1 # Thierry Parmentelat <thierry.parmentelat@inria.fr>
2 # Copyright (C) 2010 INRIA
9 from datetime import datetime, timedelta
10 from types import StringTypes
13 from Completer import Completer, CompleterTask
14 from TestSite import TestSite
15 from TestNode import TestNode, CompleterTaskNodeSsh
16 from TestUser import TestUser
17 from TestKey import TestKey
18 from TestSlice import TestSlice
19 from TestSliver import TestSliver
20 from TestBoxQemu import TestBoxQemu
21 from TestSsh import TestSsh
22 from TestApiserver import TestApiserver
23 from TestAuthSfa import TestAuthSfa
24 from PlcapiUrlScanner import PlcapiUrlScanner
26 has_sfa_cache_filename="sfa-cache"
28 # step methods must take (self) and return a boolean (options is a member of the class)
30 def standby(minutes,dry_run):
31 utils.header('Entering StandBy for %d mn'%minutes)
35 time.sleep(60*minutes)
38 def standby_generic (func):
40 minutes=int(func.__name__.split("_")[1])
41 return standby(minutes,self.options.dry_run)
44 def node_mapper (method):
45 def map_on_nodes(self,*args, **kwds):
47 node_method = TestNode.__dict__[method.__name__]
48 for test_node in self.all_nodes():
49 if not node_method(test_node, *args, **kwds): overall=False
51 # maintain __name__ for ignore_result
52 map_on_nodes.__name__=method.__name__
53 # restore the doc text
54 map_on_nodes.__doc__=TestNode.__dict__[method.__name__].__doc__
57 def slice_mapper (method):
58 def map_on_slices(self):
60 slice_method = TestSlice.__dict__[method.__name__]
61 for slice_spec in self.plc_spec['slices']:
62 site_spec = self.locate_site (slice_spec['sitename'])
63 test_site = TestSite(self,site_spec)
64 test_slice=TestSlice(self,test_site,slice_spec)
65 if not slice_method(test_slice,self.options): overall=False
67 # maintain __name__ for ignore_result
68 map_on_slices.__name__=method.__name__
69 # restore the doc text
70 map_on_slices.__doc__=TestSlice.__dict__[method.__name__].__doc__
73 # run a step but return True so that we can go on
74 def ignore_result (method):
76 # ssh_slice_ignore->ssh_slice
77 ref_name=method.__name__.replace('_ignore','').replace('force_','')
78 ref_method=TestPlc.__dict__[ref_name]
79 result=ref_method(self)
80 print "Actual (but ignored) result for %(ref_name)s is %(result)s"%locals()
81 return Ignored (result)
82 name=method.__name__.replace('_ignore','').replace('force_','')
83 ignoring.__name__=name
84 ignoring.__doc__="ignored version of " + name
87 # a variant that expects the TestSlice method to return a list of CompleterTasks that
88 # are then merged into a single Completer run to avoid wating for all the slices
89 # esp. useful when a test fails of course
90 # because we need to pass arguments we use a class instead..
91 class slice_mapper__tasks (object):
92 # could not get this to work with named arguments
93 def __init__ (self,timeout_minutes,silent_minutes,period_seconds):
94 self.timeout=timedelta(minutes=timeout_minutes)
95 self.silent=timedelta(minutes=silent_minutes)
96 self.period=timedelta(seconds=period_seconds)
97 def __call__ (self, method):
99 # compute augmented method name
100 method_name = method.__name__ + "__tasks"
101 # locate in TestSlice
102 slice_method = TestSlice.__dict__[ method_name ]
105 for slice_spec in self.plc_spec['slices']:
106 site_spec = self.locate_site (slice_spec['sitename'])
107 test_site = TestSite(self,site_spec)
108 test_slice=TestSlice(self,test_site,slice_spec)
109 tasks += slice_method (test_slice, self.options)
110 return Completer (tasks).run (decorator_self.timeout, decorator_self.silent, decorator_self.period)
111 # restore the doc text from the TestSlice method even if a bit odd
112 wrappee.__name__ = method.__name__
113 wrappee.__doc__ = slice_method.__doc__
116 def auth_sfa_mapper (method):
119 auth_method = TestAuthSfa.__dict__[method.__name__]
120 for auth_spec in self.plc_spec['sfa']['auth_sfa_specs']:
121 test_auth=TestAuthSfa(self,auth_spec)
122 if not auth_method(test_auth,self.options): overall=False
124 # restore the doc text
125 actual.__doc__=TestAuthSfa.__dict__[method.__name__].__doc__
129 def __init__ (self,result):
139 'plcvm_delete','plcvm_timestamp','plcvm_create', SEP,
140 'plc_install', 'plc_configure', 'plc_start', SEP,
141 'keys_fetch', 'keys_store', 'keys_clear_known_hosts', SEP,
142 'plcapi_urls','speed_up_slices', SEP,
143 'initscripts', 'sites', 'nodes', 'slices', 'nodegroups', 'leases', SEP,
144 # slices created under plcsh interactively seem to be fine but these ones don't have the tags
145 # keep this our of the way for now
146 'check_vsys_defaults_ignore', SEP,
147 # run this first off so it's easier to re-run on another qemu box
148 'qemu_kill_mine', SEP,
149 'nodestate_reinstall', 'qemu_local_init','bootcd', 'qemu_local_config', SEP,
150 'qemu_clean_mine', 'qemu_export', 'qemu_start', 'qemu_timestamp', SEP,
151 'sfa_install_all', 'sfa_configure', 'cross_sfa_configure', 'sfa_start', 'sfa_import', SEPSFA,
152 'sfi_configure@1', 'sfa_register_site@1','sfa_register_pi@1', SEPSFA,
153 'sfa_register_user@1', 'sfa_update_user@1', 'sfa_register_slice@1', 'sfa_renew_slice@1', SEPSFA,
154 'sfa_remove_user_from_slice@1','sfi_show_slice_researchers@1',
155 'sfa_insert_user_in_slice@1','sfi_show_slice_researchers@1', SEPSFA,
156 'sfa_discover@1', 'sfa_create_slice@1', 'sfa_check_slice_plc@1', 'sfa_update_slice@1', SEPSFA,
157 'sfi_list@1', 'sfi_show_site@1', 'sfa_utest@1', SEPSFA,
158 # we used to run plcsh_stress_test, and then ssh_node_debug and ssh_node_boot
159 # but as the stress test might take a while, we sometimes missed the debug mode..
160 'probe_kvm_iptables',
161 'ping_node', 'ssh_node_debug', 'plcsh_stress_test@1', SEP,
162 'ssh_node_boot', 'node_bmlogs', 'ssh_slice', 'ssh_slice_basics', 'check_initscripts_ignore', SEP,
163 'ssh_slice_sfa@1', 'sfa_delete_slice@1', 'sfa_delete_user@1', SEPSFA,
164 'cross_check_tcp@1', 'check_system_slice', SEP,
165 # check slices are turned off properly
166 'empty_slices', 'ssh_slice_off', 'slice_fs_deleted_ignore', SEP,
167 # check they are properly re-created with the same name
168 'fill_slices', 'ssh_slice_again_ignore', SEP,
169 'gather_logs_force', SEP,
172 'export', 'show_boxes', SEP,
173 'check_hooks', 'plc_stop', 'plcvm_start', 'plcvm_stop', SEP,
174 'delete_initscripts', 'delete_nodegroups','delete_all_sites', SEP,
175 'delete_sites', 'delete_nodes', 'delete_slices', 'keys_clean', SEP,
176 'delete_leases', 'list_leases', SEP,
178 'nodestate_show','nodestate_safeboot','nodestate_boot', SEP,
179 'qemu_list_all', 'qemu_list_mine', 'qemu_kill_all', SEP,
180 'sfa_install_core', 'sfa_install_sfatables', 'sfa_install_plc', 'sfa_install_client', SEPSFA,
181 'sfa_plcclean', 'sfa_dbclean', 'sfa_stop','sfa_uninstall', 'sfi_clean', SEPSFA,
182 'plc_db_dump' , 'plc_db_restore', SEP,
183 'check_netflow','check_drl', SEP,
184 'debug_nodemanager', 'slice_fs_present', SEP,
185 'standby_1_through_20','yes','no',SEP,
189 def printable_steps (list):
190 single_line=" ".join(list)+" "
191 return single_line.replace(" "+SEP+" "," \\\n").replace(" "+SEPSFA+" "," \\\n")
193 def valid_step (step):
194 return step != SEP and step != SEPSFA
196 # turn off the sfa-related steps when build has skipped SFA
197 # this was originally for centos5 but is still valid
198 # for up to f12 as recent SFAs with sqlalchemy won't build before f14
200 def _has_sfa_cached (rpms_url):
201 if os.path.isfile(has_sfa_cache_filename):
202 cached=file(has_sfa_cache_filename).read()=="yes"
203 utils.header("build provides SFA (cached):%s"%cached)
205 # warning, we're now building 'sface' so let's be a bit more picky
206 # full builds are expected to return with 0 here
207 utils.header ("Checking if build provides SFA package...")
