X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=3fbcb09d6f6a2e06e106e2dd25959fb12246f646;hb=51fddd4990c9a122b26e5a2f2603f6aa5b0178bb;hp=4e9a99b8ebebd1810771dff43a45027fc3675673;hpb=1558599ffdfa4fa2dd53bdc68beaf56c2d9ad4aa;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index 4e9a99b..3fbcb09 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -17,6 +17,8 @@ from TestSliver import TestSliver from TestBox import TestBox from TestSsh import TestSsh from TestApiserver import TestApiserver +from TestSliceSfa import TestSliceSfa +from TestUserSfa import TestUserSfa # step methods must take (self) and return a boolean (options is a member of the class) @@ -44,6 +46,8 @@ def node_mapper (method): test_node = TestNode (self,test_site,node_spec) if not node_method(test_node): overall=False return overall + # restore the doc text + actual.__doc__=method.__doc__ return actual def slice_mapper_options (method): @@ -56,29 +60,55 @@ 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__=method.__doc__ + return actual + +def slice_mapper_options_sfa (method): + def actual(self): + test_plc=self + overall=True + slice_method = TestSliceSfa.__dict__[method.__name__] + for slice_spec in self.plc_spec['sfa']['slices_sfa']: + site_spec = self.locate_site (slice_spec['sitename']) + test_site = TestSite(self,site_spec) + test_slice=TestSliceSfa(test_plc,test_site,slice_spec) + if not slice_method(test_slice,self.options): overall=False + return overall + # restore the doc text + actual.__doc__=method.__doc__ return actual SEP='' class TestPlc: - default_steps = ['uninstall','install','install_rpm', - 'configure', 'start', SEP, - 'store_keys', 'clear_known_hosts', 'initscripts', SEP, - 'sites', 'nodes', 'slices', 'nodegroups', SEP, - 'init_node','bootcd', 'configure_qemu', 'export_qemu', - 'kill_all_qemus', 'reinstall_node','start_node', SEP, - 'nodes_booted', 'nodes_ssh', 'check_slice', - 'check_initscripts', 'check_tcp', 'plcsh_stress_test', SEP, - 'force_gather_logs', 'force_kill_qemus', 'force_record_tracker','force_free_tracker' ] - other_steps = [ 'stop_all_vservers','fresh_install', 'cache_rpm', 'stop', 'vs_start', SEP, - 'clean_initscripts', 'clean_all_sites', - 'clean_sites', 'clean_nodes', - 'clean_slices', 'clean_keys', SEP, - 'show_boxes', 'list_all_qemus', 'list_qemus', SEP, - 'db_dump' , 'db_restore', ' cleanup_tracker', - 'standby_1 through 20' - ] + default_steps = [ + 'display', 'local_pre', SEP, + 'delete','create','install', 'configure', 'start', SEP, + 'fetch_keys', 'store_keys', 'clear_known_hosts', SEP, + 'initscripts', 'sites', 'nodes', 'slices', 'nodegroups', SEP, + 'reinstall_node', 'init_node','bootcd', 'configure_qemu', 'export_qemu', + 'kill_all_qemus', '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, + 'install_sfa', 'configure_sfa', 'import_sfa', 'start_sfa', SEP, + 'setup_sfa', 'add_sfa', 'update_sfa', SEP, + 'view_sfa', 'check_slice_sfa', 'delete_sfa', 'stop_sfa', SEP, + 'check_tcp', 'check_hooks', SEP, + 'force_gather_logs', 'force_local_post', + ] + other_steps = [ + 'fresh_install', 'stop', 'vs_start', SEP, + 'clean_initscripts', 'clean_nodegroups','clean_all_sites', SEP, + 'clean_sites', 'clean_nodes', 'clean_slices', 'clean_keys', SEP, + 'populate' , SEP, + 'show_boxes', 'list_all_qemus', 'list_qemus', 'kill_qemus', SEP, + 'db_dump' , 'db_restore', SEP, + 'local_list','local_cleanup',SEP, + 'standby_1 through 20', + ] @staticmethod def printable_steps (list): @@ -179,6 +209,37 @@ class TestPlc: return slice raise Exception,"Cannot locate slice %s"%slicename + def all_sliver_objs (self): + result=[] + for slice_spec in self.plc_spec['slices']: + slicename = slice_spec['slice_fields']['name'] + for nodename in slice_spec['nodenames']: + result.append(self.locate_sliver_obj (nodename,slicename)) + return result + + def locate_sliver_obj (self,nodename,slicename): + (site,node) = self.locate_node(nodename) + slice = self.locate_slice (slicename) + # build objects + test_site = TestSite (self, site) + test_node = TestNode (self, test_site,node) + # xxx the slice site is assumed to be the node site - mhh - probably harmless + test_slice = TestSlice (self, test_site, slice) + return TestSliver (self, test_node, test_slice) + + def locate_first_node(self): + nodename=self.plc_spec['slices'][0]['nodenames'][0] + (site,node) = self.locate_node(nodename) + test_site = TestSite (self, site) + test_node = TestNode (self, test_site,node) + return test_node + + def locate_first_sliver (self): + slice_spec=self.plc_spec['slices'][0] + slicename=slice_spec['slice_fields']['name'] + nodename=slice_spec['nodenames'][0] + return self.locate_sliver_obj(nodename,slicename) + # all different hostboxes used in this plc def gather_hostBoxes(self): # maps on sites and nodes, return [ (host_box,test_node) ] @@ -206,6 +267,7 @@ class TestPlc: # make this a valid step def kill_all_qemus(self): + "all qemu boxes: kill all running qemus (even of former runs)" # 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 @@ -236,58 +298,140 @@ class TestPlc: node.kill_qemu() return True + #################### display config + def display (self): + "show test configuration after localization" + self.display_pass (1) + self.display_pass (2) + return True - ### utility methods for handling the pool of IP addresses allocated to plcs - # Logic - # (*) running plcs are recorded in the file named ~/running-test-plcs - # (*) this file contains a line for each running plc, older first - # (*) each line contains the vserver name + the hostname of the (vserver) testbox where it sits - # (*) the free_tracker method performs a vserver stop on the oldest entry - # (*) the record_tracker method adds an entry at the bottom of the file - # (*) the cleanup_tracker method stops all known vservers and removes the tracker file - - TRACKER_FILE="~/running-test-plcs" + # entry point + def display_pass (self,passno): + for (key,val) in self.plc_spec.iteritems(): + if passno == 2: + if key == 'sites': + for site in val: + self.display_site_spec(site) + for node in site['nodes']: + self.display_node_spec(node) + elif key=='initscripts': + for initscript in val: + self.display_initscript_spec (initscript) + elif key=='slices': + for slice in val: + self.display_slice_spec (slice) + elif key=='keys': + for key in val: + self.display_key_spec (key) + elif passno == 1: + if key not in ['sites','initscripts','slices','keys']: + print '+ ',key,':',val + + def display_site_spec (self,site): + print '+ ======== site',site['site_fields']['name'] + for (k,v) in site.iteritems(): + if k=='nodes': + if v: + print '+ ','nodes : ', + for node in v: + print node['node_fields']['hostname'],'', + print '' + elif k=='users': + if v: + print '+ users : ', + for user in v: + print user['name'],'', + print '' + elif k == 'site_fields': + print '+ login_base',':',v['login_base'] + elif k == 'address_fields': + pass + else: + print '+ ',k, + PrettyPrinter(indent=8,depth=2).pprint(v) + + def display_initscript_spec (self,initscript): + print '+ ======== initscript',initscript['initscript_fields']['name'] + + def display_key_spec (self,key): + print '+ ======== key',key['name'] + + def display_slice_spec (self,slice): + print '+ ======== slice',slice['slice_fields']['name'] + for (k,v) in slice.iteritems(): + if k=='nodenames': + if v: + print '+ nodes : ', + for nodename in v: + print nodename,'', + print '' + elif k=='usernames': + if v: + print '+ users : ', + for username in v: + print username,'', + print '' + elif k=='slice_fields': + print '+ fields',':', + print 'max_nodes=',v['max_nodes'], + print '' + else: + print '+ ',k,v - def record_tracker (self): - command="echo %s %s >> %s"%(self.vservername,self.test_ssh.hostname,TestPlc.TRACKER_FILE) - (code,output) = utils.output_of (self.test_ssh.actual_command(command)) - if code != 0: - print "WARNING : COULD NOT record_tracker %s as a running plc on %s"%(self.vservername,self.test_ssh.hostname) - return False - print "Recorded %s in running plcs on host %s"%(self.vservername,self.test_ssh.hostname) - return True + def display_node_spec (self,node): + print "+ node",node['name'],"host_box=",node['host_box'], + print "hostname=",node['node_fields']['hostname'], + print "ip=",node['interface_fields']['ip'] + - def free_tracker (self): - command="head -1 %s"%TestPlc.TRACKER_FILE - (code,line) = utils.output_of(self.test_ssh.actual_command(command)) - if code != 0: - print "No entry found in %s on %s"%(TestPlc.TRACKER_FILE,self.test_ssh.hostname) - return False - try: - [vserver_to_stop,hostname] = line.split() - except: - print "WARNING: free_tracker: Could not parse %s - skipped"%TestPlc.TRACKER_FILE - return False - stop_command = "vserver --silent %s stop"%vserver_to_stop - utils.system(self.test_ssh.actual_command(stop_command)) - x=TestPlc.TRACKER_FILE - flush_command = "tail --lines=+2 %s > %s.tmp ; mv %s.tmp %s"%(x,x,x,x) - utils.system(self.test_ssh.actual_command(flush_command)) + # another entry point for just showing the boxes involved + def display_mapping (self): + TestPlc.display_mapping_plc(self.plc_spec) return True - # this should/could stop only the ones in TRACKER_FILE if that turns out to be reliable - def cleanup_tracker (self): - stop_all = "cd /vservers ; for i in * ; do vserver --silent $i stop ; done" - utils.system(self.test_ssh.actual_command(stop_all)) - clean_tracker = "rm -f %s"%TestPlc.TRACKER_FILE - utils.system(self.test_ssh.actual_command(clean_tracker)) + @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']) + for site_spec in plc_spec['sites']: + for node_spec in site_spec['nodes']: + TestPlc.display_mapping_node(node_spec) - def uninstall(self): + @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'] + + 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_cleanup (self): + "run site-dependant cleanup script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_cleanup(self) + + def delete(self): + "vserver delete the test myplc" self.run_in_host("vserver --silent %s delete"%self.vservername) return True ### install - def install(self): + def create (self): + "vserver creation (no install done)" if self.is_local(): # a full path for the local calls build_dir=os.path.dirname(sys.argv[0]) @@ -302,14 +446,15 @@ class TestPlc: if self.run_in_host(build_checkout) != 0: return False # the repo url is taken from arch-rpms-url - # with the last step (i386.) removed + # with the last step (i386) removed repo_url = self.options.arch_rpms_url for level in [ 'arch' ]: repo_url = os.path.dirname(repo_url) - if self.options.arch == "i386": - personality_option="-p linux32" - else: - personality_option="-p linux64" + # pass the vbuild-nightly options to vtest-init-vserver + test_env_options="" + 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" vserver_name = self.vservername vserver_options="--netdev eth0 --interface %s"%self.vserverip @@ -318,15 +463,34 @@ class TestPlc: vserver_options += " --hostname %s"%vserver_hostname except: pass - create_vserver="%(build_dir)s/%(script)s %(personality_option)s %(vserver_name)s %(repo_url)s -- %(vserver_options)s"%locals() + 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): - return self.run_in_guest("yum -y install myplc-native")==0 + def install(self): + "yum install myplc, noderepo, and the plain bootstrapfs" + + # workaround for getting pgsql5.2 on centos5 + if self.options.fcdistro == "centos5": + self.run_in_guest("rpm -Uvh http://yum.pgsqlrpms.org/8.2/pgdg-centos-8.2-4.noarch.rpm") + self.run_in_guest("rpm -Uvh http://download.fedora.redhat.com/pub/epel/5/i386/epel-release-5-3.noarch.rpm") + + if self.options.personality == "linux32": + arch = "i386" + elif self.options.personality == "linux64": + arch = "x86_64" + else: + raise Exception, "Unsupported personality %r"%self.options.personality + + nodefamily="%s-%s-%s"%(self.options.pldistro,self.options.fcdistro,arch) + return \ + self.run_in_guest("yum -y install myplc")==0 and \ + self.run_in_guest("yum -y install noderepo-%s"%nodefamily)==0 and \ + self.run_in_guest("yum -y install bootstrapfs-%s-plain"%nodefamily)==0 ### def configure(self): + "run plc-config-tty" tmpname='%s.plc-config-tty'%(self.name()) fileconf=open(tmpname,'w') for var in [ 'PLC_NAME', @@ -335,6 +499,8 @@ class TestPlc: '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', @@ -350,10 +516,12 @@ class TestPlc: return True def start(self): + "service plc start" self.run_in_guest('service plc start') return True def stop(self): + "service plc stop" self.run_in_guest('service plc stop') return True @@ -361,8 +529,9 @@ class TestPlc: self.start_guest() return True - # could use a TestKey class + # stores the keys from the config for further use def store_keys(self): + "stores test users ssh keys in keys/" for key_spec in self.plc_spec['keys']: TestKey(self,key_spec).store_key() return True @@ -370,10 +539,28 @@ class TestPlc: def clean_keys(self): utils.system("rm -rf %s/keys/"%os.path(sys.argv[0])) + # 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): + "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 + overall=True + prefix = 'debug_ssh_key' + for ext in [ 'pub', 'rsa' ] : + src="/vservers/%(vservername)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): + "delete sites with PLCAPI" return self.do_sites(action="delete") def do_sites (self,action="add"): @@ -399,8 +586,10 @@ class TestPlc: self.apiserver.DeleteSite(self.auth_root(),site_id) def nodes (self): + "create nodes with PLCAPI" return self.do_nodes() def clean_nodes (self): + "delete nodes with PLCAPI" return self.do_nodes(action="delete") def do_nodes (self,action="add"): @@ -420,9 +609,15 @@ class TestPlc: test_node.create_node () return True - # create nodegroups if needed, and populate - # no need for a clean_nodegroups if we are careful enough def nodegroups (self): + "create nodegroups with PLCAPI" + return self.do_nodegroups("add") + def clean_nodegroups (self): + "delete nodegroups with PLCAPI" + return self.do_nodegroups("delete") + + # create nodegroups if needed, and populate + def do_nodegroups (self, action="add"): # 1st pass to scan contents groups_dict = {} for site_spec in self.plc_spec['sites']: @@ -438,60 +633,73 @@ class TestPlc: groups_dict[nodegroupname]=[] groups_dict[nodegroupname].append(test_node.name()) auth=self.auth_root() + overall = True for (nodegroupname,group_nodes) in groups_dict.iteritems(): - print 'nodegroups:','dealing with nodegroup',nodegroupname,'on nodes',group_nodes - # first, check if the nodetagtype is here - tag_types = self.apiserver.GetTagTypes(auth,{'tagname':nodegroupname}) - if tag_types: - tag_type_id = tag_types[0]['node_tag_type_id'] - else: - tag_type_id = self.apiserver.AddNodeTagType(auth, + if action == "add": + print 'nodegroups:','dealing with nodegroup',nodegroupname,'on nodes',group_nodes + # first, check if the nodetagtype is here + tag_types = self.apiserver.GetTagTypes(auth,{'tagname':nodegroupname}) + if tag_types: + tag_type_id = tag_types[0]['tag_type_id'] + else: + tag_type_id = self.apiserver.AddTagType(auth, {'tagname':nodegroupname, 'description': 'for nodegroup %s'%nodegroupname, 'category':'test', 'min_role_id':10}) - # create nodegroup - nodegroups = self.apiserver.GetNodeGroups (auth, {'groupname':nodegroupname}) - if not nodegroups: - self.apiserver.AddNodeGroup(auth, nodegroupname, tag_type_id, 'yes') - # set node tag on all nodes, value='yes' - overall = True - for nodename in group_nodes: - try: - self.apiserver.AddNodeTag(auth, nodename, nodegroupname, "yes") - except: - print 'node',nodename,'seems to already have tag',nodegroupname - # check anyway + print 'located tag (type)',nodegroupname,'as',tag_type_id + # create nodegroup + nodegroups = self.apiserver.GetNodeGroups (auth, {'groupname':nodegroupname}) + if not nodegroups: + self.apiserver.AddNodeGroup(auth, nodegroupname, tag_type_id, 'yes') + print 'created nodegroup',nodegroupname,'from tagname',nodegroupname,'and value','yes' + # set node tag on all nodes, value='yes' + for nodename in group_nodes: + try: + self.apiserver.AddNodeTag(auth, nodename, nodegroupname, "yes") + except: + traceback.print_exc() + print 'node',nodename,'seems to already have tag',nodegroupname + # check anyway + try: + expect_yes = self.apiserver.GetNodeTags(auth, + {'hostname':nodename, + 'tagname':nodegroupname}, + ['value'])[0]['value'] + if expect_yes != "yes": + print 'Mismatch node tag on node',nodename,'got',expect_yes + overall=False + except: + if not self.options.dry_run: + print 'Cannot find tag',nodegroupname,'on node',nodename + overall = False + else: try: - expect_yes = self.apiserver.GetNodeTags( - auth, - {'hostname':nodename, - 'tagname':nodegroupname}, - ['tagvalue'])[0]['tagvalue'] - if expect_yes != "yes": - print 'Mismatch node tag on node',nodename,'got',expect_yes - overall=False + print 'cleaning nodegroup',nodegroupname + self.apiserver.DeleteNodeGroup(auth,nodegroupname) except: - if not self.options.dry_run: - print 'Cannot find tag',nodegroupname,'on node',nodename - overall = False + traceback.print_exc() + overall=False return overall - def all_hostnames (self) : - hostnames = [] + # 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'] \ + node_infos += [ (node_spec['node_fields']['hostname'],node_spec['host_box']) \ for node_spec in site_spec['nodes'] ] - return hostnames + return node_infos + + def all_nodenames (self): return [ x[0] for x in self.all_node_infos() ] - # gracetime : during the first minutes nothing gets printed - def do_nodes_booted (self, minutes, gracetime,period=15): + # silent_minutes : during the first minutes nothing gets printed + def nodes_check_boot_state (self, target_boot_state, timeout_minutes, silent_minutes,period=15): if self.options.dry_run: print 'dry_run' return True # compute timeout - timeout = datetime.datetime.now()+datetime.timedelta(minutes=minutes) - graceout = datetime.datetime.now()+datetime.timedelta(minutes=gracetime) + timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes) + graceout = datetime.datetime.now()+datetime.timedelta(minutes=silent_minutes) # the nodes that haven't checked yet - start with a full list and shrink over time tocheck = self.all_hostnames() utils.header("checking nodes %r"%tocheck) @@ -504,21 +712,21 @@ class TestPlc: for array in tocheck_status: hostname=array['hostname'] boot_state=array['boot_state'] - if boot_state == 'boot': - utils.header ("%s has reached the 'boot' state"%hostname) + if boot_state == target_boot_state: + utils.header ("%s has reached the %s state"%(hostname,target_boot_state)) else: # if it's a real node, never mind (site_spec,node_spec)=self.locate_hostname(hostname) if TestNode.is_real_model(node_spec['node_fields']['model']): utils.header("WARNING - Real node %s in %s - ignored"%(hostname,boot_state)) # let's cheat - boot_state = 'boot' + boot_state = target_boot_state elif datetime.datetime.now() > graceout: utils.header ("%s still in '%s' state"%(hostname,boot_state)) graceout=datetime.datetime.now()+datetime.timedelta(1) status[hostname] = boot_state # refresh tocheck - tocheck = [ hostname for (hostname,boot_state) in status.iteritems() if boot_state != 'boot' ] + tocheck = [ hostname for (hostname,boot_state) in status.iteritems() if boot_state != target_boot_state ] if not tocheck: return True if datetime.datetime.now() > timeout: @@ -531,57 +739,93 @@ class TestPlc: return True def nodes_booted(self): - return self.do_nodes_booted(minutes=20,gracetime=15) + return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=20) - def do_nodes_ssh(self,minutes,gracetime,period=15): + def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=15): # compute