X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=aced6505af8a073da873c2721451ee42c0be200c;hb=25de2f0b5602100459eeaadab5204aec2fc790ea;hp=9cd218106a1c0985dca2c2e79d824369cd54efd2;hpb=2ffdcb76d2bcb838118419d8553afc3d07e4765c;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index 9cd2181..aced650 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -45,7 +45,7 @@ def node_mapper (method): 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): @@ -59,7 +59,7 @@ 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): @@ -67,13 +67,11 @@ def slice_sfa_mapper (method): 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) + test_slice=TestSliceSfa(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__=TestSliceSfa.__dict__[method.__name__].__doc__ return actual SEP='' @@ -85,20 +83,23 @@ class TestPlc: 'show', SEP, 'vs_delete','timestamp_vs','vs_create', SEP, 'plc_install', 'plc_configure', 'plc_start', SEP, - 'keys_fetch', 'keys_store', 'keys_clear_known_hosts', 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', 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_user@1', 'sfa_add_slice@1', 'sfa_discover@1', SEPSFA, + 'sfi_configure@1', 'sfa_add_site@1','sfa_add_pi@1', SEPSFA, + '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, + 'sfa_update_user@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@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', 'check_netflow', SEP, + 'check_tcp', 'check_sys_slice', SEP, + 'empty_slices', 'ssh_slice_off', 'fill_slices', SEP, 'force_gather_logs', SEP, ] other_steps = [ @@ -107,7 +108,7 @@ class TestPlc: '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_all', SEP, 'sfa_install_core', 'sfa_install_sfatables', 'sfa_install_plc', 'sfa_install_client', SEPSFA, @@ -125,15 +126,20 @@ 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 platform + # 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) @@ -146,6 +152,9 @@ class TestPlc: 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) @@ -191,7 +200,7 @@ class TestPlc: def vm_timestamp_path (self): if self.options.plcs_use_lxc: - return "/var/lib/lxc/%s.timestamp"%(self.vservername) + return "/var/lib/lxc/%s/%s.timestamp"%(self.vservername,self.vservername) else: return "/vservers/%s.timestamp"%(self.vservername) @@ -212,6 +221,11 @@ class TestPlc: 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): @@ -219,7 +233,7 @@ class TestPlc: 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.run_in_guest("rpm -q %s"%rpms)==0 + return self.yum_check_installed (rpms) def auth_root (self): return {'Username':self.plc_spec['PLC_ROOT_USER'], @@ -357,8 +371,8 @@ 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): @@ -367,8 +381,13 @@ class TestPlc: domain=socket.gethostname().split('.',1)[1] fqdn="%s.%s"%(self.plc_spec['host_box'],domain) print "export BUILD=%s"%self.options.buildname - print "export PLCHOST=%s"%fqdn - print "export GUEST=%s"%self.plc_spec['vservername'] + 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) @@ -377,7 +396,7 @@ class TestPlc: # 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: @@ -482,6 +501,7 @@ class TestPlc: # 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()) # TODO-lxc check this one # a first approx. is to store the timestamp close to the VM root like vs does @@ -681,8 +701,11 @@ 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 @@ -975,37 +998,21 @@ class TestPlc: 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): @@ -1061,45 +1068,65 @@ 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)" @@ -1120,9 +1147,12 @@ class TestPlc: return overall # painfully enough, we need to allow for some time as netflow might show up last - def check_netflow (self): - "all nodes: check that the netflow slice is alive" - return self.check_systemslice ('netflow') + 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): @@ -1130,7 +1160,7 @@ class TestPlc: test_nodes=self.all_nodes() while test_nodes: for test_node in test_nodes: - if test_node.check_systemslice (slicename): + if test_node.check_systemslice (slicename,dry_run=self.options.dry_run): utils.header ("ok") test_nodes.remove(test_node) else: @@ -1172,20 +1202,45 @@ class TestPlc: "yum