208 retcod=os.system ("curl --silent %s/ | grep -q sfa-"%rpms_url)==0
209 encoded='yes' if retcod else 'no'
210 file(has_sfa_cache_filename,'w').write(encoded)
214 def check_whether_build_has_sfa (rpms_url):
215 has_sfa=TestPlc._has_sfa_cached(rpms_url)
217 utils.header("build does provide SFA")
219 # move all steps containing 'sfa' from default_steps to other_steps
220 utils.header("SFA package not found - removing steps with sfa or sfi")
221 sfa_steps= [ step for step in TestPlc.default_steps if step.find('sfa')>=0 or step.find("sfi")>=0 ]
222 TestPlc.other_steps += sfa_steps
223 for step in sfa_steps: TestPlc.default_steps.remove(step)
225 def __init__ (self,plc_spec,options):
226 self.plc_spec=plc_spec
228 self.test_ssh=TestSsh(self.plc_spec['host_box'],self.options.buildname)
229 self.vserverip=plc_spec['vserverip']
230 self.vservername=plc_spec['vservername']
231 self.url="https://%s:443/PLCAPI/"%plc_spec['vserverip']
232 self.apiserver=TestApiserver(self.url,options.dry_run)
233 (self.ssh_node_boot_timeout,self.ssh_node_boot_silent)=plc_spec['ssh_node_boot_timers']
234 (self.ssh_node_debug_timeout,self.ssh_node_debug_silent)=plc_spec['ssh_node_debug_timers']
236 def has_addresses_api (self):
237 return self.apiserver.has_method('AddIpAddress')
240 name=self.plc_spec['name']
241 return "%s.%s"%(name,self.vservername)
244 return self.plc_spec['host_box']
247 return self.test_ssh.is_local()
249 # define the API methods on this object through xmlrpc
250 # would help, but not strictly necessary
254 def actual_command_in_guest (self,command, backslash=False):
255 raw1=self.host_to_guest(command)
256 raw2=self.test_ssh.actual_command(raw1,dry_run=self.options.dry_run, backslash=backslash)
259 def start_guest (self):
260 return utils.system(self.test_ssh.actual_command(self.start_guest_in_host(),dry_run=self.options.dry_run))
262 def stop_guest (self):
263 return utils.system(self.test_ssh.actual_command(self.stop_guest_in_host(),dry_run=self.options.dry_run))
265 def run_in_guest (self,command,backslash=False):
266 raw=self.actual_command_in_guest(command,backslash)
267 return utils.system(raw)
269 def run_in_host (self,command):
270 return self.test_ssh.run_in_buildname(command, dry_run=self.options.dry_run)
272 # backslashing turned out so awful at some point that I've turned off auto-backslashing
273 # see e.g. plc_start esp. the version for f14
274 #command gets run in the plc's vm
275 def host_to_guest(self,command):
276 # f14 still needs some extra help
277 if self.options.fcdistro == 'f14':
278 raw="virsh -c lxc:/// lxc-enter-namespace %s -- /usr/bin/env PATH=/bin:/sbin:/usr/bin:/usr/sbin %s" %(self.vservername,command)
280 raw="virsh -c lxc:/// lxc-enter-namespace %s -- /usr/bin/env %s" %(self.vservername,command)
283 # this /vservers thing is legacy...
284 def vm_root_in_host(self):
285 return "/vservers/%s/"%(self.vservername)
287 def vm_timestamp_path (self):
288 return "/vservers/%s/%s.timestamp"%(self.vservername,self.vservername)
290 #start/stop the vserver
291 def start_guest_in_host(self):
292 return "virsh -c lxc:/// start %s"%(self.vservername)
294 def stop_guest_in_host(self):
295 return "virsh -c lxc:/// destroy %s"%(self.vservername)
298 def run_in_guest_piped (self,local,remote):
299 return utils.system(local+" | "+self.test_ssh.actual_command(self.host_to_guest(remote),keep_stdin=True))
301 def yum_check_installed (self, rpms):
302 if isinstance (rpms, list):
304 return self.run_in_guest("rpm -q %s"%rpms)==0
306 # does a yum install in the vs, ignore yum retcod, check with rpm
307 def yum_install (self, rpms):
308 if isinstance (rpms, list):
310 self.run_in_guest("yum -y install %s"%rpms)
311 # yum-complete-transaction comes with yum-utils, that is in vtest.pkgs
312 self.run_in_guest("yum-complete-transaction -y")
313 return self.yum_check_installed (rpms)
315 def auth_root (self):
316 return {'Username':self.plc_spec['PLC_ROOT_USER'],
317 'AuthMethod':'password',
318 'AuthString':self.plc_spec['PLC_ROOT_PASSWORD'],
319 'Role' : self.plc_spec['role']
321 def locate_site (self,sitename):
322 for site in self.plc_spec['sites']:
323 if site['site_fields']['name'] == sitename:
325 if site['site_fields']['login_base'] == sitename:
327 raise Exception,"Cannot locate site %s"%sitename
329 def locate_node (self,nodename):
330 for site in self.plc_spec['sites']:
331 for node in site['nodes']:
332 if node['name'] == nodename:
334 raise Exception,"Cannot locate node %s"%nodename
336 def locate_hostname (self,hostname):
337 for site in self.plc_spec['sites']:
338 for node in site['nodes']:
339 if node['node_fields']['hostname'] == hostname:
341 raise Exception,"Cannot locate hostname %s"%hostname
343 def locate_key (self,key_name):
344 for key in self.plc_spec['keys']:
345 if key['key_name'] == key_name:
347 raise Exception,"Cannot locate key %s"%key_name
349 def locate_private_key_from_key_names (self, key_names):
350 # locate the first avail. key
352 for key_name in key_names:
353 key_spec=self.locate_key(key_name)
354 test_key=TestKey(self,key_spec)
355 publickey=test_key.publicpath()
356 privatekey=test_key.privatepath()
357 if os.path.isfile(publickey) and os.path.isfile(privatekey):
359 if found: return privatekey
362 def locate_slice (self, slicename):
363 for slice in self.plc_spec['slices']:
364 if slice['slice_fields']['name'] == slicename:
366 raise Exception,"Cannot locate slice %s"%slicename
368 def all_sliver_objs (self):
370 for slice_spec in self.plc_spec['slices']:
371 slicename = slice_spec['slice_fields']['name']
372 for nodename in slice_spec['nodenames']:
373 result.append(self.locate_sliver_obj (nodename,slicename))
376 def locate_sliver_obj (self,nodename,slicename):
377 (site,node) = self.locate_node(nodename)
378 slice = self.locate_slice (slicename)
380 test_site = TestSite (self, site)
381 test_node = TestNode (self, test_site,node)
382 # xxx the slice site is assumed to be the node site - mhh - probably harmless
383 test_slice = TestSlice (self, test_site, slice)
384 return TestSliver (self, test_node, test_slice)
386 def locate_first_node(self):
387 nodename=self.plc_spec['slices'][0]['nodenames'][0]
388 (site,node) = self.locate_node(nodename)
389 test_site = TestSite (self, site)
390 test_node = TestNode (self, test_site,node)
393 def locate_first_sliver (self):
394 slice_spec=self.plc_spec['slices'][0]
395 slicename=slice_spec['slice_fields']['name']
396 nodename=slice_spec['nodenames'][0]
397 return self.locate_sliver_obj(nodename,slicename)
399 # all different hostboxes used in this plc
400 def get_BoxNodes(self):
401 # maps on sites and nodes, return [ (host_box,test_node) ]
403 for site_spec in self.plc_spec['sites']:
404 test_site = TestSite (self,site_spec)
405 for node_spec in site_spec['nodes']:
406 test_node = TestNode (self, test_site, node_spec)
407 if not test_node.is_real():
408 tuples.append( (test_node.host_box(),test_node) )
409 # transform into a dict { 'host_box' -> [ test_node .. ] }
411 for (box,node) in tuples:
412 if not result.has_key(box):
415 result[box].append(node)
418 # a step for checking this stuff
419 def show_boxes (self):
420 'print summary of nodes location'
421 for (box,nodes) in self.get_BoxNodes().iteritems():
422 print box,":"," + ".join( [ node.name() for node in nodes ] )
425 # make this a valid step
426 def qemu_kill_all(self):
427 'kill all qemu instances on the qemu boxes involved by this setup'
428 # this is the brute force version, kill all qemus on that host box
429 for (box,nodes) in self.get_BoxNodes().iteritems():
430 # pass the first nodename, as we don't push template-qemu on testboxes
431 nodedir=nodes[0].nodedir()
432 TestBoxQemu(box,self.options.buildname).qemu_kill_all(nodedir)
435 # make this a valid step
436 def qemu_list_all(self):
437 'list all qemu instances on the qemu boxes involved by this setup'
438 for (box,nodes) in self.get_BoxNodes().iteritems():
439 # this is the brute force version, kill all qemus on that host box
440 TestBoxQemu(box,self.options.buildname).qemu_list_all()
443 # kill only the qemus related to this test
444 def qemu_list_mine(self):
445 'list qemu instances for our nodes'
446 for (box,nodes) in self.get_BoxNodes().iteritems():
447 # the fine-grain version
452 # kill only the qemus related to this test
453 def qemu_clean_mine(self):
454 'cleanup (rm -rf) qemu instances for our nodes'
455 for (box,nodes) in self.get_BoxNodes().iteritems():
456 # the fine-grain version
461 # kill only the right qemus
462 def qemu_kill_mine(self):
463 'kill the qemu instances for our nodes'
464 for (box,nodes) in self.get_BoxNodes().iteritems():
465 # the fine-grain version
470 #################### display config
472 "show test configuration after localization"
477 # uggly hack to make sure 'run export' only reports about the 1st plc
478 # to avoid confusion - also we use 'inri_slice1' in various aliases..
