +184,7 @@ class TestPlc: '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, + 'sfa_get_expires', SEPSFA, 'plc_db_dump' , 'plc_db_restore', SEP, 'check_netflow','check_drl', SEP, 'debug_nodemanager', 'slice_fs_present', SEP, @@ -273,11 +279,14 @@ class TestPlc: # see e.g. plc_start esp. the version for f14 #command gets run in the plc's vm def host_to_guest(self,command): + vservername=self.vservername + personality=self.options.personality + raw="%(personality)s virsh -c lxc:/// lxc-enter-namespace %(vservername)s"%locals() # f14 still needs some extra help if self.options.fcdistro == 'f14': - raw="virsh -c lxc:/// lxc-enter-namespace %s -- /usr/bin/env PATH=/bin:/sbin:/usr/bin:/usr/sbin %s" %(self.vservername,command) + raw +=" -- /usr/bin/env PATH=/bin:/sbin:/usr/bin:/usr/sbin %(command)s" %locals() else: - raw="virsh -c lxc:/// lxc-enter-namespace %s -- /usr/bin/env %s" %(self.vservername,command) + raw +=" -- /usr/bin/env %(command)s"%locals() return raw # this /vservers thing is legacy... @@ -1021,8 +1030,8 @@ class TestPlc: return False def message (self): return "CompleterTaskBootState with node %s"%self.hostname - def failure_message (self): - return "node %s in state %s - expected %s"%(self.hostname,self.last_boot_state,target_boot_state) + def failure_epilogue (self): + print "node %s in state %s - expected %s"%(self.hostname,self.last_boot_state,target_boot_state) timeout = timedelta(minutes=timeout_minutes) graceout = timedelta(minutes=silent_minutes) @@ -1031,7 +1040,8 @@ class TestPlc: utils.header("checking nodes boot state (expected %s)"%target_boot_state) tasks = [ CompleterTaskBootState (self,hostname) \ for (hostname,_) in self.all_node_infos() ] - return Completer (tasks).run (timeout, graceout, period) + message = 'check_boot_state={}'.format(target_boot_state) + return Completer (tasks, message=message).run (timeout, graceout, period) def nodes_booted(self): return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=28) @@ -1042,28 +1052,28 @@ class TestPlc: return True # probing nodes - def check_nodes_ping(self,timeout_seconds=120,period_seconds=10): - class CompleterTaskPingNode (CompleterTask): + def check_nodes_ping(self, timeout_seconds=30, period_seconds=10): + class CompleterTaskPingNode(CompleterTask): def __init__ (self, hostname): self.hostname=hostname - def run(self,silent): + def run(self, silent): command="ping -c 1 -w 1 %s >& /dev/null"%self.hostname return utils.system (command, silent=silent)==0 - def failure_message (self): - return "Cannot ping node with name %s"%self.hostname + def failure_epilogue (self): + print "Cannot ping node with name %s"%self.hostname timeout=timedelta (seconds=timeout_seconds) graceout=timeout period=timedelta (seconds=period_seconds) node_infos = self.all_node_infos() tasks = [ CompleterTaskPingNode (h) for (h,_) in node_infos ] - return Completer (tasks).run (timeout, graceout, period) + return Completer (tasks, message='ping_node').run (timeout, graceout, period) # ping node before we try to reach ssh, helpful for troubleshooting failing bootCDs def ping_node (self): "Ping nodes" return self.check_nodes_ping () - def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period_seconds=15): + def check_nodes_ssh(self, debug, timeout_minutes, silent_minutes, period_seconds=15): # various delays timeout = timedelta(minutes=timeout_minutes) graceout = timedelta(minutes=silent_minutes) @@ -1071,15 +1081,18 @@ class TestPlc: vservername=self.vservername if debug: message="debug" + completer_message = 'ssh_node_debug' local_key = "keys/%(vservername)s-debug.rsa"%locals() else: message="boot" + completer_message = 'ssh_node_boot' local_key = "keys/key_admin.rsa" utils.header("checking ssh access to nodes (expected in %s mode)"%message) node_infos = self.all_node_infos() - tasks = [ CompleterTaskNodeSsh (nodename, qemuname, local_key, boot_state=message) \ + tasks = [ CompleterTaskNodeSsh (nodename, qemuname, local_key, + boot_state=message, dry_run=self.options.dry_run) \ for (nodename,qemuname) in node_infos ] - return Completer (tasks).run (timeout, graceout, period) + return Completer (tasks, message=completer_message).run (timeout, graceout, period) def ssh_node_debug(self): "Tries to ssh into nodes in debug mode with the debug ssh key" @@ -1134,8 +1147,8 @@ class TestPlc: return self.test_sliver.check_initscript_stamp (self.stamp) def message (self): return "initscript checker for %s"%self.test_sliver.name() - def failure_message (self): - return "initscript stamp %s not found in sliver %s"%(self.stamp,self.test_sliver.name()) + def failure_epilogue (self): + print "initscript stamp %s not found in sliver %s"%(self.stamp,self.test_sliver.name()) tasks=[] for slice_spec in self.plc_spec['slices']: @@ -1152,7 +1165,7 @@ class TestPlc: test_node = TestNode (self,test_site,node) test_sliver = TestSliver (self, test_node, test_slice) tasks.append ( CompleterTaskInitscript (test_sliver, stamp)) - return Completer (tasks).run (timedelta(minutes=5), timedelta(minutes=4), timedelta(seconds=10)) + return Completer (tasks, message='check_initscripts').run (timedelta(minutes=5), timedelta(minutes=4), timedelta(seconds=10)) def check_initscripts(self): "check that the initscripts have triggered" @@ -1213,9 +1226,9 @@ class TestPlc: def ssh_slice(self): pass @slice_mapper__tasks(20,19,15) def ssh_slice_off (self): pass - @slice_mapper__tasks(2,1,15) + @slice_mapper__tasks(1,1,15) def slice_fs_present(self): pass - @slice_mapper__tasks(2,1,15) + @slice_mapper__tasks(1,1,15) def slice_fs_deleted(self): pass # use another name so we can exclude/ignore it from the tests on the nightly command line @@ -1235,20 +1248,28 @@ class TestPlc: return PlcapiUrlScanner (self.auth_root(),ip=self.vserverip).scan() def speed_up_slices (self): - "tweak nodemanager settings on all nodes using a conf file" + "tweak nodemanager cycle (wait time) to 30+/-10 s" + return self._speed_up_slices (30,10) + def super_speed_up_slices (self): + "dev mode: tweak nodemanager cycle (wait time) to 5+/-1 s" + return self._speed_up_slices (5,1) + + def _speed_up_slices (self, p, r): # 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.write('OPTIONS="-p %s -r %s -d"\n'%(p,r)) 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',}) + if not self.apiserver.GetConfFiles (self.auth_root(), + {'dest':'/etc/sysconfig/nodemanager'}): + self.apiserver.AddConfFile (self.auth_root(), + {'dest':'/etc/sysconfig/nodemanager', + 'source':'PlanetLabConf/nodemanager', + 'postinstall_cmd':'service nm restart',}) return True def debug_nodemanager (self): @@ -1289,20 +1310,23 @@ class TestPlc: for spec in specs: port = spec['port'] # server side - 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=20): - overall=False + # the issue here is that we have the server run in background + # and so we have no clue if it took off properly or not + # looks like in some cases it does not + 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=20): + overall = False break # idem for the client side - c_test_sliver = self.locate_sliver_obj_cross (spec['client_node'],spec['client_slice'],other_plcs) + c_test_sliver = self.locate_sliver_obj_cross (spec['client_node'], spec['client_slice'], other_plcs) # use nodename from locatesd sliver, unless 'client_connect' is set if 'client_connect' in spec: destination = spec['client_connect'] else: - destination=s_test_sliver.test_node.name() - if not c_test_sliver.run_tcp_client(destination,port): - overall=False + destination = s_test_sliver.test_node.name() + if not c_test_sliver.run_tcp_client(destination, port): + overall = False return overall # painfully enough, we need to allow for some time as netflow might show up last @@ -1327,14 +1351,14 @@ class TestPlc: return self.test_node._check_system_slice (slicename, dry_run=self.dry_run) def message (self): return "System slice %s @ %s"%(slicename, self.test_node.name()) - def failure_message (self): - return "COULD not find system slice %s @ %s"%(slicename, self.test_node.name()) + def failure_epilogue (self): + print "COULD not find system slice %s @ %s"%(slicename, self.test_node.name()) timeout = timedelta(minutes=timeout_minutes) silent = timedelta (0) period = timedelta (seconds=period_seconds) tasks = [ CompleterTaskSystemSlice (test_node, self.options.dry_run) \ for test_node in self.all_nodes() ] - return Completer (tasks) . run (timeout, silent, period) + return Completer (tasks, message='_check_system_slice') . run (timeout, silent, period) def plcsh_stress_test (self): "runs PLCAPI stress test, that checks Add/Update/Delete on all types - preserves contents" @@ -1508,11 +1532,11 @@ class TestPlc: def aggregate_xml_line(self): port=self.plc_spec['sfa']['neighbours-port'] return '' % \ - (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'],port) + (self.vserverip,self.plc_spec['sfa']['settings']['SFA_REGISTRY_ROOT_AUTH'],port) def registry_xml_line(self): return '' % \ - (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH']) + (self.vserverip,self.plc_spec['sfa']['settings']['SFA_REGISTRY_ROOT_AUTH']) # a cross step that takes all other plcs in argument @@ -1530,11 +1554,11 @@ class TestPlc: " ".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 + and self.test_ssh.copy_abs(reg_fname,'/%s/etc/sfa/registries.xml'%self.vm_root_in_host())==0 def sfa_import(self): "use sfaadmin to import from plc" - auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'] + auth=self.plc_spec['sfa']['settings']['SFA_REGISTRY_ROOT_AUTH'] return self.run_in_guest('sfaadmin reg import_registry')==0 def sfa_start(self): @@ -1568,6 +1592,17 @@ class TestPlc: self.run_in_guest("rm -rf /root/sfi") return True + def sfa_rspec_empty(self): + "expose a static empty rspec (ships with the tests module) in the sfi directory" + filename="empty-rspec.xml" + overall=True + for slice_spec in self.plc_spec['sfa']['auth_sfa_specs']: + test_slice=TestAuthSfa(self,slice_spec) + in_vm = test_slice.sfi_path() + remote="%s/%s"%(self.vm_root_in_host(),in_vm) + if self.test_ssh.copy_abs (filename, remote) !=0: overall=False + return overall + @auth_sfa_mapper def sfa_register_site (self): pass @auth_sfa_mapper @@ -1581,12 +1616,24 @@ class TestPlc: @auth_sfa_mapper def sfa_renew_slice(self): pass @auth_sfa_mapper + def sfa_get_expires(self): pass + @auth_sfa_mapper def sfa_discover(self): pass @auth_sfa_mapper - def sfa_create_slice(self): pass + def sfa_rspec(self): pass + @auth_sfa_mapper + def sfa_allocate(self): pass + @auth_sfa_mapper + def sfa_allocate_empty(self): pass + @auth_sfa_mapper + def sfa_provision(self): pass + @auth_sfa_mapper + def sfa_provision_empty(self): pass @auth_sfa_mapper def sfa_check_slice_plc(self): pass @auth_sfa_mapper + def sfa_check_slice_plc_empty(self): pass + @auth_sfa_mapper def sfa_update_slice(self): pass @auth_sfa_mapper def sfa_remove_user_from_slice(self): pass @@ -1798,3 +1845,4 @@ class TestPlc: # convenience for debugging the test logic def yes (self): return True def no (self): return False + def fail (self): return False