timeout - timeout = datetime.datetime.now()+datetime.timedelta(minutes=minutes) - graceout = datetime.datetime.now()+datetime.timedelta(minutes=gracetime) - tocheck = self.all_hostnames() -# self.scan_publicKeys(tocheck) - utils.header("checking Connectivity on nodes %r"%tocheck) - while tocheck: - for hostname in tocheck: - # try to ssh in nodes - node_test_ssh = TestSsh (hostname,key="/etc/planetlab/root_ssh_key.rsa") - success=self.run_in_guest(node_test_ssh.actual_command("hostname"))==0 - if success: - utils.header('The node %s is sshable -->'%hostname) - # refresh tocheck - tocheck.remove(hostname) + timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes) + graceout = datetime.datetime.now()+datetime.timedelta(minutes=silent_minutes) + vservername=self.vservername + if debug: + message="debug" + local_key = "keys/%(vservername)s-debug.rsa"%locals() + else: + message="boot" + local_key = "keys/key1.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 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 + success = utils.system ( command, silent=datetime.datetime.now() < graceout) + if success==0: + utils.header('Successfully entered root@%s (%s)'%(hostname,message)) + # 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) - elif datetime.datetime.now() > graceout: - utils.header("Could not ssh-enter root context on %s"%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(self): - return self.do_nodes_ssh(minutes=6,gracetime=4) + def nodes_ssh_debug(self): + "Tries to ssh into nodes in debug mode with the debug ssh key" + return self.check_nodes_ssh(debug=True,timeout_minutes=30,silent_minutes=5) + + def nodes_ssh_boot(self): + "Tries to ssh into nodes in production mode with the root ssh key" + return self.check_nodes_ssh(debug=False,timeout_minutes=30,silent_minutes=15) @node_mapper - def init_node (self): pass + def init_node (self): + "all nodes : init a clean local directory for holding node-dep stuff like iso image..." + pass @node_mapper - def bootcd (self): pass + def bootcd (self): + "all nodes: invoke GetBootMedium and store result locally" + pass @node_mapper - def configure_qemu (self): pass + def configure_qemu (self): + "all nodes: compute qemu config qemu.conf and store it locally" + pass @node_mapper - def reinstall_node (self): pass + def reinstall_node (self): + "all nodes: mark PLCAPI boot_state as reinstall" + pass @node_mapper - def export_qemu (self): pass + def export_qemu (self): + "all nodes: push local node-dep directory on the qemu box" + pass + ### 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_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']: @@ -600,15 +844,18 @@ class TestPlc: 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): + "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'])) @@ -619,10 +866,13 @@ class TestPlc: print 'deletion went wrong - probably did not exist' return True + ### manage slices def slices (self): + "create slices with PLCAPI" return self.do_slices() def clean_slices (self): + "delete slices with PLCAPI" return self.do_slices("delete") def do_slices (self, action="add"): @@ -640,33 +890,22 @@ class TestPlc: return True @slice_mapper_options - def check_slice(self): pass + def check_slice(self): + "tries to ssh-enter the slice with the user key, to ensure slice creation" + pass @node_mapper - def clear_known_hosts (self): pass + def clear_known_hosts (self): + "remove test nodes entries from the local known_hosts file" + pass @node_mapper - def start_node (self) : pass - - def all_sliver_objs (self): - result=[] - for slice_spec in self.plc_spec['slices']: - slicename = slice_spec['slice_fields']['name'] - for nodename in slice_spec['nodenames']: - result.append(self.locate_sliver_obj (nodename,slicename)) - return result - - def locate_sliver_obj (self,nodename,slicename): - (site,node) = self.locate_node(nodename) - slice = self.locate_slice (slicename) - # build objects - test_site = TestSite (self, site) - test_node = TestNode (self, test_site,node) - # xxx the slice site is assumed to be the node site - mhh - probably harmless - test_slice = TestSlice (self, test_site, slice) - return TestSliver (self, test_node, test_slice) + def start_node (self) : + "all nodes: start the qemu instance (also runs qemu-bridge-init start)" + pass def check_tcp (self): + "check TCP connectivity between 2 slices (or in loopback if only one is defined)" specs = self.plc_spec['tcp_test'] overall=True for spec in specs: @@ -684,24 +923,214 @@ class TestPlc: return overall 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" + 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) + self.test_ssh.copy_abs("plcsh_stress_test.py",remote) command = location command += " -- --check" - if self.options.small_test: + 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 + + ### install_sfa_rpm + def install_sfa(self): + "yum install sfa, sfa-plc and sfa-client" + 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 sfa sfa-client sfa-plc sfa-sfatables")==0 + + ### + def configure_sfa(self): + "run sfa-config-tty" + tmpname='%s.sfa-config-tty'%(self.name()) + fileconf=open(tmpname,'w') + for var in [ 'SFA_REGISTRY_ROOT_AUTH', + 'SFA_REGISTRY_LEVEL1_AUTH', + 'SFA_REGISTRY_HOST', + 'SFA_AGGREGATE_HOST', + 'SFA_SM_HOST', + 'SFA_PLC_USER', + 'SFA_PLC_PASSWORD', + 'SFA_PLC_DB_HOST', + 'SFA_PLC_DB_USER', + 'SFA_PLC_DB_PASSWORD', + 'SFA_PLC_URL']: + fileconf.write ('e %s\n%s\n'%(var,self.plc_spec['sfa'][var])) + 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') + utils.system('rm %s'%tmpname) + return True + + def import_sfa(self): + "sfa-import-plc" + auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'] + return self.run_in_guest('sfa-import-plc.py')==0 +# not needed anymore +# self.run_in_guest('cp /etc/sfa/authorities/%s/%s.pkey /etc/sfa/authorities/server.key'%(auth,auth)) + + def start_sfa(self): + "service sfa start" + return self.run_in_guest('service sfa start')==0 + + def setup_sfa(self): + "sfi client configuration" + dir_name=".sfi" + if os.path.exists(dir_name): + utils.system('rm -rf %s'%dir_name) + utils.system('mkdir %s'%dir_name) + file_name=dir_name + os.sep + 'fake-pi1.pkey' + fileconf=open(file_name,'w') + fileconf.write (self.plc_spec['keys'][0]['private']) + fileconf.close() + + file_name=dir_name + os.sep + 'sfi_config' + fileconf=open(file_name,'w') + SFI_AUTH=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH']+".main" + fileconf.write ("SFI_AUTH='%s'"%SFI_AUTH) + fileconf.write('\n') + SFI_USER=SFI_AUTH+'.fake-pi1' + fileconf.write ("SFI_USER='%s'"%SFI_USER) + fileconf.write('\n') + SFI_REGISTRY='http://' + self.plc_spec['sfa']['SFA_PLC_DB_HOST'] + ':12345/' + fileconf.write ("SFI_REGISTRY='%s'"%SFI_REGISTRY) + fileconf.write('\n') + SFI_SM='http://' + self.plc_spec['sfa']['SFA_PLC_DB_HOST'] + ':12347/' + fileconf.write ("SFI_SM='%s'"%SFI_SM) + fileconf.write('\n') + fileconf.close() + + file_name=dir_name + os.sep + 'person.xml' + fileconf=open(file_name,'w') + for record in self.plc_spec['sfa']['sfa_person_xml']: + person_record=record + fileconf.write(person_record) + fileconf.write('\n') + fileconf.close() + + file_name=dir_name + os.sep + 'slice.xml' + fileconf=open(file_name,'w') + for record in self.plc_spec['sfa']['sfa_slice_xml']: + slice_record=record + #slice_record=self.plc_spec['sfa']['sfa_slice_xml'] + fileconf.write(slice_record) + fileconf.write('\n') + fileconf.close() + + file_name=dir_name + os.sep + 'slice.rspec' + fileconf=open(file_name,'w') + slice_rspec='' + for (key, value) in self.plc_spec['sfa']['sfa_slice_rspec'].items(): + slice_rspec +=value + fileconf.write(slice_rspec) + fileconf.write('\n') + fileconf.close() + location = "root/" + remote="/vservers/%s/%s"%(self.vservername,location) + self.test_ssh.copy_abs(dir_name, remote, recursive=True) + + #utils.system('cat %s'%tmpname) + utils.system('rm -rf %s'%dir_name) + return True + + def add_sfa(self): + "run sfi.py add (on Registry) and sfi.py create (on SM) to form new objects" + test_plc=self + test_user_sfa=TestUserSfa(test_plc,self.plc_spec['sfa']) + success=test_user_sfa.add_user() + + for slice_spec in self.plc_spec['sfa']['slices_sfa']: + site_spec = self.locate_site (slice_spec['sitename']) + test_site = TestSite(self,site_spec) + test_slice_sfa=TestSliceSfa(test_plc,test_site,slice_spec) + success1=test_slice_sfa.add_slice() + success2=test_slice_sfa.create_slice() + return success and success1 and success2 + + def update_sfa(self): + "run sfi.py update (on Registry) and sfi.py create (on SM) on existing objects" + test_plc=self + test_user_sfa=TestUserSfa(test_plc,self.plc_spec['sfa']) + success1=test_user_sfa.update_user() + + for slice_spec in self.plc_spec['sfa']['slices_sfa']: + site_spec = self.locate_site (slice_spec['sitename']) + test_site = TestSite(self,site_spec) + test_slice_sfa=TestSliceSfa(test_plc,test_site,slice_spec) + success2=test_slice_sfa.update_slice() + return success1 and success2 + + def view_sfa(self): + "run sfi.py list and sfi.py show (both on Registry) and sfi.py slices and sfi.py resources (both on SM)" + auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'] + return \ + self.run_in_guest("sfi.py -d /root/.sfi/ list %s.main"%auth)==0 and \ + self.run_in_guest("sfi.py -d /root/.sfi/ show %s.main"%auth)==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_mapper_options_sfa + def check_slice_sfa(self): + "tries to ssh-enter the SFA slice" + pass + + def delete_sfa(self): + "run sfi.py delete (on SM), sfi.py remove (on Registry)" + test_plc=self + test_user_sfa=TestUserSfa(test_plc,self.plc_spec['sfa']) + success1=test_user_sfa.delete_user() + for slice_spec in self.plc_spec['sfa']['slices_sfa']: + site_spec = self.locate_site (slice_spec['sitename']) + test_site = TestSite(self,site_spec) + test_slice_sfa=TestSliceSfa(test_plc,test_site,slice_spec) + success2=test_slice_sfa.delete_slice() + + return success1 and success2 + + def stop_sfa(self): + "service sfa stop" + return self.run_in_guest('service sfa stop')==0 + + 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="/vservers/%s/%s"%(self.vservername,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): - # (1) get the plc's /var/log and store it locally in logs/myplc.var-log./* + "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./* # (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./* - # (1) + # (1.a) print "-------------------- TestPlc.gather_logs : PLC's /var/log" self.gather_var_logs () + # (1.b) + print "-------------------- TestPlc.gather_logs : PLC's /var/lib/psql/data/pg_log/" + self.gather_pgsql_logs () # (2) print "-------------------- TestPlc.gather_logs : nodes's QEMU logs" for site_spec in self.plc_spec['sites']: @@ -726,9 +1155,17 @@ class TestPlc: return True def gather_var_logs (self): + utils.system("mkdir -p logs/myplc.var-log.%s"%self.name()) to_plc = self.actual_command_in_guest("tar -C /var/log/ -cf - .") command = to_plc + "| tar -C logs/myplc.var-log.%s -xf -"%self.name() - utils.system("mkdir -p logs/myplc.var-log.%s"%self.name()) + utils.system(command) + command = "chmod a+r,a+x logs/myplc.var-log.%s/httpd"%self.name() + utils.system(command) + + def gather_pgsql_logs (self): + utils.system("mkdir -p logs/myplc.pgsql-log.%s"%self.name()) + to_plc = self.actual_command_in_guest("tar -C /var/lib/pgsql/data/pg_log/ -cf - .") + command = to_plc + "| tar -C logs/myplc.pgsql-log.%s -xf -"%self.name() utils.system(command) def gather_nodes_var_logs (self): @@ -736,9 +1173,9 @@ class TestPlc: 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/key1.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) @@ -815,4 +1252,3 @@ class TestPlc: def standby_19(): pass @standby_generic def standby_20(): pass -