481 "print cut'n paste-able stuff to export env variables to your shell"
482 # guess local domain from hostname
483 if TestPlc.exported_id>1:
484 print "export GUESTHOSTNAME%d=%s"%(TestPlc.exported_id,self.plc_spec['vservername'])
486 TestPlc.exported_id+=1
487 domain=socket.gethostname().split('.',1)[1]
488 fqdn="%s.%s"%(self.plc_spec['host_box'],domain)
489 print "export BUILD=%s"%self.options.buildname
490 print "export PLCHOSTLXC=%s"%fqdn
491 print "export GUESTNAME=%s"%self.plc_spec['vservername']
492 vplcname=self.plc_spec['vservername'].split('-')[-1]
493 print "export GUESTHOSTNAME=%s.%s"%(vplcname,domain)
494 # find hostname of first node
495 (hostname,qemubox) = self.all_node_infos()[0]
496 print "export KVMHOST=%s.%s"%(qemubox,domain)
497 print "export NODE=%s"%(hostname)
501 always_display_keys=['PLC_WWW_HOST','nodes','sites',]
502 def show_pass (self,passno):
503 for (key,val) in self.plc_spec.iteritems():
504 if not self.options.verbose and key not in TestPlc.always_display_keys: continue
508 self.display_site_spec(site)
509 for node in site['nodes']:
510 self.display_node_spec(node)
511 elif key=='initscripts':
512 for initscript in val:
513 self.display_initscript_spec (initscript)
516 self.display_slice_spec (slice)
519 self.display_key_spec (key)
521 if key not in ['sites','initscripts','slices','keys']:
522 print '+ ',key,':',val
524 def display_site_spec (self,site):
525 print '+ ======== site',site['site_fields']['name']
526 for (k,v) in site.iteritems():
527 if not self.options.verbose and k not in TestPlc.always_display_keys: continue
530 print '+ ','nodes : ',
532 print node['node_fields']['hostname'],'',
538 print user['name'],'',
540 elif k == 'site_fields':
541 print '+ login_base',':',v['login_base']
542 elif k == 'address_fields':
548 def display_initscript_spec (self,initscript):
549 print '+ ======== initscript',initscript['initscript_fields']['name']
551 def display_key_spec (self,key):
552 print '+ ======== key',key['key_name']
554 def display_slice_spec (self,slice):
555 print '+ ======== slice',slice['slice_fields']['name']
556 for (k,v) in slice.iteritems():
569 elif k=='slice_fields':
570 print '+ fields',':',
571 print 'max_nodes=',v['max_nodes'],
576 def display_node_spec (self,node):
577 print "+ node=%s host_box=%s"%(node['name'],node['host_box']),
578 print "hostname=",node['node_fields']['hostname'],
579 print "ip=",node['interface_fields']['ip']
580 if self.options.verbose:
581 utils.pprint("node details",node,depth=3)
583 # another entry point for just showing the boxes involved
584 def display_mapping (self):
585 TestPlc.display_mapping_plc(self.plc_spec)
589 def display_mapping_plc (plc_spec):
590 print '+ MyPLC',plc_spec['name']
591 # WARNING this would not be right for lxc-based PLC's - should be harmless though
592 print '+\tvserver address = root@%s:/vservers/%s'%(plc_spec['host_box'],plc_spec['vservername'])
593 print '+\tIP = %s/%s'%(plc_spec['PLC_API_HOST'],plc_spec['vserverip'])
594 for site_spec in plc_spec['sites']:
595 for node_spec in site_spec['nodes']:
596 TestPlc.display_mapping_node(node_spec)
599 def display_mapping_node (node_spec):
600 print '+ NODE %s'%(node_spec['name'])
601 print '+\tqemu box %s'%node_spec['host_box']
602 print '+\thostname=%s'%node_spec['node_fields']['hostname']
604 # write a timestamp in /vservers/<>.timestamp
605 # cannot be inside the vserver, that causes vserver .. build to cough
606 def plcvm_timestamp (self):
607 "Create a timestamp to remember creation date for this plc"
609 # TODO-lxc check this one
610 # a first approx. is to store the timestamp close to the VM root like vs does
611 stamp_path=self.vm_timestamp_path ()
612 stamp_dir = os.path.dirname (stamp_path)
613 utils.system(self.test_ssh.actual_command("mkdir -p %s"%stamp_dir))
614 return utils.system(self.test_ssh.actual_command("echo %d > %s"%(now,stamp_path)))==0
616 # this is called inconditionnally at the beginning of the test sequence
617 # just in case this is a rerun, so if the vm is not running it's fine
618 def plcvm_delete(self):
619 "vserver delete the test myplc"
620 stamp_path=self.vm_timestamp_path()
621 self.run_in_host("rm -f %s"%stamp_path)
622 self.run_in_host("virsh -c lxc:// destroy %s"%self.vservername)
623 self.run_in_host("virsh -c lxc:// undefine %s"%self.vservername)
624 self.run_in_host("rm -fr /vservers/%s"%self.vservername)
628 # historically the build was being fetched by the tests
629 # now the build pushes itself as a subdir of the tests workdir
630 # so that the tests do not have to worry about extracting the build (svn, git, or whatever)
631 def plcvm_create (self):
632 "vserver creation (no install done)"
633 # push the local build/ dir to the testplc box
635 # a full path for the local calls
636 build_dir=os.path.dirname(sys.argv[0])
637 # sometimes this is empty - set to "." in such a case
638 if not build_dir: build_dir="."
