X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=system%2FTestPlc.py;h=463e79410eda680e1053ffe0d88a71b1f15525a4;hb=1c87dd36c45ad0896491f51aa8bdf63f8668da25;hp=490ddd7b012c05cc78064fd1f4e2d496c5e4551b;hpb=c05187b73f49c2ec7361f20a062bb17e33dbe5ed;p=tests.git diff --git a/system/TestPlc.py b/system/TestPlc.py index 490ddd7..463e794 100644 --- a/system/TestPlc.py +++ b/system/TestPlc.py @@ -38,14 +38,11 @@ def standby_generic (func): return actual def node_mapper (method): - def actual(self): + def actual(self,*args, **kwds): overall=True node_method = TestNode.__dict__[method.__name__] - for site_spec in self.plc_spec['sites']: - test_site = TestSite (self,site_spec) - for node_spec in site_spec['nodes']: - test_node = TestNode (self,test_site,node_spec) - if not node_method(test_node): overall=False + for test_node in self.all_nodes(): + if not node_method(test_node, *args, **kwds): overall=False return overall # restore the doc text actual.__doc__=method.__doc__ @@ -101,7 +98,7 @@ class TestPlc: 'ssh_node_debug@1', 'plcsh_stress_test@1', SEP, 'ssh_node_boot@1', 'ssh_slice', 'check_initscripts', SEP, 'ssh_slice_sfa@1', 'sfa_delete_slice@1', 'sfa_delete_user@1', SEPSFA, - 'check_tcp', SEP, + 'check_tcp', 'check_sys_slice', SEP, 'force_gather_logs', SEP, ] other_steps = [ @@ -182,30 +179,32 @@ class TestPlc: #command gets run in the plc's vm def host_to_guest(self,command): if self.options.plcs_use_lxc: - # XXX TODO-lxc how to run a command in the plc context from an lxc-based host - return "TODO-lxc TestPlc.host_to_guest" + return "ssh -o StrictHostKeyChecking=no %s %s"%(self.vserverip,command) else: return "vserver %s exec %s"%(self.vservername,command) - def vm_root_in_guest(self): + def vm_root_in_host(self): + if self.options.plcs_use_lxc: + return "/var/lib/lxc/%s/rootfs/"%(self.vservername) + else: + return "/vservers/%s"%(self.vservername) + + def vm_timestamp_path (self): if self.options.plcs_use_lxc: - # TODO-lxc - return "TODO TestPlc.vm_root_in_guest" + return "/var/lib/lxc/%s/%s.timestamp"%(self.vservername,self.vservername) else: - return "/vservers/%s"%self.vservername + return "/vservers/%s.timestamp"%(self.vservername) #start/stop the vserver def start_guest_in_host(self): if self.options.plcs_use_lxc: - # XXX TODO-lxc how to run a command in the plc context from an lxc-based host - return "TODO-lxc TestPlc.start_guest_in_host" + return "lxc-start --daemon --name=%s"%(self.vservername) else: return "vserver %s start"%(self.vservername) def stop_guest_in_host(self): if self.options.plcs_use_lxc: - # XXX TODO-lxc how to run a command in the plc context from an lxc-based host - return "TODO-lxc TestPlc.stop_guest_in_host" + return "lxc-stop --name=%s"%(self.vservername) else: return "vserver %s stop"%(self.vservername) @@ -368,8 +367,13 @@ 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) @@ -486,19 +490,20 @@ class TestPlc: 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 - stamp_path="%s.timestamp"%self.vm_root_in_guest() + stamp_path=self.vm_timestamp_path () + stamp_dir = os.path.dirname (stamp_path) + utils.system(self.test_ssh.actual_command("mkdir -p %s"%stamp_dir)) return utils.system(self.test_ssh.actual_command("echo %d > %s"%(now,stamp_path)))==0 # this is called inconditionnally at the beginning of the test sequence # just in case this is a rerun, so if the vm is not running it's fine def vs_delete(self): "vserver delete the test myplc" - stamp_path="%s.timestamp"%self.vm_root_in_guest() + stamp_path=self.vm_timestamp_path() self.run_in_host("rm -f %s"%stamp_path) if self.options.plcs_use_lxc: - # TODO-lxc : how to trash a VM altogether and the related timestamp as well - # might make sense to test that this has been done - unlike for vs - print "TODO TestPlc.vs_delete" + self.run_in_host("lxc-stop --name %s"%self.vservername) + self.run_in_host("lxc-destroy --name %s"%self.vservername) return True else: self.run_in_host("vserver --silent %s delete"%self.vservername) @@ -534,7 +539,6 @@ class TestPlc: test_env_options += " -d %s"%self.options.pldistro test_env_options += " -f %s"%self.options.fcdistro if self.options.plcs_use_lxc: - # TODO-lxc : might need some tweaks script="vtest-init-lxc.sh" else: script="vtest-init-vserver.sh" @@ -571,7 +575,7 @@ class TestPlc: pkgs_list.append ("slicerepo-%s"%nodefamily) pkgs_list.append ("myplc") pkgs_list.append ("noderepo-%s"%nodefamily) - pkgs_list.append ("bootstrapfs-%s-plain"%nodefamily) + pkgs_list.append ("nodeimage-%s-plain"%nodefamily) pkgs_string=" ".join(pkgs_list) return self.yum_install (pkgs_list) @@ -647,7 +651,7 @@ class TestPlc: if not os.path.isdir(dir): os.mkdir(dir) vservername=self.vservername - vm_root=self.vm_root_in_guest() + vm_root=self.vm_root_in_host() overall=True prefix = 'debug_ssh_key' for ext in [ 'pub', 'rsa' ] : @@ -846,6 +850,15 @@ class TestPlc: overall=False return overall + # a list of TestNode objs + def all_nodes (self): + nodes=[] + for site_spec in self.plc_spec['sites']: + test_site = TestSite (self,site_spec) + for node_spec in site_spec['nodes']: + nodes.append(TestNode (self,test_site,node_spec)) + return nodes + # return a list of tuples (nodename,qemuname) def all_node_infos (self) : node_infos = [] @@ -911,7 +924,7 @@ class TestPlc: return True def nodes_booted(self): - return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=20) + return self.nodes_check_boot_state('boot',timeout_minutes=30,silent_minutes=28) def check_nodes_ssh(self,debug,timeout_minutes,silent_minutes,period=15): # compute timeout @@ -960,11 +973,11 @@ class TestPlc: def ssh_node_debug(self): "Tries to ssh into nodes in debug mode with the debug ssh key" - return self.check_nodes_ssh(debug=True,timeout_minutes=10,silent_minutes=5) + return self.check_nodes_ssh(debug=True,timeout_minutes=10,silent_minutes=8) def ssh_node_boot(self): "Tries to ssh into nodes in production mode with the root ssh key" - return self.check_nodes_ssh(debug=False,timeout_minutes=40,silent_minutes=15) + return self.check_nodes_ssh(debug=False,timeout_minutes=40,silent_minutes=38) @node_mapper def qemu_local_init (self): @@ -1111,12 +1124,39 @@ class TestPlc: overall=False return overall + # painfully enough, we need to allow for some time as netflow might show up last + def check_sys_slice (self): + "all nodes: check that a system slice is alive" +# would probably make more sense to check for netflow, +# but that one is currently not working in the lxc distro +# return self.check_systemslice ('netflow') + return self.check_systemslice ('drl') + + # we have the slices up already here, so it should not take too long + def check_systemslice (self, slicename, timeout_minutes=5, period=15): + timeout = datetime.datetime.now()+datetime.timedelta(minutes=timeout_minutes) + test_nodes=self.all_nodes() + while test_nodes: + for test_node in test_nodes: + if test_node.check_systemslice (slicename,dry_run=self.options.dry_run): + utils.header ("ok") + test_nodes.remove(test_node) + else: + print '.', + if not test_nodes: + return True + if datetime.datetime.now () > timeout: + for test_node in test_nodes: + utils.header ("can't find system slice %s in %s"%(slicename,test_node.name())) + return False + time.sleep(period) + return True + def plcsh_stress_test (self): "runs PLCAPI stress test, that checks Add/Update/Delete on all types - preserves contents" # install the stress-test in the plc image location = "/usr/share/plc_api/plcsh_stress_test.py" - # TODO-lxc - remote="%s/%s"%(self.vm_root_in_guest(),location) + remote="%s/%s"%(self.vm_root_in_host(),location) self.test_ssh.copy_abs("plcsh_stress_test.py",remote) command = location command += " -- --check" @@ -1285,8 +1325,8 @@ class TestPlc: file(reg_fname,"w").write("%s\n" % \ " ".join([ plc.registry_xml_line() for plc in other_plcs ])) utils.header ("(Over)wrote %s"%reg_fname) - return self.test_ssh.copy_abs(agg_fname,'/%s/etc/sfa/aggregates.xml'%self.vm_root_in_guest())==0 \ - and self.test_ssh.copy_abs(reg_fname,'/%s/etc/sfa/registries.xml'%self.vm_root_in_guest())==0 + return self.test_ssh.copy_abs(agg_fname,'/%s/etc/sfa/aggregates.xml'%self.vm_root_in_host())==0 \ + and self.test_ssh.copy_abs(reg_fname,'/%s/etc/sfa/registries.xml'%self.vm_root_in_host())==0 def sfa_import(self): "sfa-import-plc" @@ -1316,7 +1356,7 @@ class TestPlc: test_slice.sfi_config(dir_name) # push into the remote /root/sfi area location = test_slice.sfi_path() - remote="%s/%s"%(self.vm_root_in_guest(),location) + remote="%s/%s"%(self.vm_root_in_host(),location) self.test_ssh.mkdir(remote,abs=True) # need to strip last level or remote otherwise we get an extra dir level self.test_ssh.copy_abs(dir_name, os.path.dirname(remote), recursive=True) @@ -1391,7 +1431,7 @@ class TestPlc: "creates random entries in the PLCAPI" # install the stress-test in the plc image location = "/usr/share/plc_api/plcsh_stress_test.py" - remote="%s/%s"%(self.vm_root_in_guest(),location) + remote="%s/%s"%(self.vm_root_in_host(),location) self.test_ssh.copy_abs("plcsh_stress_test.py",remote) command = location command += " -- --preserve --short-names" @@ -1405,6 +1445,7 @@ 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./* @@ -1414,6 +1455,9 @@ 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']: @@ -1451,6 +1495,12 @@ 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)