X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=3e7970d0552b5c18a900bdfac6827e4dad691722;hb=acc29368268d995c2965a917409dba89d2373397;hp=d4172e1ae1606b5265c85c0e9282dc19657c50c2;hpb=82e22465b2dc1e78b9c6acdee2b6004424b3ffdd;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index d4172e1..3e7970d 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -1,4 +1,6 @@ -# $Id$ +# Thierry Parmentelat +# Copyright (C) 2010 INRIA +# import os, os.path import datetime import time @@ -17,6 +19,8 @@ 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 +48,8 @@ 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): @@ -56,42 +62,77 @@ 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='' +SEPSFA='' class TestPlc: default_steps = [ - 'display','uninstall','install','install_rpm', - 'configure', 'start', 'fetch_keys', SEP, - 'store_keys', 'clear_known_hosts', 'initscripts', SEP, - 'sites', 'nodes', 'slices', 'nodegroups', SEP, - 'init_node','bootcd', 'configure_qemu', 'export_qemu', - 'kill_all_qemus', 'reinstall_node','start_node', SEP, + 'display', 'resources_pre', SEP, + 'delete_vs','create_vs','install', 'configure', 'start', SEP, + 'fetch_keys', 'store_keys', 'clear_known_hosts', SEP, + 'initscripts', 'sites', 'nodes', 'slices', 'nodegroups', 'leases', 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, - 'check_tcp', SEP, - 'force_gather_logs', 'force_kill_qemus', 'force_record_tracker','force_free_tracker', + 'install_sfa', 'configure_sfa', 'import_sfa', 'start_sfa', SEPSFA, + 'setup_sfa', 'add_sfa', 'update_sfa', 'view_sfa', SEPSFA, + 'nodes_ssh_debug', 'nodes_ssh_boot', 'check_slice', 'check_initscripts', SEPSFA, + 'check_slice_sfa', 'delete_sfa', 'stop_sfa', SEPSFA, + 'check_tcp', 'check_hooks', SEP, + 'force_gather_logs', 'force_resources_post', SEP, ] other_steps = [ - 'stop_all_vservers','fresh_install', 'cache_rpm', 'stop', 'vs_start', SEP, - 'check_sanity', SEP, + 'show_boxes', 'resources_list','resources_release','resources_release_plc','resources_release_qemu',SEP, + 'stop', 'vs_start', SEP, 'clean_initscripts', 'clean_nodegroups','clean_all_sites', SEP, - 'clean_sites', 'clean_nodes', - 'clean_slices', 'clean_keys', SEP, - 'show_boxes', 'list_all_qemus', 'list_qemus', SEP, - 'db_dump' , 'db_restore', 'cleanup_trackers', 'cleanup_all_trackers', - 'standby_1 through 20', + 'clean_sites', 'clean_nodes', 'clean_slices', 'clean_keys', SEP, + 'clean_leases', 'list_leases', SEP, + 'populate' , SEP, + 'list_all_qemus', 'list_qemus', 'kill_qemus', SEP, + 'db_dump' , 'db_restore', SEP, + 'standby_1 through 20',SEP, ] @staticmethod def printable_steps (list): - return " ".join(list).replace(" "+SEP+" "," \\\n") + single_line=" ".join(list)+" " + return single_line.replace(" "+SEP+" "," \\\n").replace(" "+SEPSFA+" "," \\\n") @staticmethod def valid_step (step): - return step != SEP + return step != SEP and step != SEPSFA + + # turn off the sfa-related steps when build has skipped SFA + # this is originally for centos5 as recent SFAs won't build on this platformb + @staticmethod + def check_whether_build_has_sfa (rpms_url): + retcod=os.system ("curl --silent %s/ | grep -q sfa"%rpms_url) + # full builds are expected to return with 0 here + if retcod!=0: + # 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 ] + 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 @@ -237,12 +278,14 @@ class TestPlc: # a step for checking this stuff def show_boxes (self): + 'print summary of nodes location' for (box,nodes) in self.gather_hostBoxes().iteritems(): print box,":"," + ".join( [ node.name() for node in nodes ] ) return True # make this a valid step def kill_all_qemus(self): + 'kill all qemu instances on the qemu boxes involved by this setup' # this is the brute force version, kill all qemus on that host box for (box,nodes) in self.gather_hostBoxes().iteritems(): # pass the first nodename, as we don't push template-qemu on testboxes @@ -252,6 +295,7 @@ class TestPlc: # make this a valid step def list_all_qemus(self): + 'list all qemu instances on the qemu boxes involved by this setup' for (box,nodes) in