639 build_dir += "/build"
641 # use a standard name - will be relative to remote buildname
643 # remove for safety; do *not* mkdir first, otherwise we end up with build/build/
644 self.test_ssh.rmdir(build_dir)
645 self.test_ssh.copy(build_dir,recursive=True)
646 # the repo url is taken from arch-rpms-url
647 # with the last step (i386) removed
648 repo_url = self.options.arch_rpms_url
649 for level in [ 'arch' ]:
650 repo_url = os.path.dirname(repo_url)
652 # invoke initvm (drop support for vs)
653 script="lbuild-initvm.sh"
655 # pass the vbuild-nightly options to [lv]test-initvm
656 script_options += " -p %s"%self.options.personality
657 script_options += " -d %s"%self.options.pldistro
658 script_options += " -f %s"%self.options.fcdistro
659 script_options += " -r %s"%repo_url
660 vserver_name = self.vservername
662 vserver_hostname=socket.gethostbyaddr(self.vserverip)[0]
663 script_options += " -n %s"%vserver_hostname
665 print "Cannot reverse lookup %s"%self.vserverip
666 print "This is considered fatal, as this might pollute the test results"
668 create_vserver="%(build_dir)s/%(script)s %(script_options)s %(vserver_name)s"%locals()
669 return self.run_in_host(create_vserver) == 0
672 def plc_install(self):
673 "yum install myplc, noderepo, and the plain bootstrapfs"
675 # workaround for getting pgsql8.2 on centos5
676 if self.options.fcdistro == "centos5":
677 self.run_in_guest("rpm -Uvh http://download.fedora.redhat.com/pub/epel/5/i386/epel-release-5-3.noarch.rpm")
680 if self.options.personality == "linux32":
682 elif self.options.personality == "linux64":
685 raise Exception, "Unsupported personality %r"%self.options.personality
686 nodefamily="%s-%s-%s"%(self.options.pldistro,self.options.fcdistro,arch)
689 pkgs_list.append ("slicerepo-%s"%nodefamily)
690 pkgs_list.append ("myplc")
691 pkgs_list.append ("noderepo-%s"%nodefamily)
692 pkgs_list.append ("nodeimage-%s-plain"%nodefamily)
693 pkgs_string=" ".join(pkgs_list)
694 return self.yum_install (pkgs_list)
697 def mod_python(self):
698 """yum install mod_python, useful on f18 and above so as to avoid broken wsgi"""
699 return self.yum_install ( [ 'mod_python' ] )
702 def plc_configure(self):
704 tmpname='%s.plc-config-tty'%(self.name())
705 fileconf=open(tmpname,'w')
706 for var in [ 'PLC_NAME',
711 'PLC_MAIL_SUPPORT_ADDRESS',
714 # Above line was added for integrating SFA Testing
720 'PLC_RESERVATION_GRANULARITY',
722 'PLC_OMF_XMPP_SERVER',
725 fileconf.write ('e %s\n%s\n'%(var,self.plc_spec[var]))
726 fileconf.write('w\n')
727 fileconf.write('q\n')
729 utils.system('cat %s'%tmpname)
730 self.run_in_guest_piped('cat %s'%tmpname,'plc-config-tty')
731 utils.system('rm %s'%tmpname)
734 # f14 is a bit odd in this respect, although this worked fine in guests up to f18
735 # however using a vplc guest under f20 requires this trick
736 # the symptom is this: service plc start
737 # Starting plc (via systemctl): Failed to get D-Bus connection: \
738 # Failed to connect to socket /org/freedesktop/systemd1/private: Connection refused
739 # weird thing is the doc says f14 uses upstart by default and not systemd
740 # so this sounds kind of harmless
741 def start_service (self,service): return self.start_stop_service (service,'start')
742 def stop_service (self,service): return self.start_stop_service (service,'stop')
744 def start_stop_service (self, service,start_or_stop):
745 "utility to start/stop a service with the special trick for f14"
746 if self.options.fcdistro != 'f14':
747 return self.run_in_guest ("service %s %s"%(service,start_or_stop))==0
749 # patch /sbin/service so it does not reset environment
750 self.run_in_guest ('sed -i -e \\"s,env -i,env,\\" /sbin/service')
751 # this is because our own scripts in turn call service
752 return self.run_in_guest("SYSTEMCTL_SKIP_REDIRECT=true service %s %s"%(service,start_or_stop))==0
756 return self.start_service ('plc')
760 return self.stop_service ('plc')
762 def plcvm_start (self):
763 "start the PLC vserver"
767 def plcvm_stop (self):
768 "stop the PLC vserver"
772 # stores the keys from the config for further use
773 def keys_store(self):
774 "stores test users ssh keys in keys/"
775 for key_spec in self.plc_spec['keys']:
776 TestKey(self,key_spec).store_key()
779 def keys_clean(self):
780 "removes keys cached in keys/"
781 utils.system("rm -rf ./keys")
784 # fetches the ssh keys in the plc's /etc/planetlab and stores them in keys/
785 # for later direct access to the nodes
786 def keys_fetch(self):
787 "gets ssh keys in /etc/planetlab/ and stores them locally in keys/"
789 if not os.path.isdir(dir):
791 vservername=self.vservername
792 vm_root=self.vm_root_in_host()
794 prefix = 'debug_ssh_key'
795 for ext in [ 'pub', 'rsa' ] :
796 src="%(vm_root)s/etc/planetlab/%(prefix)s.%(ext)s"%locals()
797 dst="keys/%(vservername)s-debug.%(ext)s"%locals()
798 if self.test_ssh.fetch(src,dst) != 0: overall=False
802 "create sites with PLCAPI"
803 return self.do_sites()
805 def delete_sites (self):
806 "delete sites with PLCAPI"
807 return self.do_sites(action="delete")
809 def do_sites (self,action="add"):
810 for site_spec in self.plc_spec['sites']:
811 test_site = TestSite (self,site_spec)
812 if (action != "add"):
813 utils.header("Deleting site %s in %s"%(test_site.name(),self.name()))
814 test_site.delete_site()
815 # deleted with the site
816 #test_site.delete_users()
819 utils.header("Creating site %s & users in %s"%(test_site.name(),self.name()))
820 test_site.create_site()
821 test_site.create_users()
824 def delete_all_sites (self):
825 "Delete all sites in PLC, and related objects"
826 print 'auth_root',self.auth_root()
827 sites = self.apiserver.GetSites(self.auth_root(), {}, ['site_id','login_base'])
829 # keep automatic site - otherwise we shoot in our own foot, root_auth is not valid anymore
830 if site['login_base']==self.plc_spec['PLC_SLICE_PREFIX']: continue
831 site_id=site['site_id']
832 print 'Deleting site_id',site_id
833 self.apiserver.DeleteSite(self.auth_root(),site_id)
837 "create nodes with PLCAPI"
838 return self.do_nodes()
839 def delete_nodes (self):
840 "delete nodes with PLCAPI"
841 return self.do_nodes(action="delete")
843 def do_nodes (self,action="add"):
844 for site_spec in self.plc_spec['sites']:
845 test_site = TestSite (self,site_spec)
847 utils.header("Deleting nodes in site %s"%test_site.name())
848 for node_spec in site_spec['nodes']:
849 test_node=TestNode(self,test_site,node_spec)
850 utils.header("Deleting %s"%test_node.name())
851 test_node.delete_node()
853 utils.header("Creating nodes for site %s in %s"%(test_site.name(),self.name()))
854 for node_spec in site_spec['nodes']:
855 utils.pprint('Creating node %s'%node_spec,node_spec)
856 test_node = TestNode (self,test_site,node_spec)
857 test_node.create_node ()
860 def nodegroups (self):
861 "create nodegroups with PLCAPI"
862 return self.do_nodegroups("add")
863 def delete_nodegroups (self):
864 "delete nodegroups with PLCAPI"
865 return self.do_nodegroups("delete")
869 def translate_timestamp (start,grain,timestamp):
870 if timestamp < TestPlc.YEAR: return start+timestamp*grain
871 else: return timestamp
874 def timestamp_printable (timestamp):
875 return time.strftime('%m-%d %H:%M:%S UTC',time.gmtime(timestamp))
878 "create leases (on reservable nodes only, use e.g. run -c default -c resa)"
880 grain=self.apiserver.GetLeaseGranularity(self.auth_root())
881 print 'API answered grain=',grain
882 start=(now/grain)*grain
884 # find out all nodes that are reservable
885 nodes=self.all_reservable_nodenames()
887 utils.header ("No reservable node found - proceeding without leases")
890 # attach them to the leases as specified in plc_specs
891 # this is where the 'leases' field gets interpreted as relative of absolute
892 for lease_spec in self.plc_spec['leases']:
893 # skip the ones that come with a null slice id
894 if not lease_spec['slice']: continue
895 lease_spec['t_from']=TestPlc.translate_timestamp(start,grain,lease_spec['t_from'])
896 lease_spec['t_until']=TestPlc.translate_timestamp(start,grain,lease_spec['t_until'])
897 lease_addition=self.apiserver.AddLeases(self.auth_root(),nodes,
898 lease_spec['slice'],lease_spec['t_from'],lease_spec['t_until'])
899 if lease_addition['errors']:
900 utils.header("Cannot create leases, %s"%lease_addition['errors'])
903 utils.header('Leases on nodes %r for %s from %d (%s) until %d (%s)'%\
904 (nodes,lease_spec['slice'],
905 lease_spec['t_from'],TestPlc.timestamp_printable(lease_spec['t_from']),
906 lease_spec['t_until'],TestPlc.timestamp_printable(lease_spec['t_until'])))
910 def delete_leases (self):
911 "remove all leases in the myplc side"
912 lease_ids= [ l['lease_id'] for l in self.apiserver.GetLeases(self.auth_root())]
913 utils.header("Cleaning leases %r"%lease_ids)