install sfa-plc" return self.yum_install("sfa-plc") - def sfa_install_client(self): - "yum install sfa-client" - return self.yum_install("sfa-client") - 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.py")==0 or \ - self.run_in_guest("sfa-nuke-plc.py") or \ - self.run_in_guest("sfaadmin.py registry nuke") - 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" @@ -1193,13 +1248,17 @@ class TestPlc: sfa_spec=self.plc_spec['sfa'] for sfa_slice_spec in sfa_spec['sfa_slice_specs']: - slicename='%s_%s'%(sfa_slice_spec['login_base'],sfa_slice_spec['slicename']) - try: self.apiserver.DeleteSlice(self.auth_root(),slicename) - except: print "Slice %s already absent from PLC db"%slicename - - username="%s@%s"%(sfa_slice_spec['regularuser'],sfa_slice_spec['domain']) - try: self.apiserver.DeletePerson(self.auth_root(),username) - except: print "User %s already absent from PLC db"%username + login_base=sfa_slice_spec['login_base'] + try: self.apiserver.DeleteSite (self.auth_root(),login_base) + except: print "Site %s already absent from PLC db"%login_base + + for key in ['piuser','regularuser']: + username="%s@%s"%(sfa_slice_spec[key],sfa_slice_spec['domain']) + 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 @@ -1277,6 +1336,8 @@ class TestPlc: 'SFA_DB_PASSWORD', 'SFA_DB_NAME', 'SFA_API_LOGLEVEL', + 'SFA_GENERIC_FLAVOUR', + 'SFA_AGGREGATE_ENABLED', ]: if self.plc_spec['sfa'].has_key(var): fileconf.write ('e %s\n%s\n'%(var,self.plc_spec['sfa'][var])) @@ -1323,9 +1384,8 @@ class TestPlc: def sfa_import(self): "sfa-import-plc" auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'] - return self.run_in_guest('sfa-import.py')==0 or \ - self.run_in_guest('sfa-import-plc.py')==0 or \ - self.run_in_guest('sfaadmin.py registry import_registry')==0 + 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)) @@ -1339,13 +1399,12 @@ class TestPlc: utils.header("DRY RUN - skipping step") return True sfa_spec=self.plc_spec['sfa'] - # cannot use sfa_slice_mapper to pass dir_name + # cannot use slice_sfa_mapper to pass dir_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) - dir_name=self.confsubdir("dot-sfi/%s"%slice_spec['slicename'],clean=True,dry_run=self.options.dry_run) - test_slice.sfi_config(dir_name) + test_slice=TestSliceSfa(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) @@ -1361,58 +1420,35 @@ class TestPlc: return True @slice_sfa_mapper - def sfa_add_user(self): - "run sfi.py add" - pass - + def sfa_add_site (self): pass @slice_sfa_mapper - def sfa_update_user(self): - "run sfi.py update" - + def sfa_add_pi (self): pass @slice_sfa_mapper - def sfa_add_slice(self): - "run sfi.py add (on Registry) from slice.xml" - pass - + def sfa_add_user(self): pass @slice_sfa_mapper - def sfa_discover(self): - "discover resources into resouces_in.rspec" - pass - + def sfa_update_user(self): pass @slice_sfa_mapper - def sfa_create_slice(self): - "run sfi.py create (on SM) - 1st time" - pass - + def sfa_add_slice(self): 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 - + def sfa_discover(self): pass @slice_sfa_mapper - def sfa_update_slice(self): - "run sfi.py create (on SM) on existing object" - pass - + def sfa_create_slice(self): pass @slice_sfa_mapper - def sfa_view(self): - "various registry-related calls" - pass - + def sfa_check_slice_plc(self): pass @slice_sfa_mapper - def ssh_slice_sfa(self): - "tries to ssh-enter the SFA slice" - pass - + def sfa_update_slice(self): pass @slice_sfa_mapper - def sfa_delete_user(self): - "run sfi.py delete" - pass - + def sfi_list(self): pass @slice_sfa_mapper - def sfa_delete_slice(self): - "run sfi.py delete (on SM), sfi.py remove (on Registry) to clean slices" - pass + def sfi_show(self): pass + @slice_sfa_mapper + def sfi_slices(self): pass + @slice_sfa_mapper + def ssh_slice_sfa(self): pass + @slice_sfa_mapper + def sfa_delete_user(self): pass + @slice_sfa_mapper + def sfa_delete_slice(self): pass def sfa_stop(self): "service sfa stop" @@ -1437,6 +1473,7 @@ 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./* @@ -1446,6 +1483,9 @@ 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']: @@ -1483,6 +1523,12 @@ 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)