self.gather_hostBoxes().iteritems(): # this is the brute force version, kill all qemus on that host box TestBox(box,self.options.buildname).list_all_qemus() @@ -259,6 +303,7 @@ class TestPlc: # kill only the right qemus def list_qemus(self): + 'list qemu instances for our nodes' for (box,nodes) in self.gather_hostBoxes().iteritems(): # the fine-grain version for node in nodes: @@ -267,6 +312,7 @@ class TestPlc: # kill only the right qemus def kill_qemus(self): + 'kill the qemu instances for our nodes' for (box,nodes) in self.gather_hostBoxes().iteritems(): # the fine-grain version for node in nodes: @@ -275,13 +321,16 @@ class TestPlc: #################### display config def display (self): + "show test configuration after localization" self.display_pass (1) self.display_pass (2) return True # entry point + always_display_keys=['PLC_WWW_HOST','nodes','sites',] def display_pass (self,passno): for (key,val) in self.plc_spec.iteritems(): + if not self.options.verbose and key not in TestPlc.always_display_keys: continue if passno == 2: if key == 'sites': for site in val: @@ -298,65 +347,67 @@ class TestPlc: for key in val: self.display_key_spec (key) elif passno == 1: - if key not in ['sites','initscripts','slices','keys']: - print '* ',key,':',val + if key not in ['sites','initscripts','slices','keys', 'sfa']: + print '+ ',key,':',val def display_site_spec (self,site): - print '* ======== site',site['site_fields']['name'] + print '+ ======== site',site['site_fields']['name'] for (k,v) in site.iteritems(): + if not self.options.verbose and k not in TestPlc.always_display_keys: continue if k=='nodes': if v: - print '* ','nodes : ', + print '+ ','nodes : ', for node in v: print node['node_fields']['hostname'],'', print '' elif k=='users': if v: - print '* users : ', + print '+ users : ', for user in v: print user['name'],'', print '' elif k == 'site_fields': - print '* login_base',':',v['login_base'] + print '+ login_base',':',v['login_base'] elif k == 'address_fields': pass else: - print '* ',k, - PrettyPrinter(indent=8,depth=2).pprint(v) + print '+ ', + utils.pprint(k,v) def display_initscript_spec (self,initscript): - print '* ======== initscript',initscript['initscript_fields']['name'] + print '+ ======== initscript',initscript['initscript_fields']['name'] def display_key_spec (self,key): - print '* ======== key',key['name'] + print '+ ======== key',key['name'] def display_slice_spec (self,slice): - print '* ======== slice',slice['slice_fields']['name'] + print '+ ======== slice',slice['slice_fields']['name'] for (k,v) in slice.iteritems(): if k=='nodenames': if v: - print '* nodes : ', + print '+ nodes : ', for nodename in v: print nodename,'', print '' elif k=='usernames': if v: - print '* users : ', + print '+ users : ', for username in v: print username,'', print '' elif k=='slice_fields': - print '* fields',':', + print '+ fields',':', print 'max_nodes=',v['max_nodes'], print '' else: - print '* ',k,v + print '+ ',k,v def display_node_spec (self,node): - print "* node",node['name'],"host_box=",node['host_box'], + print "+ node=%s host_box=%s"%(node['name'],node['host_box']), print "hostname=",node['node_fields']['hostname'], print "ip=",node['interface_fields']['ip'] - + if self.options.verbose: + utils.pprint("node details",node,depth=3) # another entry point for just showing the boxes involved def display_mapping (self): @@ -365,112 +416,61 @@ class TestPlc: @staticmethod def display_mapping_plc (plc_spec): - print '* MyPLC',plc_spec['name'] - print '*\tvserver address = root@%s:/vservers/%s'%(plc_spec['hostname'],plc_spec['vservername']) - print '*\tIP = %s/%s'%(plc_spec['PLC_API_HOST'],plc_spec['vserverip']) + print '+ MyPLC',plc_spec['name'] + 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) @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'] - - ### 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=os.environ['HOME']+"/running-test-plcs" - # how many concurrent plcs are we keeping alive - adjust with the IP pool size - TRACKER_KEEP_VSERVERS = 12 - - def record_tracker (self): - try: - lines=file(TestPlc.TRACKER_FILE).readlines() - except: - lines=[] - - this_line="%s %s\n"%(self.vservername,self.test_ssh.hostname) - for line in lines: - if line==this_line: - print 'this vserver is already included in %s'%TestPlc.TRACKER_FILE - return True - if self.options.dry_run: - print 'dry_run: record_tracker - skipping tracker update' - return True - tracker=file(TestPlc.TRACKER_FILE,"w") - for line in lines+[this_line]: - tracker.write(line) - tracker.close() - print "Recorded %s in running plcs on host %s"%(self.vservername,self.test_ssh.hostname) - return True - - def free_tracker (self, keep_vservers=None): - if not keep_vservers: keep_vservers=TestPlc.TRACKER_KEEP_VSERVERS - try: - lines=file(TestPlc.TRACKER_FILE).readlines() - except: - print 'dry_run: free_tracker - skipping tracker update' - return True - how_many = len(lines) - keep_vservers - # nothing todo until we have more than keep_vservers in the tracker - if how_many <= 0: - print 'free_tracker : limit %d not reached'%keep_vservers - return True - to_stop = lines[:how_many] - to_keep = lines[how_many:] - for line in to_stop: - print '>%s<'%line - [vname,hostname]=line.split() - command=TestSsh(hostname).actual_command("vserver --silent %s stop"%vname) - utils.system(command) - if self.options.dry_run: - print 'dry_run: free_tracker would stop %d vservers'%len(to_stop) - for line in to_stop: print line, - print 'dry_run: free_tracker would keep %d vservers'%len(to_keep) - for line in to_keep: print line, - return True - print "Storing %d remaining vservers in %s"%(len(to_keep),TestPlc.TRACKER_FILE) - tracker=open(TestPlc.TRACKER_FILE,"w") - for line in to_keep: - tracker.write(line) - tracker.close() - return True - - # this should/could stop only the ones in TRACKER_FILE if that turns out to be reliable - def cleanup_trackers (self): - try: - for line in TestPlc.TRACKER_FILE.readlines(): - [vname,hostname]=line.split() - stop="vserver --silent %s stop"%vname - command=TestSsh(hostname).actual_command(stop) - utils.system(command) - clean_tracker = "rm -f %s"%TestPlc.TRACKER_FILE - utils.system(self.test_ssh.actual_command(clean_tracker)) - except: - return True - - # this should/could stop only the ones in TRACKER_FILE if that turns out to be reliable - def cleanup_all_trackers (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)) - return True - - def uninstall(self): + print '+ NODE %s'%(node_spec['name']) + print '+\tqemu box %s'%node_spec['host_box'] + print '+\thostname=%s'%node_spec['node_fields']['hostname'] + + def resources_pre (self): + "run site-dependant pre-test script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_pre(self) + + def resources_post (self): + "run site-dependant post-test script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_post(self) + + def resources_list (self): + "run site-dependant list script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_list(self) + + def resources_release (self): + "run site-dependant release script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_release(self) + + def resources_release_plc (self): + "run site-dependant release script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_release_plc(self) + + def resources_release_qemu (self): + "run site-dependant release script as defined in LocalTestResources" + from LocalTestResources import local_resources + return local_resources.step_release_qemu(self) + + def delete_vs(self): + "vserver delete the test myplc" self.run_in_host("vserver --silent %s delete"%self.vservername) return True ### install - def install(self): + # historically the build was being fetched by the tests + # now the build pushes itself as a subdir of the tests workdir + # so that the tests do not have to worry about extracting the build (svn, git, or whatever) + def create_vs (self): + "vserver creation (no install done)" + # push the local build/ dir to the testplc box if self.is_local(): # a full path for the local calls build_dir=os.path.dirname(sys.argv[0]) @@ -480,10 +480,9 @@ class TestPlc: else: # use a standard name - will be relative to remote buildname