914 self.apiserver.DeleteLeases(self.auth_root(),lease_ids)
917 def list_leases (self):
918 "list all leases known to the myplc"
919 leases = self.apiserver.GetLeases(self.auth_root())
922 current=l['t_until']>=now
923 if self.options.verbose or current:
924 utils.header("%s %s from %s until %s"%(l['hostname'],l['name'],
925 TestPlc.timestamp_printable(l['t_from']),
926 TestPlc.timestamp_printable(l['t_until'])))
929 # create nodegroups if needed, and populate
930 def do_nodegroups (self, action="add"):
931 # 1st pass to scan contents
933 for site_spec in self.plc_spec['sites']:
934 test_site = TestSite (self,site_spec)
935 for node_spec in site_spec['nodes']:
936 test_node=TestNode (self,test_site,node_spec)
937 if node_spec.has_key('nodegroups'):
938 nodegroupnames=node_spec['nodegroups']
939 if isinstance(nodegroupnames,StringTypes):
940 nodegroupnames = [ nodegroupnames ]
941 for nodegroupname in nodegroupnames:
942 if not groups_dict.has_key(nodegroupname):
943 groups_dict[nodegroupname]=[]
944 groups_dict[nodegroupname].append(test_node.name())
945 auth=self.auth_root()
947 for (nodegroupname,group_nodes) in groups_dict.iteritems():
949 print 'nodegroups:','dealing with nodegroup',nodegroupname,'on nodes',group_nodes
950 # first, check if the nodetagtype is here
951 tag_types = self.apiserver.GetTagTypes(auth,{'tagname':nodegroupname})
953 tag_type_id = tag_types[0]['tag_type_id']
955 tag_type_id = self.apiserver.AddTagType(auth,
956 {'tagname':nodegroupname,
957 'description': 'for nodegroup %s'%nodegroupname,
959 print 'located tag (type)',nodegroupname,'as',tag_type_id
961 nodegroups = self.apiserver.GetNodeGroups (auth, {'groupname':nodegroupname})
963 self.apiserver.AddNodeGroup(auth, nodegroupname, tag_type_id, 'yes')
964 print 'created nodegroup',nodegroupname,'from tagname',nodegroupname,'and value','yes'
965 # set node tag on all nodes, value='yes'
966 for nodename in group_nodes:
968 self.apiserver.AddNodeTag(auth, nodename, nodegroupname, "yes")
970 traceback.print_exc()
971 print 'node',nodename,'seems to already have tag',nodegroupname
974 expect_yes = self.apiserver.GetNodeTags(auth,
975 {'hostname':nodename,
976 'tagname':nodegroupname},
977 ['value'])[0]['value']
978 if expect_yes != "yes":
979 print 'Mismatch node tag on node',nodename,'got',expect_yes
982 if not self.options.dry_run:
983 print 'Cannot find tag',nodegroupname,'on node',nodename
987 print 'cleaning nodegroup',nodegroupname
988 self.apiserver.DeleteNodeGroup(auth,nodegroupname)
990 traceback.print_exc()
994 # a list of TestNode objs
995 def all_nodes (self):
997 for site_spec in self.plc_spec['sites']:
998 test_site = TestSite (self,site_spec)
999 for node_spec in site_spec['nodes']:
1000 nodes.append(TestNode (self,test_site,node_spec))
1003 # return a list of tuples (nodename,qemuname)
1004 def all_node_infos (self) :
1006 for site_spec in self.plc_spec['sites']:
1007 node_infos += [ (node_spec['node_fields']['hostname'],node_spec['host_box']) \
1008 for node_spec in site_spec['nodes'] ]
1011 def all_nodenames (self): return [ x[0] for x in self.all_node_infos() ]
1012 def all_reservable_nodenames (self):
1014 for site_spec in self.plc_spec['sites']:
1015 for node_spec in site_spec['nodes']:
1016 node_fields=node_spec['node_fields']
1017 if 'node_type' in node_fields and node_fields['node_type']=='reservable':
1018 res.append(node_fields['hostname'])
1021 # silent_minutes : during the first <silent_minutes> minutes nothing gets printed
1022 def nodes_check_boot_state (self, target_boot_state, timeout_minutes, silent_minutes,period_seconds=15):
1023 if self.options.dry_run:
1027 class CompleterTaskBootState (CompleterTask):
1028 def __init__ (self, test_plc,hostname):
1029 self.test_plc=test_plc
1030 self.hostname=hostname
1031 self.last_boot_state='undef'
1032 def actual_run (self):
1034 node = self.test_plc.apiserver.GetNodes(self.test_plc.auth_root(), [ self.hostname ],
1036 self.last_boot_state = node['boot_state']
1037 return self.last_boot_state == target_boot_state
1041 return "CompleterTaskBootState with node %s"%self.hostname
1042 def failure_message (self):
1043 return "node %s in state %s - expected %s"%(self.hostname,self.last_boot_state,target_boot_state)
1045 timeout = timedelta(minutes=timeout_minutes)
1046 graceout = timedelta(minutes=silent_minutes)
1047 period = timedelta(seconds=period_seconds)
1048 # the nodes that haven't checked yet - start with a full list and shrink over time
1049 utils.header("checking nodes boot state (expected %s)"%target_boot_state)
1050 tasks = [ CompleterTaskBootState (self,hostname) \
1051 for (hostname,_) in self.all_node_infos() ]
1052 return Completer (tasks).run (timeout, graceout, period)
1054 def nodes_booted(self):
1055 return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=28)
1057 def probe_kvm_iptables (self):
1058 (_,kvmbox) = self.all_node_infos()[0]
1059 TestSsh(kvmbox).run("iptables-save")
1063 def check_nodes_ping(self,timeout_seconds=120,period_seconds=10):
1064 class CompleterTaskPingNode (CompleterTask):
1065 def __init__ (self, hostname):
1066 self.hostname=hostname
1067 def run(self,silent):
1068 command="ping -c 1 -w 1 %s >& /dev/null"%self.hostname
1069 return utils.system (command, silent=silent)==0
1070 def failure_message (self):
1071 return "Cannot ping node with name %s"%self.hostname
1072 timeout=timedelta (seconds=timeout_seconds)
1074 period=timedelta (seconds=period_seconds)
1075 node_infos = self.all_node_infos()
1076 tasks = [ CompleterTaskPingNode (h) for (h,_) in node_infos ]
1077 return Completer (tasks).run (timeout, graceout, period)
1079 # ping node before we try to reach ssh, helpful for troubleshooting failing bootCDs
1080 def ping_node (self):
1082 return self.check_nodes_ping ()
1084 def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period_seconds=15):
1086 timeout = timedelta(minutes=timeout_minutes)
1087 graceout = timedelta(minutes=silent_minutes)
1088 period = timedelta(seconds=period_seconds)
1089 vservername=self.vservername
1092 local_key = "keys/%(vservername)s-debug.rsa"%locals()
1095 local_key = "keys/key_admin.rsa"
1096 utils.header("checking ssh access to nodes (expected in %s mode)"%message)
1097 node_infos = self.all_node_infos()
1098 tasks = [ CompleterTaskNodeSsh (nodename, qemuname, local_key, boot_state=message) \
1099 for (nodename,qemuname) in node_infos ]
1100 return Completer (tasks).run (timeout, graceout, period)
1102 def ssh_node_debug(self):
1103 "Tries to ssh into nodes in debug mode with the debug ssh key"
1104 return self.check_nodes_ssh(debug=True,
1105 timeout_minutes=self.ssh_node_debug_timeout,
1106 silent_minutes=self.ssh_node_debug_silent)
1108 def ssh_node_boot(self):
1109 "Tries to ssh into nodes in production mode with the root ssh key"
1110 return self.check_nodes_ssh(debug=False,
1111 timeout_minutes=self.ssh_node_boot_timeout,
1112 silent_minutes=self.ssh_node_boot_silent)
1114 def node_bmlogs(self):
1115 "Checks that there's a non-empty dir. /var/log/bm/raw"
1116 return utils.system(self.actual_command_in_guest("ls /var/log/bm/raw"))==0
1119 def qemu_local_init (self): pass
1121 def bootcd (self): pass
1123 def qemu_local_config (self): pass
1125 def nodestate_reinstall (self): pass
1127 def nodestate_safeboot (self): pass
1129 def nodestate_boot (self): pass
1131 def nodestate_show (self): pass
1133 def qemu_export (self): pass
1135 ### check hooks : invoke scripts from hooks/{node,slice}
1136 def check_hooks_node (self):
1137 return self.locate_first_node().check_hooks()
1138 def check_hooks_sliver (self) :
1139 return self.locate_first_sliver().check_hooks()
1141 def check_hooks (self):
1142 "runs unit tests in the node and slice contexts - see hooks/{node,slice}"
1143 return self.check_hooks_node() and self.check_hooks_sliver()
1146 def do_check_initscripts(self):
1147 class CompleterTaskInitscript (CompleterTask):
1148 def __init__ (self, test_sliver, stamp):
1149 self.test_sliver=test_sliver
1151 def actual_run (self):
1152 return self.test_sliver.check_initscript_stamp (self.stamp)
1154 return "initscript checker for %s"%self.test_sliver.name()
1155 def failure_message (self):
1156 return "initscript stamp %s not found in sliver %s"%(self.stamp,self.test_sliver.name())
1159 for slice_spec in self.plc_spec['slices']:
1160 if not slice_spec.has_key('initscriptstamp'):
1162 stamp=slice_spec['initscriptstamp']
1163 slicename=slice_spec['slice_fields']['name']
1164 for nodename in slice_spec['nodenames']:
1165 print 'nodename',nodename,'slicename',slicename,'stamp',stamp
1166 (site,node) = self.locate_node (nodename)
1167 # xxx - passing the wrong site - probably harmless
1168 test_site = TestSite (self,site)
