X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=ccb90a53045fc2db09cadfdca886e193b4a8244a;hb=3b3d113e498e6a9c1ed0b9d1a12fb9d49c025d5c;hp=fbd32950f6a6abea38c8c1649184643c7dbcaa3c;hpb=f8436f56fe24e8e545c9b3669897b8332dddffbc;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index fbd3295..ccb90a5 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -19,8 +19,7 @@ from TestSliver import TestSliver from TestBoxQemu import TestBoxQemu from TestSsh import TestSsh from TestApiserver import TestApiserver -from TestSliceSfa import TestSliceSfa -from TestUserSfa import TestUserSfa +from TestAuthSfa import TestAuthSfa # step methods must take (self) and return a boolean (options is a member of the class) @@ -39,17 +38,14 @@ 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__=method.__doc__ + actual.__doc__=TestNode.__dict__[method.__name__].__doc__ return actual def slice_mapper (method): @@ -63,21 +59,19 @@ def slice_mapper (method): if not slice_method(test_slice,self.options): overall=False return overall # restore the doc text - actual.__doc__=method.__doc__ + actual.__doc__=TestSlice.__dict__[method.__name__].__doc__ return actual -def slice_sfa_mapper (method): +def auth_sfa_mapper (method): def actual(self): overall=True - slice_method = TestSliceSfa.__dict__[method.__name__] - for slice_spec in self.plc_spec['sfa']['sfa_slice_specs']: - site_spec = self.locate_site (slice_spec['sitename']) - test_site = TestSite(self,site_spec) - test_slice=TestSliceSfa(self,test_site,slice_spec) + slice_method = TestAuthSfa.__dict__[method.__name__] + for slice_spec in self.plc_spec['sfa']['auth_sfa_specs']: + test_slice=TestAuthSfa(self,slice_spec) if not slice_method(test_slice,self.options): overall=False return overall # restore the doc text - actual.__doc__=method.__doc__ + actual.__doc__=TestAuthSfa.__dict__[method.__name__].__doc__ return actual SEP='' @@ -86,37 +80,40 @@ SEPSFA='' class TestPlc: default_steps = [ - 'show', 'timestamp_plc', 'timestamp_qemu', SEP, - 'vs_delete','vs_create','plc_install', 'plc_configure', 'plc_start', SEP, - 'keys_fetch', 'keys_store', 'keys_clear_known_hosts', SEP, + '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', 'qemu_export', 'qemu_kill_all', 'qemu_start', SEP, - 'sfa_install', 'sfa_configure', 'cross_sfa_configure', 'sfa_import', 'sfa_start', SEPSFA, - 'sfi_configure@1', 'sfa_add_user@1', 'sfa_add_slice@1', 'sfa_discover@1', SEPSFA, - 'sfa_create_slice@1', 'sfa_check_slice_plc@1', SEPSFA, - 'sfa_update_user@1', 'sfa_update_slice@1', 'sfa_view@1', 'sfa_utest@1',SEPSFA, + '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_discover@1', SEPSFA, + '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', 'plcsh_stress_test@1', SEP, - 'ssh_node_boot', 'ssh_slice', 'check_initscripts', SEP, + '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, - 'check_tcp', SEP, - 'force_gather_logs', 'force_local_post', SEP, + 'check_tcp', 'check_sys_slice', SEP, + 'empty_slices', 'ssh_slice_off', 'fill_slices', SEP, + 'force_gather_logs', SEP, ] other_steps = [ - 'check_hooks', - 'free_all', - 'show_boxes', 'local_list','local_rel','local_rel_plc','local_rel_qemu',SEP, - 'plc_stop', 'vs_start', 'vs_stop', SEP, + '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, + 'populate', SEP, 'nodestate_show','nodestate_safeboot','nodestate_boot', SEP, - 'qemu_list_all', 'qemu_list_mine', 'qemu_kill_mine', SEP, - 'sfa_plcclean', 'sfa_dbclean', 'sfa_stop','sfa_uninstall', 'sfi_clean', 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, - 'standby_1 through 20',SEP, + 'standby_1_through_20',SEP, ] @staticmethod @@ -128,37 +125,41 @@ class TestPlc: return step != SEP and step != SEPSFA # turn off the sfa-related steps when build has skipped SFA - # this is originally for centos5 as recent SFAs won't build on this platformb + # 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: + if retcod==0: + utils.header("build does provide SFA") + else: # move all steps containing 'sfa' from default_steps to other_steps - sfa_steps= [ step for step in TestPlc.default_steps if step.find('sfa')>=0 ] + 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() @@ -183,21 +184,56 @@ class TestPlc: 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) + 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): - return "vserver %s stop"%(self.vservername) + 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', @@ -226,11 +262,24 @@ 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']: @@ -334,13 +383,32 @@ class TestPlc: #################### display config def show (self): "show test configuration after localization" - self.display_pass (1) - self.display_pass (2) + 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 always_display_keys=['PLC_WWW_HOST','nodes','sites',] - def display_pass (self,passno): + 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: @@ -390,7 +458,7 @@ class TestPlc: 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'] @@ -429,7 +497,8 @@ 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']) + # 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']: @@ -441,46 +510,31 @@ class TestPlc: print '+\tqemu box %s'%node_spec['host_box'] print '+\thostname=%s'%node_spec['node_fields']['hostname'] - # write a timestamp in /vservers/<>/ - def timestamp_plc (self): + # 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()) - utils.system(self.test_ssh.actual_command("mkdir -p /vservers/%s"%self.vservername)) - return utils.system(self.test_ssh.actual_command("echo %d > /vservers/%s/timestamp"%(now,self.vservername)))==0 + # 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 - def local_pre (self): - "run site-dependant pre-test script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_pre(self) - - def local_post (self): - "run site-dependant post-test script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_post(self) - - def local_list (self): - "run site-dependant list script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_list(self) - - def local_rel (self): - "run site-dependant release script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_release(self) - - def local_rel_plc (self): - "run site-dependant release script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_release_plc(self) - - def local_rel_qemu (self): - "run site-dependant release script as defined in LocalTestResources" - from LocalTestResources import local_resources - return local_resources.step_release_qemu(self) - + # 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" - self.run_in_host("vserver --silent %s delete"%self.vservername) - return True + 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 # historically the build was being fetched by the tests @@ -511,7 +565,10 @@ 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: @@ -545,10 +602,9 @@ class TestPlc: pkgs_list.append ("slicerepo-%s"%nodefamily) pkgs_list.append ("myplc") pkgs_list.append ("noderepo-%s"%nodefamily) - pkgs_list.append ("bootstrapfs-%s-plain"%nodefamily) + pkgs_list.append ("nodeimage-%s-plain"%nodefamily) pkgs_string=" ".join(pkgs_list) - self.run_in_guest("yum -y install %s"%pkgs_string) - return self.run_in_guest("rpm -q %s"%pkgs_string)==0 + return self.yum_install (pkgs_list) ### def plc_configure(self): @@ -562,7 +618,7 @@ class TestPlc: 'PLC_MAIL_ENABLED', 'PLC_MAIL_SUPPORT_ADDRESS', 'PLC_DB_HOST', - 'PLC_DB_PASSWORD', +# 'PLC_DB_PASSWORD', # Above line was added for integrating SFA Testing 'PLC_API_HOST', 'PLC_WWW_HOST', @@ -622,10 +678,11 @@ class TestPlc: if not os.path.isdir(dir): os.mkdir(dir) vservername=self.vservername + vm_root=self.vm_root_in_host() overall=True 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 @@ -656,8 +713,11 @@ class TestPlc: 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 @@ -820,6 +880,15 @@ class TestPlc: overall=False return overall + # 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 = [] @@ -885,7 +954,7 @@ class TestPlc: return True def nodes_booted(self): - return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=20) + return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=28) def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=15): # compute timeout @@ -897,7 +966,7 @@ class TestPlc: local_key = "keys/%(vservername)s-debug.rsa"%locals() else: message="boot" - local_key = "keys/key1.rsa" + 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: @@ -934,44 +1003,28 @@ class TestPlc: 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=5) + return self.check_nodes_ssh(debug=True,timeout_minutes=10,silent_minutes=8) 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=15) + return self.check_nodes_ssh(debug=False,timeout_minutes=40,silent_minutes=38) @node_mapper - def qemu_local_init (self): - "all nodes : init a clean local directory for holding node-dep stuff like iso image..." - pass + def qemu_local_init (self): pass @node_mapper - def bootcd (self): - "all nodes: invoke GetBootMedium and store result locally" - pass + def bootcd (self): pass @node_mapper - def qemu_local_config (self): - "all nodes: compute qemu config qemu.conf and store it locally" - pass + def qemu_local_config (self): pass @node_mapper - def nodestate_reinstall (self): - "all nodes: mark PLCAPI boot_state as reinstall" - pass + def nodestate_reinstall (self): pass @node_mapper - def nodestate_safeboot (self): - "all nodes: mark PLCAPI boot_state as safeboot" - pass + def nodestate_safeboot (self): pass @node_mapper - def nodestate_boot (self): - "all nodes: mark PLCAPI boot_state as boot" - pass + def nodestate_boot (self): pass @node_mapper - def nodestate_show (self): - "all nodes: show PLCAPI boot_state" - pass + def nodestate_show (self): pass @node_mapper - def qemu_export (self): - "all nodes: push local node-dep directory on the qemu box" - pass + def qemu_export (self): pass ### check hooks : invoke scripts from hooks/{node,slice} def check_hooks_node (self): @@ -1027,45 +1080,65 @@ class TestPlc: ### manage slices def slices (self): "create slices with PLCAPI" - return self.do_slices() + return self.do_slices(action="add") def delete_slices (self): "delete slices with PLCAPI" - return self.do_slices("delete") + return self.do_slices(action="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 - def ssh_slice(self): - "tries to ssh-enter the slice with the user key, to ensure slice creation" - pass + def ssh_slice(self): pass + @slice_mapper + def ssh_slice_off (self): pass @node_mapper - def keys_clear_known_hosts (self): - "remove test nodes entries from the local known_hosts file" - 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 + @node_mapper - def qemu_start (self) : - "all nodes: start the qemu instance (also runs qemu-bridge-init start)" - pass + def qemu_start (self) : pass @node_mapper - def timestamp_qemu (self) : - "all nodes: start the qemu instance (also runs qemu-bridge-init start)" - pass + def timestamp_qemu (self) : pass def check_tcp (self): "check TCP connectivity between 2 slices (or in loopback if only one is defined)" @@ -1085,11 +1158,39 @@ class TestPlc: overall=False return overall + # painfully enough, we need to allow for some time as netflow might show up last + def check_sys_slice (self): + "all nodes: check that a system slice is alive" +# would probably make more sense to check for netflow, +# but that one is currently not working in the lxc distro +# return self.check_systemslice ('netflow') + return self.check_systemslice ('drl') + + # we have the slices up already here, so it should not take too long + def check_systemslice (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_systemslice (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) + remote="%s/%s"%(self.vm_root_in_host(),location) self.test_ssh.copy_abs("plcsh_stress_test.py",remote) command = location command += " -- --check" @@ -1101,31 +1202,76 @@ class TestPlc: # 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 - ### sfa_install_rpm - def sfa_install(self): - "yum install sfa, sfa-plc and sfa-client" - # ignore yum retcod - self.run_in_guest("yum -y install sfa sfa-client sfa-plc sfa-sfatables") - return self.run_in_guest("rpm -q sfa sfa-client sfa-plc sfa-sfatables")==0 + 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" - self.run_in_guest("sfa-nuke-plc.py")==0 - return True + 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'] - slicename='%s_%s'%(sfa_spec['login_base'],sfa_spec['slicename']) - try: self.apiserver.DeleteSlice(self.auth_root(),slicename) - except: print "Slice %s already absent from PLC db"%slicename + 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 - username="%s@%s"%(sfa_spec['regularuser'],sfa_spec['domain']) - try: self.apiserver.DeletePerson(self.auth_root(),username) - except: print "User %s already absent from PLC db"%username + 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 @@ -1191,20 +1337,25 @@ class TestPlc: fileconf=open(tmpname,'w') for var in [ 'SFA_REGISTRY_ROOT_AUTH', 'SFA_INTERFACE_HRN', -# 'SFA_REGISTRY_LEVEL1_AUTH', + 'SFA_REGISTRY_LEVEL1_AUTH', 'SFA_REGISTRY_HOST', 'SFA_AGGREGATE_HOST', 'SFA_SM_HOST', + 'SFA_PLC_URL', 'SFA_PLC_USER', 'SFA_PLC_PASSWORD', - 'SFA_PLC_DB_HOST', - 'SFA_PLC_DB_USER', - 'SFA_PLC_DB_PASSWORD', - 'SFA_PLC_URL', + 'SFA_DB_HOST', + 'SFA_DB_USER', + 'SFA_DB_PASSWORD', + 'SFA_DB_NAME', + 'SFA_API_LOGLEVEL', + 'SFA_GENERIC_FLAVOUR', + 'SFA_AGGREGATE_ENABLED', ]: - fileconf.write ('e %s\n%s\n'%(var,self.plc_spec['sfa'][var])) + 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 ['SFA_API_DEBUG']: + for var in []: val='false' if self.plc_spec['sfa'][var]: val='true' fileconf.write ('e %s\n%s\n'%(var,val)) @@ -1240,13 +1391,14 @@ class TestPlc: file(reg_fname,"w").write("%s\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,'/vservers/%s/etc/sfa/aggregates.xml'%self.vservername)==0 \ - and self.test_ssh.copy_abs(reg_fname,'/vservers/%s/etc/sfa/registries.xml'%self.vservername)==0 + 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('sfa-import-plc.py')==0 + 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)) @@ -1255,131 +1407,61 @@ class TestPlc: return self.run_in_guest('service sfa start')==0 def sfi_configure(self): - "Create /root/.sfi on the plc side for sfi client configuration" + "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'] - dir_name=self.confsubdir("dot-sfi",clean=True,dry_run=self.options.dry_run) - if self.options.dry_run: return True - file_name=dir_name + os.sep + sfa_spec['piuser'] + '.pkey' - fileconf=open(file_name,'w') - fileconf.write (self.plc_spec['keys'][0]['private']) - fileconf.close() - utils.header ("(Over)wrote %s"%file_name) - - file_name=dir_name + os.sep + 'sfi_config' - fileconf=open(file_name,'w') - SFI_AUTH="%s.%s"%(sfa_spec['SFA_REGISTRY_ROOT_AUTH'],sfa_spec['login_base']) - fileconf.write ("SFI_AUTH='%s'"%SFI_AUTH) - fileconf.write('\n') - SFI_USER=SFI_AUTH + '.' + sfa_spec['piuser'] - fileconf.write ("SFI_USER='%s'"%SFI_USER) - fileconf.write('\n') - SFI_REGISTRY='http://' + sfa_spec['SFA_PLC_DB_HOST'] + ':12345/' - fileconf.write ("SFI_REGISTRY='%s'"%SFI_REGISTRY) - fileconf.write('\n') - SFI_SM='http://' + sfa_spec['SFA_PLC_DB_HOST'] + ':12347/' - fileconf.write ("SFI_SM='%s'"%SFI_SM) - fileconf.write('\n') - fileconf.close() - utils.header ("(Over)wrote %s"%file_name) - - file_name=dir_name + os.sep + 'person.xml' - fileconf=open(file_name,'w') - for record in sfa_spec['sfa_person_xml']: - person_record=record - fileconf.write(person_record) - fileconf.write('\n') - fileconf.close() - utils.header ("(Over)wrote %s"%file_name) - - file_name=dir_name + os.sep + 'slice.xml' - fileconf=open(file_name,'w') - for record in sfa_spec['sfa_slice_xml']: - slice_record=record - #slice_record=sfa_spec['sfa_slice_xml'] - fileconf.write(slice_record) - fileconf.write('\n') - utils.header ("(Over)wrote %s"%file_name) - fileconf.close() - - file_name=dir_name + os.sep + 'slice.rspec' - fileconf=open(file_name,'w') - slice_rspec='' - for (key, value) in sfa_spec['sfa_slice_rspec'].items(): - slice_rspec +=value - fileconf.write(slice_rspec) - fileconf.write('\n') - fileconf.close() - utils.header ("(Over)wrote %s"%file_name) - - # push to the remote root's .sfi - location = "root/.sfi" - remote="/vservers/%s/%s"%(self.vservername,location) - self.test_ssh.copy_abs(dir_name, remote, recursive=True) + # 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") + "clean up /root/sfi on the plc side" + self.run_in_guest("rm -rf /root/sfi") return True - def sfa_add_user(self): - "run sfi.py add using person.xml" - return TestUserSfa(self).add_user() - - def sfa_update_user(self): - "run sfi.py update using person.xml" - return TestUserSfa(self).update_user() - - @slice_sfa_mapper - def sfa_add_slice(self): - "run sfi.py add (on Registry) from slice.xml" - pass - - @slice_sfa_mapper - def sfa_discover(self): - "discover resources into resouces_in.rspec" - pass - - @slice_sfa_mapper - def sfa_create_slice(self): - "run sfi.py create (on SM) - 1st time" - pass - - @slice_sfa_mapper - def sfa_check_slice_plc(self): - "check sfa_create_slice at the plcs - all local nodes should be in slice" - pass - - @slice_sfa_mapper - def sfa_update_slice(self): - "run sfi.py create (on SM) on existing object" - pass - - def sfa_view(self): - "run sfi.py list and sfi.py show (both on Registry) and sfi.py slices and sfi.py resources (both on SM)" - sfa_spec=self.plc_spec['sfa'] - auth=sfa_spec['SFA_REGISTRY_ROOT_AUTH'] - return \ - self.run_in_guest("sfi.py -d /root/.sfi/ list %s.%s"%(auth,sfa_spec['login_base']))==0 and \ - self.run_in_guest("sfi.py -d /root/.sfi/ show %s.%s"%(auth,sfa_spec['login_base']))==0 and \ - self.run_in_guest("sfi.py -d /root/.sfi/ slices")==0 and \ - self.run_in_guest("sfi.py -d /root/.sfi/ resources -o resources")==0 - - @slice_sfa_mapper - def ssh_slice_sfa(self): - "tries to ssh-enter the SFA slice" - pass - - def sfa_delete_user(self): - "run sfi.py delete (on SM) for user" - test_user_sfa=TestUserSfa(self) - return test_user_sfa.delete_user() - - @slice_sfa_mapper - def sfa_delete_slice(self): - "run sfi.py delete (on SM), sfi.py remove (on Registry) to clean slices" - pass + @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_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" @@ -1390,7 +1472,7 @@ class TestPlc: "creates random entries in the PLCAPI" # install the stress-test in the plc image location = "/usr/share/plc_api/plcsh_stress_test.py" - remote="/vservers/%s/%s"%(self.vservername,location) + 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" @@ -1404,6 +1486,7 @@ class TestPlc: "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./* # (1.b) get the plc's /var/lib/pgsql/data/pg_log/ -> logs/myplc.pgsql-log./* + # (1.c) get the plc's /root/sfi -> logs/sfi./ # (2) get all the nodes qemu log and store it as logs/node.qemu..log # (3) get the nodes /var/log and store is as logs/node.var-log./* # (4) as far as possible get the slice's /var/log as logs/sliver.var-log./* @@ -1413,6 +1496,9 @@ 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']: @@ -1450,12 +1536,18 @@ 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="keys/key1.rsa") + 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()) @@ -1496,6 +1588,9 @@ class TestPlc: 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