build_dir="build" - # run checkout in any case - would do an update if already exists - build_checkout = "svn checkout %s %s"%(self.options.build_url,build_dir) - if self.run_in_host(build_checkout) != 0: - return False + # remove for safety; do *not* mkdir first, otherwise we end up with build/build/ + self.test_ssh.rmdir(build_dir) + self.test_ssh.copy(build_dir,recursive=True) # the repo url is taken from arch-rpms-url # with the last step (i386) removed repo_url = self.options.arch_rpms_url @@ -501,25 +500,40 @@ class TestPlc: vserver_hostname=socket.gethostbyaddr(self.vserverip)[0] vserver_options += " --hostname %s"%vserver_hostname except: - pass + print "Cannot reverse lookup %s"%self.vserverip + print "This is considered fatal, as this might pollute the test results" + return False create_vserver="%(build_dir)s/%(script)s %(test_env_options)s %(vserver_name)s %(repo_url)s -- %(vserver_options)s"%locals() return self.run_in_host(create_vserver) == 0 ### install_rpm - def install_rpm(self): + def install(self): + "yum install myplc, noderepo, and the plain bootstrapfs" + + # workaround for getting pgsql8.2 on centos5 + if self.options.fcdistro == "centos5": + self.run_in_guest("rpm -Uvh http://download.fedora.redhat.com/pub/epel/5/i386/epel-release-5-3.noarch.rpm") + 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) + + # try to install slicerepo - not fatal yet + self.run_in_guest("yum -y install slicerepo-%s"%nodefamily) + return \ - self.run_in_guest("yum -y install myplc-native")==0 and \ - self.run_in_guest("yum -y install noderepo-%s-%s"%(self.options.pldistro,arch))==0 and \ - self.run_in_guest("yum -y install bootstrapfs-%s-%s-plain"%(self.options.pldistro,arch))==0 + 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', @@ -528,11 +542,16 @@ 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', 'PLC_NET_DNS1', - 'PLC_NET_DNS2']: + 'PLC_NET_DNS2', + 'PLC_RESERVATION_GRANULARITY', + 'PLC_OMF_ENABLED', + ]: fileconf.write ('e %s\n%s\n'%(var,self.plc_spec[var])) fileconf.write('w\n') fileconf.write('q\n') @@ -543,39 +562,40 @@ 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 def vs_start (self): + "start the PLC vserver" self.start_guest() return True # 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 def clean_keys(self): + "removes keys cached in keys/" 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 = 'root_ssh_key' - for ext in [ 'pub', 'rsa' ] : - src="/vservers/%(vservername)s/etc/planetlab/%(prefix)s.%(ext)s"%locals() - dst="keys/%(vservername)s.%(ext)s"%locals() - if self.test_ssh.fetch(src,dst) != 0: overall=False prefix = 'debug_ssh_key' for ext in [ 'pub', 'rsa' ] : src="/vservers/%(vservername)s/etc/planetlab/%(prefix)s.%(ext)s"%locals() @@ -584,9 +604,11 @@ class TestPlc: 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"): @@ -605,6 +627,7 @@ class TestPlc: return True def clean_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: @@ -612,8 +635,10 @@ 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"): @@ -634,10 +659,74 @@ class TestPlc: return True 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") + YEAR = 365*24*3600 + @staticmethod + def translate_timestamp (start,grain,timestamp): + if timestamp < TestPlc.YEAR: return start+timestamp*grain + else: return timestamp + + @staticmethod + def timestamp_printable (timestamp): + return time.strftime('%m-%d %H:%M:%S UTC',time.gmtime(timestamp)) + + def leases(self): + "create leases (on reservable nodes only, use e.g. run -c default -c resa)" + now=int(time.time()) + grain=self.apiserver.GetLeaseGranularity(self.auth_root()) + print 'API answered grain=',grain + start=(now/grain)*grain + start += grain + # find out all nodes that are reservable + nodes=self.all_reservable_nodenames() + if not nodes: + utils.header ("No reservable node found - proceeding without leases") + return True + ok=True + # attach them to the leases as specified in plc_specs + # this is where the 'leases' field gets interpreted as relative of absolute + for lease_spec in self.plc_spec['leases']: + # skip the ones that come with a null