1169 test_slice = TestSlice (self,test_site,slice_spec)
1170 test_node = TestNode (self,test_site,node)
1171 test_sliver = TestSliver (self, test_node, test_slice)
1172 tasks.append ( CompleterTaskInitscript (test_sliver, stamp))
1173 return Completer (tasks).run (timedelta(minutes=5), timedelta(minutes=4), timedelta(seconds=10))
1175 def check_initscripts(self):
1176 "check that the initscripts have triggered"
1177 return self.do_check_initscripts()
1179 def initscripts (self):
1180 "create initscripts with PLCAPI"
1181 for initscript in self.plc_spec['initscripts']:
1182 utils.pprint('Adding Initscript in plc %s'%self.plc_spec['name'],initscript)
1183 self.apiserver.AddInitScript(self.auth_root(),initscript['initscript_fields'])
1186 def delete_initscripts (self):
1187 "delete initscripts with PLCAPI"
1188 for initscript in self.plc_spec['initscripts']:
1189 initscript_name = initscript['initscript_fields']['name']
1190 print('Attempting to delete %s in plc %s'%(initscript_name,self.plc_spec['name']))
1192 self.apiserver.DeleteInitScript(self.auth_root(),initscript_name)
1193 print initscript_name,'deleted'
1195 print 'deletion went wrong - probably did not exist'
1200 "create slices with PLCAPI"
1201 return self.do_slices(action="add")
1203 def delete_slices (self):
1204 "delete slices with PLCAPI"
1205 return self.do_slices(action="delete")
1207 def fill_slices (self):
1208 "add nodes in slices with PLCAPI"
1209 return self.do_slices(action="fill")
1211 def empty_slices (self):
1212 "remove nodes from slices with PLCAPI"
1213 return self.do_slices(action="empty")
1215 def do_slices (self, action="add"):
1216 for slice in self.plc_spec['slices']:
1217 site_spec = self.locate_site (slice['sitename'])
1218 test_site = TestSite(self,site_spec)
1219 test_slice=TestSlice(self,test_site,slice)
1220 if action == "delete":
1221 test_slice.delete_slice()
1222 elif action=="fill":
1223 test_slice.add_nodes()
1224 elif action=="empty":
1225 test_slice.delete_nodes()
1227 test_slice.create_slice()
1230 @slice_mapper__tasks(20,10,15)
1231 def ssh_slice(self): pass
1232 @slice_mapper__tasks(20,19,15)
1233 def ssh_slice_off (self): pass
1234 @slice_mapper__tasks(2,1,15)
1235 def slice_fs_present(self): pass
1236 @slice_mapper__tasks(2,1,15)
1237 def slice_fs_deleted(self): pass
1239 # use another name so we can exclude/ignore it from the tests on the nightly command line
1240 def ssh_slice_again(self): return self.ssh_slice()
1241 # note that simply doing ssh_slice_again=ssh_slice would kind of work too
1242 # but for some reason the ignore-wrapping thing would not
1245 def ssh_slice_basics(self): pass
1247 def check_vsys_defaults(self): pass
1250 def keys_clear_known_hosts (self): pass
1252 def plcapi_urls (self):
1253 return PlcapiUrlScanner (self.auth_root(),ip=self.vserverip).scan()
1255 def speed_up_slices (self):
1256 "tweak nodemanager settings on all nodes using a conf file"
1257 # create the template on the server-side
1258 template="%s.nodemanager"%self.name()
1259 template_file = open (template,"w")
1260 template_file.write('OPTIONS="-p 30 -r 11 -d"\n')
1261 template_file.close()
1262 in_vm="/var/www/html/PlanetLabConf/nodemanager"
1263 remote="%s/%s"%(self.vm_root_in_host(),in_vm)
1264 self.test_ssh.copy_abs(template,remote)
1266 self.apiserver.AddConfFile (self.auth_root(),
1267 {'dest':'/etc/sysconfig/nodemanager',
1268 'source':'PlanetLabConf/nodemanager',
1269 'postinstall_cmd':'service nm restart',})
1272 def debug_nodemanager (self):
1273 "sets verbose mode for nodemanager, and speeds up cycle even more (needs speed_up_slices first)"
1274 template="%s.nodemanager"%self.name()
1275 template_file = open (template,"w")
1276 template_file.write('OPTIONS="-p 10 -r 6 -v -d"\n')
1277 template_file.close()
1278 in_vm="/var/www/html/PlanetLabConf/nodemanager"
1279 remote="%s/%s"%(self.vm_root_in_host(),in_vm)
1280 self.test_ssh.copy_abs(template,remote)
1284 def qemu_start (self) : pass
1287 def qemu_timestamp (self) : pass
1289 # when a spec refers to a node possibly on another plc
1290 def locate_sliver_obj_cross (self, nodename, slicename, other_plcs):
1291 for plc in [ self ] + other_plcs:
1293 return plc.locate_sliver_obj (nodename, slicename)
1296 raise Exception, "Cannot locate sliver %s@%s among all PLCs"%(nodename,slicename)
1298 # implement this one as a cross step so that we can take advantage of different nodes
1299 # in multi-plcs mode
1300 def cross_check_tcp (self, other_plcs):
1301 "check TCP connectivity between 2 slices (or in loopback if only one is defined)"
1302 if 'tcp_specs' not in self.plc_spec or not self.plc_spec['tcp_specs']:
1303 utils.header ("check_tcp: no/empty config found")
1305 specs = self.plc_spec['tcp_specs']
1310 s_test_sliver = self.locate_sliver_obj_cross (spec['server_node'],spec['server_slice'],other_plcs)
1311 if not s_test_sliver.run_tcp_server(port,timeout=20):
1315 # idem for the client side
1316 c_test_sliver = self.locate_sliver_obj_cross (spec['client_node'],spec['client_slice'],other_plcs)
1317 # use nodename from locatesd sliver, unless 'client_connect' is set
1318 if 'client_connect' in spec:
1319 destination = spec['client_connect']
1321 destination=s_test_sliver.test_node.name()
1322 if not c_test_sliver.run_tcp_client(destination,port):
1326 # painfully enough, we need to allow for some time as netflow might show up last
1327 def check_system_slice (self):
1328 "all nodes: check that a system slice is alive"
1329 # netflow currently not working in the lxc distro
1330 # drl not built at all in the wtx distro
1331 # if we find either of them we're happy
1332 return self.check_netflow() or self.check_drl()
1335 def check_netflow (self): return self._check_system_slice ('netflow')
1336 def check_drl (self): return self._check_system_slice ('drl')
1338 # we have the slices up already here, so it should not take too long
1339 def _check_system_slice (self, slicename, timeout_minutes=5, period_seconds=15):
1340 class CompleterTaskSystemSlice (CompleterTask):
1341 def __init__ (self, test_node, dry_run):
1342 self.test_node=test_node
1343 self.dry_run=dry_run
1344 def actual_run (self):
1345 return self.test_node._check_system_slice (slicename, dry_run=self.dry_run)
1347 return "System slice %s @ %s"%(slicename, self.test_node.name())
1348 def failure_message (self):
1349 return "COULD not find system slice %s @ %s"%(slicename, self.test_node.name())
1350 timeout = timedelta(minutes=timeout_minutes)
1351 silent = timedelta (0)
1352 period = timedelta (seconds=period_seconds)
1353 tasks = [ CompleterTaskSystemSlice (test_node, self.options.dry_run) \
1354 for test_node in self.all_nodes() ]
1355 return Completer (tasks) . run (timeout, silent, period)
1357 def plcsh_stress_test (self):
1358 "runs PLCAPI stress test, that checks Add/Update/Delete on all types - preserves contents"
1359 # install the stress-test in the plc image
1360 location = "/usr/share/plc_api/plcsh_stress_test.py"
1361 remote="%s/%s"%(self.vm_root_in_host(),location)
1362 self.test_ssh.copy_abs("plcsh_stress_test.py",remote)
1364 command += " -- --check"
1365 if self.options.size == 1:
1366 command += " --tiny"
1367 return ( self.run_in_guest(command) == 0)
1369 # populate runs the same utility without slightly different options
1370 # in particular runs with --preserve (dont cleanup) and without --check
1371 # also it gets run twice, once with the --foreign option for creating fake foreign entries
1373 def sfa_install_all (self):
1374 "yum install sfa sfa-plc sfa-sfatables sfa-client"
1375 return self.yum_install ("sfa sfa-plc sfa-sfatables sfa-client")
1377 def sfa_install_core(self):
1379 return self.yum_install ("sfa")
1381 def sfa_install_plc(self):
1382 "yum install sfa-plc"
1383 return self.yum_install("sfa-plc")
1385 def sfa_install_sfatables(self):
1386 "yum install sfa-sfatables"
1387 return self.yum_install ("sfa-sfatables")
1389 # for some very odd reason, this sometimes fails with the following symptom
1390 # # yum install sfa-client
1391 # Setting up Install Process
1393 # Downloading Packages:
1394 # Running rpm_check_debug
1395 # Running Transaction Test
1396 # Transaction Test Succeeded
1397 # Running Transaction
1398 # Transaction couldn't start:
1399 # installing package sfa-client-2.1-7.onelab.2012.05.23.i686 needs 68KB on the / filesystem
1400 # [('installing package sfa-client-2.1-7.onelab.2012.05.23.i686 needs 68KB on the / filesystem', (9, '/', 69632L))]
1401 # even though in the same context I have
1402 # [2012.05.23--f14-32-sfastd1-1-vplc07] / # df -h
1403 # Filesystem Size Used Avail Use% Mounted on
1404 # /dev/hdv1 806G 264G 501G 35% /
1405 # none 16M 36K 16M 1% /tmp
1407 # so as a workaround, we first try yum install, and then invoke rpm on the cached rpm...
1408 def sfa_install_client(self):
1409 "yum install sfa-client"
1410 first_try=self.yum_install("sfa-client")
1411 if first_try: return True
1412 utils.header ("********** Regular yum failed - special workaround in place, 2nd chance")
1413 (code,cached_rpm_path)=utils.output_of(self.actual_command_in_guest('find /var/cache/yum -name sfa-client\*.rpm'))
1414 utils.header("rpm_path=<<%s>>"%rpm_path)
1416 self.run_in_guest("rpm -i %s"%cached_rpm_path)
1417 return self.yum_check_installed ("sfa-client")
1419 def sfa_dbclean(self):
1420 "thoroughly wipes off the SFA database"
1421 return self.run_in_guest("sfaadmin reg nuke")==0 or \
1422 self.run_in_guest("sfa-nuke.py")==0 or \
1423 self.run_in_guest("sfa-nuke-plc.py")==0
1425 def sfa_fsclean(self):
1426 "cleanup /etc/sfa/trusted_roots and /var/lib/sfa"
1427 self.run_in_guest("rm -rf /etc/sfa/trusted_roots /var/lib/sfa/authorities")
1430 def sfa_plcclean(self):
1431 "cleans the PLC entries that were created as a side effect of running the script"
1433 sfa_spec=self.plc_spec['sfa']
1435 for auth_sfa_spec in sfa_spec['auth_sfa_specs']:
1436 login_base=auth_sfa_spec['login_base']
1437 try: self.apiserver.DeleteSite (self.auth_root(),login_base)
1438 except: print "Site %s already absent from PLC db"%login_base
1440 for spec_name in ['pi_spec','user_spec']:
1441 user_spec=auth_sfa_spec[spec_name]
1442 username=user_spec['email']
1443 try: self.apiserver.DeletePerson(self.auth_root(),username)
1445 # this in fact is expected as sites delete their members
1446 #print "User %s already absent from PLC db"%username
1449 print "REMEMBER TO RUN sfa_import AGAIN"
1452 def sfa_uninstall(self):
1453 "uses rpm to uninstall sfa - ignore result"
1454 self.run_in_guest("rpm -e sfa sfa-sfatables sfa-client sfa-plc")
1455 self.run_in_guest("rm -rf /var/lib/sfa")
1456 self.run_in_guest("rm -rf /etc/sfa")
1457 self.run_in_guest("rm -rf /var/log/sfa_access.log /var/log/sfa_import_plc.log /var/log/sfa.daemon")
1459 self.run_in_guest("rpm -e --noscripts sfa-plc")
1462 ### run unit tests for SFA
1463 # NOTE: for some reason on f14/i386, yum install sfa-tests fails for no reason
1464 # Running Transaction
1465 # Transaction couldn't start:
1466 # installing package sfa-tests-1.0-21.onelab.i686 needs 204KB on the / filesystem
1467 # [('installing package sfa-tests-1.0-21.onelab.i686 needs 204KB on the / filesystem', (9, '/', 208896L))]
1468 # no matter how many Gbs are available on the testplc
1469 # could not figure out what's wrong, so...
1470 # if the yum install phase fails, consider the test is successful
1471 # other combinations will eventually run it hopefully
1472 def sfa_utest(self):
1473 "yum install sfa-tests and run SFA unittests"
1474 self.run_in_guest("yum -y install sfa-tests")
1475 # failed to install - forget it
1476 if self.run_in_guest("rpm -q sfa-tests")!=0:
1477 utils.header("WARNING: SFA unit tests failed to install, ignoring")
1479 return self.run_in_guest("/usr/share/sfa/tests/testAll.py")==0
1483 dirname="conf.%s"%self.plc_spec['name']
1484 if not os.path.isdir(dirname):
1485 utils.system("mkdir -p %s"%dirname)
1486 if not os.path.isdir(dirname):
1487 raise Exception,"Cannot create config dir for plc %s"%self.name()
1490 def conffile(self,filename):
1491 return "%s/%s"%(self.confdir(),filename)
1492 def confsubdir(self,dirname,clean,dry_run=False):
1493 subdirname="%s/%s"%(self.confdir(),dirname)
1495 utils.system("rm -rf %s"%subdirname)
1496 if not os.path.isdir(subdirname):
1497 utils.system("mkdir -p %s"%subdirname)
1498 if not dry_run and not os.path.isdir(subdirname):
1499 raise "Cannot create config subdir %s for plc %s"%(dirname,self.name())
1502 def conffile_clean (self,filename):
1503 filename=self.conffile(filename)
1504 return utils.system("rm -rf %s"%filename)==0
1507 def sfa_configure(self):
1508 "run sfa-config-tty"
1509 tmpname=self.conffile("sfa-config-tty")
1510 fileconf=open(tmpname,'w')
1511 for var in [ 'SFA_REGISTRY_ROOT_AUTH',
1512 'SFA_INTERFACE_HRN',
1513 'SFA_REGISTRY_LEVEL1_AUTH',
1514 'SFA_REGISTRY_HOST',
1515 'SFA_AGGREGATE_HOST',
1525 'SFA_GENERIC_FLAVOUR',
1526 'SFA_AGGREGATE_ENABLED',
1528 if self.plc_spec['sfa'].has_key(var):
1529 fileconf.write ('e %s\n%s\n'%(var,self.plc_spec['sfa'][var]))
1530 # the way plc_config handles booleans just sucks..
1533 if self.plc_spec['sfa'][var]: val='true'
1534 fileconf.write ('e %s\n%s\n'%(var,val))
1535 fileconf.write('w\n')
1536 fileconf.write('R\n')
1537 fileconf.write('q\n')
1539 utils.system('cat %s'%tmpname)
1540 self.run_in_guest_piped('cat %s'%tmpname,'sfa-config-tty')
1543 def aggregate_xml_line(self):
1544 port=self.plc_spec['sfa']['neighbours-port']
1545 return '<aggregate addr="%s" hrn="%s" port="%r"/>' % \
1546 (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'],port)
1548 def registry_xml_line(self):
1549 return '<registry addr="%s" hrn="%s" port="12345"/>' % \
1550 (self.vserverip,self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH'])
1553 # a cross step that takes all other plcs in argument
1554 def cross_sfa_configure(self, other_plcs):
1555 "writes aggregates.xml and registries.xml that point to all other PLCs in the test"
1556 # of course with a single plc, other_plcs is an empty list
1559 agg_fname=self.conffile("agg.xml")
1560 file(agg_fname,"w").write("<aggregates>%s</aggregates>\n" % \
1561 " ".join([ plc.aggregate_xml_line() for plc in other_plcs ]))
1562 utils.header ("(Over)wrote %s"%agg_fname)
1563 reg_fname=self.conffile("reg.xml")
1564 file(reg_fname,"w").write("<registries>%s</registries>\n" % \
1565 " ".join([ plc.registry_xml_line() for plc in other_plcs ]))
1566 utils.header ("(Over)wrote %s"%reg_fname)
1567 return self.test_ssh.copy_abs(agg_fname,'/%s/etc/sfa/aggregates.xml'%self.vm_root_in_host())==0 \
1568 and self.test_ssh.copy_abs(reg_fname,'/%s/etc/sfa/registries.xml'%self.vm_root_in_host())==0
1570 def sfa_import(self):
1571 "use sfaadmin to import from plc"
1572 auth=self.plc_spec['sfa']['SFA_REGISTRY_ROOT_AUTH']
1573 return self.run_in_guest('sfaadmin reg import_registry')==0
1575 def sfa_start(self):
1577 return self.start_service('sfa')
1580 def sfi_configure(self):
1581 "Create /root/sfi on the plc side for sfi client configuration"
1582 if self.options.dry_run:
1583 utils.header("DRY RUN - skipping step")
1585 sfa_spec=self.plc_spec['sfa']
1586 # cannot use auth_sfa_mapper to pass dir_name
1587 for slice_spec in self.plc_spec['sfa']['auth_sfa_specs']:
1588 test_slice=TestAuthSfa(self,slice_spec)
1589 dir_basename=os.path.basename(test_slice.sfi_path())
1590 dir_name=self.confsubdir("dot-sfi/%s"%dir_basename,clean=True,dry_run=self.options.dry_run)
1591 test_slice.sfi_configure(dir_name)
1592 # push into the remote /root/sfi area
1593 location = test_slice.sfi_path()
1594 remote="%s/%s"%(self.vm_root_in_host(),location)
1595 self.test_ssh.mkdir(remote,abs=True)
1596 # need to strip last level or remote otherwise we get an extra dir level
1597 self.test_ssh.copy_abs(dir_name, os.path.dirname(remote), recursive=True)
1601 def sfi_clean (self):
1602 "clean up /root/sfi on the plc side"
1603 self.run_in_guest("rm -rf /root/sfi")