slice id + if not lease_spec['slice']: continue + lease_spec['t_from']=TestPlc.translate_timestamp(start,grain,lease_spec['t_from']) + lease_spec['t_until']=TestPlc.translate_timestamp(start,grain,lease_spec['t_until']) + lease_addition=self.apiserver.AddLeases(self.auth_root(),nodes, + lease_spec['slice'],lease_spec['t_from'],lease_spec['t_until']) + if lease_addition['errors']: + utils.header("Cannot create leases, %s"%lease_addition['errors']) + ok=False + else: + utils.header('Leases on nodes %r for %s from %d (%s) until %d (%s)'%\ + (nodes,lease_spec['slice'], + lease_spec['t_from'],TestPlc.timestamp_printable(lease_spec['t_from']), + lease_spec['t_until'],TestPlc.timestamp_printable(lease_spec['t_until']))) + + return ok + + def clean_leases (self): + "remove all leases in the myplc side" + lease_ids= [ l['lease_id'] for l in self.apiserver.GetLeases(self.auth_root())] + utils.header("Cleaning leases %r"%lease_ids) + self.apiserver.DeleteLeases(self.auth_root(),lease_ids) + return True + + def list_leases (self): + "list all leases known to the myplc" + leases = self.apiserver.GetLeases(self.auth_root()) + now=int(time.time()) + for l in leases: + current=l['t_until']>=now + if self.options.verbose or current: + utils.header("%s %s from %s until %s"%(l['hostname'],l['name'], + TestPlc.timestamp_printable(l['t_from']), + TestPlc.timestamp_printable(l['t_until']))) + return True + # create nodegroups if needed, and populate def do_nodegroups (self, action="add"): # 1st pass to scan contents @@ -687,7 +776,7 @@ class TestPlc: expect_yes = self.apiserver.GetNodeTags(auth, {'hostname':nodename, 'tagname':nodegroupname}, - ['tagvalue'])[0]['tagvalue'] + ['value'])[0]['value'] if expect_yes != "yes": print 'Mismatch node tag on node',nodename,'got',expect_yes overall=False @@ -704,12 +793,23 @@ class TestPlc: 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'] \ - for node_spec in site_spec['nodes'] ] - return hostnames + node_infos += [ (node_spec['node_fields']['hostname'],node_spec['host_box']) \ + for node_spec in site_spec['nodes'] ] + return node_infos + + def all_nodenames (self): return [ x[0] for x in self.all_node_infos() ] + def all_reservable_nodenames (self): + res=[] + for site_spec in self.plc_spec['sites']: + for node_spec in site_spec['nodes']: + node_fields=node_spec['node_fields'] + if 'node_type' in node_fields and node_fields['node_type']=='reservable': + res.append(node_fields['hostname']) + return res # silent_minutes : during the first minutes nothing gets printed def nodes_check_boot_state (self, target_boot_state, timeout_minutes, silent_minutes,period=15): @@ -758,9 +858,9 @@ class TestPlc: return True def nodes_booted(self): - return self.nodes_check_boot_state('boot',timeout_minutes=20,silent_minutes=15) + return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=20) - def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=20): + def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=15): # compute timeout timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes) graceout = datetime.datetime.now()+datetime.timedelta(minutes=silent_minutes) @@ -770,43 +870,35 @@ class TestPlc: local_key = "keys/%(vservername)s-debug.rsa"%locals() else: message="boot" - local_key = "keys/%(vservername)s.rsa"%locals() - tocheck = self.all_hostnames() - utils.header("checking ssh access (expected in %s mode) to nodes %r"%(message,tocheck)) + local_key = "keys/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 tocheck: - for hostname in tocheck: + while node_infos: + for node_info in node_infos: + (hostname,qemuname) = node_info # try to run 'hostname' in the node command = TestSsh (hostname,key=local_key).actual_command("hostname;uname -a") # don't spam logs - show the command only after the grace period - if datetime.datetime.now() > graceout: - success=utils.system(command) - else: - # truly silent, just print out a dot to show we're alive - print '.', - sys.stdout.flush() - command += " 2>/dev/null" - if self.options.dry_run: - print 'dry_run',command - success=0 - else: - success=os.system(command) + success = utils.system ( command, silent=datetime.datetime.now() < graceout) if success==0: utils.header('Successfully entered root@%s (%s)'%(hostname,message)) - # refresh tocheck - tocheck.remove(hostname) + # refresh node_infos + node_infos.remove(node_info) else: # we will have tried real nodes once, in case they're up - but if not, just skip (site_spec,node_spec)=self.locate_hostname(hostname) if TestNode.is_real_model(node_spec['node_fields']['model']): utils.header ("WARNING : check ssh access into real node %s - skipped"%hostname) - tocheck.remove(hostname) - if not tocheck: + node_infos.remove(node_info) + if not node_infos: return True if datetime.datetime.now() > timeout: - for hostname in tocheck: - utils.header("FAILURE to ssh into %s"%hostname) + for (hostname,qemuname) in node_infos: + utils.header("FAILURE to ssh into %s (on %s)"%(hostname,qemuname)) return False # otherwise, sleep for a while time.sleep(period) @@ -814,30 +906,43 @@ class TestPlc: return True def nodes_ssh_debug(self): - return self.check_nodes_ssh(debug=True,timeout_minutes=30,silent_minutes=10) + "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) def nodes_ssh_boot(self): - return self.check_nodes_ssh(debug=False,timeout_minutes=30,silent_minutes=10) + "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) @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 sanity : invoke scripts from qaapi/qa/tests/{node,slice} - def check_sanity_node (self): - return self.locate_first_node().check_sanity() - def check_sanity_sliver (self) : - return self.locate_first_sliver().check_sanity() + ### check hooks : invoke scripts from hooks/{node,slice} + def check_hooks_node (self): + return self.locate_first_node().check_hooks() + def check_hooks_sliver (self) : + return self.locate_first_sliver().check_hooks() - def check_sanity (self): - return self.check_sanity_node() and self.check_sanity_sliver() + def check_hooks (self): + "runs unit tests in the node and slice contexts - see hooks/{node,slice}" + return self.check_hooks_node() and self.check_hooks_sliver() ### initscripts def do_check_initscripts(self): @@ -858,15 +963,18 @@ 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'])) @@ -879,9 +987,11 @@ class TestPlc: ### 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"): @@ -899,15 +1009,22 @@ 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 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: @@ -925,17 +1042,203 @@ 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.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): + "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 @@ -989,9 +1292,9 @@ 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) @@ -1010,19 +1313,21 @@ class TestPlc: return "/root/%s-%s.sql"%(database,name) def db_dump(self): - dump=self.dbfile("planetab4") - self.run_in_guest('pg_dump -U pgsqluser planetlab4 -f '+ dump) - utils.header('Dumped planetlab4 database in %s'%dump) + 'dump the planetlab5 DB in /root in the PLC - filename has time' + dump=self.dbfile("planetab5") + self.run_in_guest('pg_dump -U pgsqluser planetlab5 -f '+ dump) + utils.header('Dumped planetlab5 database in %s'%dump) return True def db_restore(self): - dump=self.dbfile("planetab4") + 'restore the planetlab5 DB - looks broken, but run -n might help' + dump=self.dbfile("planetab5") ##stop httpd service self.run_in_guest('service httpd stop') # xxx - need another wrapper - self.run_in_guest_piped('echo drop database planetlab4','psql --user=pgsqluser template1') - self.run_in_guest('createdb -U postgres --encoding=UNICODE --owner=pgsqluser planetlab4') - self.run_in_guest('psql -U pgsqluser planetlab4 -f '+dump) + self.run_in_guest_piped('echo drop database planetlab5','psql --user=pgsqluser template1') + self.run_in_guest('createdb -U postgres --encoding=UNICODE --owner=pgsqluser planetlab5') + self.run_in_guest('psql -U pgsqluser planetlab5 -f '+dump) ##starting httpd service self.run_in_guest('service httpd start') @@ -1068,4 +1373,3 @@ class TestPlc: def standby_19(): pass @standby_generic def standby_20(): pass -