1607 def sfa_register_site (self): pass
1609 def sfa_register_pi (self): pass
1611 def sfa_register_user(self): pass
1613 def sfa_update_user(self): pass
1615 def sfa_register_slice(self): pass
1617 def sfa_renew_slice(self): pass
1619 def sfa_discover(self): pass
1621 def sfa_create_slice(self): pass
1623 def sfa_check_slice_plc(self): pass
1625 def sfa_update_slice(self): pass
1627 def sfa_remove_user_from_slice(self): pass
1629 def sfa_insert_user_in_slice(self): pass
1631 def sfi_list(self): pass
1633 def sfi_show_site(self): pass
1635 def sfi_show_slice(self): pass
1637 def sfi_show_slice_researchers(self): pass
1639 def ssh_slice_sfa(self): pass
1641 def sfa_delete_user(self): pass
1643 def sfa_delete_slice(self): pass
1647 return self.stop_service ('sfa')
1649 def populate (self):
1650 "creates random entries in the PLCAPI"
1651 # install the stress-test in the plc image
1652 location = "/usr/share/plc_api/plcsh_stress_test.py"
1653 remote="%s/%s"%(self.vm_root_in_host(),location)
1654 self.test_ssh.copy_abs("plcsh_stress_test.py",remote)
1656 command += " -- --preserve --short-names"
1657 local = (self.run_in_guest(command) == 0);
1658 # second run with --foreign
1659 command += ' --foreign'
1660 remote = (self.run_in_guest(command) == 0);
1661 return ( local and remote)
1663 def gather_logs (self):
1664 "gets all possible logs from plc's/qemu node's/slice's for future reference"
1665 # (1.a) get the plc's /var/log/ and store it locally in logs/myplc.var-log.<plcname>/*
1666 # (1.b) get the plc's /var/lib/pgsql/data/pg_log/ -> logs/myplc.pgsql-log.<plcname>/*
1667 # (1.c) get the plc's /root/sfi -> logs/sfi.<plcname>/
1668 # (2) get all the nodes qemu log and store it as logs/node.qemu.<node>.log
1669 # (3) get the nodes /var/log and store is as logs/node.var-log.<node>/*
1670 # (4) as far as possible get the slice's /var/log as logs/sliver.var-log.<sliver>/*
1672 print "-------------------- TestPlc.gather_logs : PLC's /var/log"
1673 self.gather_var_logs ()
1675 print "-------------------- TestPlc.gather_logs : PLC's /var/lib/psql/data/pg_log/"
1676 self.gather_pgsql_logs ()
1678 print "-------------------- TestPlc.gather_logs : PLC's /root/sfi/"
1679 self.gather_root_sfi ()
1681 print "-------------------- TestPlc.gather_logs : nodes's QEMU logs"
1682 for site_spec in self.plc_spec['sites']:
1683 test_site = TestSite (self,site_spec)
1684 for node_spec in site_spec['nodes']:
1685 test_node=TestNode(self,test_site,node_spec)
1686 test_node.gather_qemu_logs()
1688 print "-------------------- TestPlc.gather_logs : nodes's /var/log"
1689 self.gather_nodes_var_logs()
1691 print "-------------------- TestPlc.gather_logs : sample sliver's /var/log"
1692 self.gather_slivers_var_logs()
1695 def gather_slivers_var_logs(self):
1696 for test_sliver in self.all_sliver_objs():
1697 remote = test_sliver.tar_var_logs()
1698 utils.system("mkdir -p logs/sliver.var-log.%s"%test_sliver.name())
1699 command = remote + " | tar -C logs/sliver.var-log.%s -xf -"%test_sliver.name()
1700 utils.system(command)
1703 def gather_var_logs (self):
1704 utils.system("mkdir -p logs/myplc.var-log.%s"%self.name())
1705 to_plc = self.actual_command_in_guest("tar -C /var/log/ -cf - .")
1706 command = to_plc + "| tar -C logs/myplc.var-log.%s -xf -"%self.name()
1707 utils.system(command)
1708 command = "chmod a+r,a+x logs/myplc.var-log.%s/httpd"%self.name()
1709 utils.system(command)
1711 def gather_pgsql_logs (self):
1712 utils.system("mkdir -p logs/myplc.pgsql-log.%s"%self.name())
1713 to_plc = self.actual_command_in_guest("tar -C /var/lib/pgsql/data/pg_log/ -cf - .")
1714 command = to_plc + "| tar -C logs/myplc.pgsql-log.%s -xf -"%self.name()
1715 utils.system(command)
1717 def gather_root_sfi (self):
1718 utils.system("mkdir -p logs/sfi.%s"%self.name())
1719 to_plc = self.actual_command_in_guest("tar -C /root/sfi/ -cf - .")
1720 command = to_plc + "| tar -C logs/sfi.%s -xf -"%self.name()
1721 utils.system(command)
1723 def gather_nodes_var_logs (self):
1724 for site_spec in self.plc_spec['sites']:
1725 test_site = TestSite (self,site_spec)
1726 for node_spec in site_spec['nodes']:
1727 test_node=TestNode(self,test_site,node_spec)
1728 test_ssh = TestSsh (test_node.name(),key="keys/key_admin.rsa")
1729 command = test_ssh.actual_command("tar -C /var/log -cf - .")
1730 command = command + "| tar -C logs/node.var-log.%s -xf -"%test_node.name()
1731 utils.system("mkdir -p logs/node.var-log.%s"%test_node.name())
1732 utils.system(command)
1735 # returns the filename to use for sql dump/restore, using options.dbname if set
1736 def dbfile (self, database):
1737 # uses options.dbname if it is found
1739 name=self.options.dbname
1740 if not isinstance(name,StringTypes):
1746 return "/root/%s-%s.sql"%(database,name)
1748 def plc_db_dump(self):
1749 'dump the planetlab5 DB in /root in the PLC - filename has time'
1750 dump=self.dbfile("planetab5")
1751 self.run_in_guest('pg_dump -U pgsqluser planetlab5 -f '+ dump)
1752 utils.header('Dumped planetlab5 database in %s'%dump)
1755 def plc_db_restore(self):
1756 'restore the planetlab5 DB - looks broken, but run -n might help'
1757 dump=self.dbfile("planetab5")
1758 ##stop httpd service
1759 self.run_in_guest('service httpd stop')
1760 # xxx - need another wrapper
1761 self.run_in_guest_piped('echo drop database planetlab5','psql --user=pgsqluser template1')
1762 self.run_in_guest('createdb -U postgres --encoding=UNICODE --owner=pgsqluser planetlab5')
1763 self.run_in_guest('psql -U pgsqluser planetlab5 -f '+dump)
1764 ##starting httpd service
1765 self.run_in_guest('service httpd start')
1767 utils.header('Database restored from ' + dump)
1770 def create_ignore_steps ():
1771 for step in TestPlc.default_steps + TestPlc.other_steps:
1772 # default step can have a plc qualifier
1773 if '@' in step: (step,qualifier)=step.split('@')
1774 # or be defined as forced or ignored by default
1775 for keyword in ['_ignore','_force']:
1776 if step.endswith (keyword): step=step.replace(keyword,'')
1777 if step == SEP or step == SEPSFA : continue
1778 method=getattr(TestPlc,step)
1780 wrapped=ignore_result(method)
1781 # wrapped.__doc__ = method.__doc__ + " (run in ignore-result mode)"
1782 setattr(TestPlc, name, wrapped)
1785 # def ssh_slice_again_ignore (self): pass
1787 # def check_initscripts_ignore (self): pass
1789 def standby_1_through_20(self):
1790 """convenience function to wait for a specified number of minutes"""
1793 def standby_1(): pass
1795 def standby_2(): pass
1797 def standby_3(): pass
1799 def standby_4(): pass
1801 def standby_5(): pass
1803 def standby_6(): pass
1805 def standby_7(): pass
1807 def standby_8(): pass
1809 def standby_9(): pass
1811 def standby_10(): pass
1813 def standby_11(): pass
1815 def standby_12(): pass
1817 def standby_13(): pass
1819 def standby_14(): pass
1821 def standby_15(): pass
1823 def standby_16(): pass
1825 def standby_17(): pass
1827 def standby_18(): pass
1829 def standby_19(): pass
1831 def standby_20(): pass
1833 # convenience for debugging the test logic
1834 def yes (self): return True
1835 def no (self): return False