X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=monitor%2Fbootman.py;h=96bc740c2f36b94fd0474eaa8c3bbbe1ce129c7d;hb=cccbcf3069da9b6fe5e5817b5370bf1339bf42eb;hp=bfb295c102d8b6615c01af09122881c4b8dffe86;hpb=16b7bc48471b0695e72320fbcf761db695af032e;p=monitor.git diff --git a/monitor/bootman.py b/monitor/bootman.py index bfb295c..96bc740 100755 --- a/monitor/bootman.py +++ b/monitor/bootman.py @@ -42,6 +42,8 @@ api = plc.getAuthAPI() fb = None +class ExceptionDoubleSSHError(Exception): pass + class NodeConnection: def __init__(self, connection, node, config): self.node = node @@ -66,18 +68,34 @@ class NodeConnection: return "unknown" def get_dmesg(self): + t_stamp = time.strftime("%Y-%m-%d-%H:%M") self.c.modules.os.system("dmesg > /var/log/dmesg.bm.log") - download(self.c, "/var/log/dmesg.bm.log", "log/dmesg.%s.log" % self.node) - log = open("log/dmesg.%s.log" % self.node, 'r') + download(self.c, "/var/log/dmesg.bm.log", "%s/history/%s-dmesg.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, t_stamp, self.node)) + os.system("cp %s/history/%s-dmesg.%s.log %s/dmesg.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, t_stamp, self.node, config.MONITOR_BOOTMANAGER_LOG, self.node)) + log = open("%s/dmesg.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, self.node), 'r') return log def get_bootmanager_log(self): - download(self.c, "/tmp/bm.log", "log/bm.%s.log.gz" % self.node) - #os.system("zcat log/bm.%s.log.gz > log/bm.%s.log" % (self.node, self.node)) - os.system("cp log/bm.%s.log.gz log/bm.%s.log" % (self.node, self.node)) - log = open("log/bm.%s.log" % self.node, 'r') + t_stamp = time.strftime("%Y-%m-%d-%H:%M") + download(self.c, "/tmp/bm.log", "%s/history/%s-bm.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, t_stamp, self.node)) + os.system("cp %s/history/%s-bm.%s.log %s/bm.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, t_stamp, self.node, config.MONITOR_BOOTMANAGER_LOG, self.node)) + log = open("%s/bm.%s.log" % (config.MONITOR_BOOTMANAGER_LOG, self.node), 'r') return log + +# def get_dmesg(self): +# self.c.modules.os.system("dmesg > /var/log/dmesg.bm.log") +# download(self.c, "/var/log/dmesg.bm.log", "log/dmesg.%s.log" % self.node) +# log = open("log/dmesg.%s.log" % self.node, 'r') +# return log +# +# def get_bootmanager_log(self): +# download(self.c, "/tmp/bm.log", "log/bm.%s.log.gz" % self.node) +# #os.system("zcat log/bm.%s.log.gz > log/bm.%s.log" % (self.node, self.node)) +# os.system("cp log/bm.%s.log.gz log/bm.%s.log" % (self.node, self.node)) +# log = open("log/bm.%s.log" % self.node, 'r') +# return log + def dump_plconf_file(self): c = self.c self.c.modules.sys.path.append("/tmp/source/") @@ -104,6 +122,28 @@ class NodeConnection: else: print " Unable to read Node Configuration" + def fsck_repair_node(self): + c = self.c + self.c.modules.sys.path.append("/tmp/source/") + self.c.modules.os.chdir('/tmp/source') + # TODO: restart + # TODO: set boot state to node's actually boot state. + # could be 'boot' or 'safeboot' + self.c.modules.os.chdir('/tmp/source') + if self.c.modules.os.path.exists('/tmp/BM_RUNNING'): + print "Running MANUAL FSCK already... try again soon." + else: + print "Running MANUAL fsck on %s" % self.node + cmd = "( touch /tmp/BM_RUNNING ; " + \ + " fsck -v -f -y /dev/planetlab/root &> out.fsck ; " + \ + " fsck -v -f -y /dev/planetlab/vserver >> out.fsck 2>&1 ; " + \ + " python ./BootManager.py %s &> server.log < /dev/null ; " + \ + " rm -f /tmp/BM_RUNNING " + \ + ") &" + cmd = cmd % self.get_nodestate() + self.c.modules.os.system(cmd) + #self.restart_bootmanager('boot') + pass def compare_and_repair_nodekeys(self): c = self.c @@ -155,6 +195,16 @@ class NodeConnection: def set_nodestate(self, state='boot'): return api.UpdateNode(self.node, {'boot_state' : state}) + def get_nodestate(self): + try: + return api.GetNodes(self.node, ['boot_state'])[0]['boot_state'] + except: + traceback.print_exc() + # NOTE: use last cached value from plc + fbnode = FindbadNodeRecord.get_latest_by(hostname=self.node).to_dict() + return fbnode['plc_node_stats']['boot_state'] + + def restart_node(self, state='boot'): api.UpdateNode(self.node, {'boot_state' : state}) @@ -230,7 +280,8 @@ class PlanetLabSession: return # COPY Rpyc files to host - cmd = "rsync -qv -az -e ssh %(monitordir)s/Rpyc/ %(user)s@%(hostname)s:Rpyc 2> /dev/null" % args + #cmd = "rsync -vvv -az -e ssh %(monitordir)s/Rpyc/ %(user)s@%(hostname)s:Rpyc 2> /dev/null" % args + cmd = """rsync -vvv -az -e "ssh -o BatchMode=yes" %(monitordir)s/Rpyc/ %(user)s@%(hostname)s:Rpyc""" % args if self.verbose: print cmd print cmd # TODO: Add timeout @@ -243,12 +294,14 @@ class PlanetLabSession: print "\tUNKNOWN SSH KEY FOR %s; making an exception" % self.node #print "MAKE EXPLICIT EXCEPTION FOR %s" % self.node k = SSHKnownHosts(); k.updateDirect(self.node); k.write(); del k + print "trying: ", cmd + print [ "%s=%s" % (a, os.environ[a]) for a in filter(lambda x: 'SSH' in x, os.environ.keys()) ] ret = localos.system(cmd, timeout) print ret if ret != 0: print "\tFAILED TWICE" #sys.exit(1) - raise Exception("Failed twice trying to login with updated ssh host key") + raise ExceptionDoubleSSHError("Failed twice trying to login with updated ssh host key") t1 = time.time() # KILL any already running servers. @@ -341,14 +394,16 @@ class DebugInterface: print traceback.print_exc() return False + msg = "ERROR setting up session for %s" % self.hostname try: if config == None: self.session = PlanetLabSession(self.hostname, False, True) else: self.session = PlanetLabSession(self.hostname, config.nosetup, config.verbose) - except Exception, e: - msg = "ERROR setting up session for %s" % self.hostname + except ExceptionDoubleSSHError, e: print msg + return False + except Exception, e: traceback.print_exc() email_exception(msg) return False @@ -361,6 +416,10 @@ class DebugInterface: try: time.sleep(self.session.timeout*5) conn = self.session.get_connection(config) + except EOFError: + # failed twice... no need to report this really, it's just in a + # weird state... + return False except: traceback.print_exc() email_exception(self.hostname) @@ -399,7 +458,7 @@ class DebugInterface: ]: sequences.update({n : "restart_bootmanager_boot"}) - # conn.restart_bootmanager('rins') + # conn.restart_bootmanager('reinstall') for n in [ "bminit-cfg-auth-getplc-installinit-validate-exception-modulefail-update-debug-done", "bminit-cfg-auth-getplc-update-installinit-validate-exception-modulefail-update-debug-done", "bminit-cfg-auth-getplc-installinit-validate-bmexceptmount-exception-noinstall-update-debug-done", @@ -422,13 +481,20 @@ class DebugInterface: "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-implementerror-update-debug-done", "bminit-cfg-auth-getplc-installinit-validate-exception-bmexceptmount-exception-noinstall-update-debug-done", "bminit-cfg-auth-getplc-update-installinit-validate-exception-bmexceptmount-exception-noinstall-update-debug-done", + "bminit-cfg-auth-getplc-update-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-validate-bmexceptvgscan-done", + "bminit-cfg-auth-getplc-update-installinit-validate-exception-noinstall-update-debug-validate-done", + "bminit-cfg-auth-getplc-installinit-validate-bmexceptvgscan-exception-noinstall-update-debug-validate-bmexceptvgscan-done", ]: sequences.update({n : "restart_bootmanager_rins"}) # repair_node_keys - sequences.update({"bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-done": "repair_node_keys"}) + for n in ["bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-validate-exception-done", + "bminit-cfg-auth-bootcheckfail-authfail-exception-update-bootupdatefail-authfail-debug-done", + "bminit-cfg-auth-bootcheckfail-authfail-exception-update-debug-validate-exception-done", + ]: + sequences.update({n: "repair_node_keys"}) - # conn.restart_node('rins') + # conn.restart_node('reinstall') for n in ["bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done", "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-disk-update4-exception-chrootfail-update-debug-done", "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-exception-chrootfail-update-debug-done", @@ -451,17 +517,29 @@ class DebugInterface: "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nospace-update-debug-done", "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-exception-downloadfail-update-debug-done", "bminit-cfg-auth-getplc-update-installinit-validate-implementerror-update-debug-done", + "bminit-cfg-auth-getplc-exception-update-bootupdatefail-debug-done", ]: sequences.update({n: "restart_node_boot"}) + # fsck_repair + for n in ["bminit-cfg-auth-getplc-update-installinit-validate-fsckabort-exception-fsckfail-bmexceptmount-exception-noinstall-update-debug-validate-fsckabort-exception-fsckfail-bmexceptmount-done", + "bminit-cfg-auth-getplc-installinit-validate-exception-fsckfail-exception-noinstall-update-debug-validate-exception-fsckfail-done", + "bminit-cfg-auth-getplc-update-installinit-validate-exception-fsckfail-exception-noinstall-update-debug-validate-exception-fsckfail-done" + ]: + sequences.update({n : "fsck_repair"}) + # update_node_config_email for n in ["bminit-cfg-exception-nocfg-update-bootupdatefail-nonode-debug-done", "bminit-cfg-exception-update-bootupdatefail-nonode-debug-done", + "bminit-cfg-exception-update-bootupdatefail-nonode-debug-validate-exception-done", + "bminit-cfg-exception-nocfg-update-bootupdatefail-nonode-debug-validate-exception-done", "bminit-cfg-auth-bootcheckfail-nonode-exception-update-bootupdatefail-nonode-debug-done", + "bminit-cfg-exception-noconfig-update-debug-validate-exception-done", ]: sequences.update({n : "update_node_config_email"}) for n in [ "bminit-cfg-exception-nodehostname-update-debug-done", + "bminit-cfg-update-exception-nodehostname-update-debug-validate-exception-done", "bminit-cfg-update-exception-nodehostname-update-debug-done", ]: sequences.update({n : "nodenetwork_email"}) @@ -573,12 +651,15 @@ class DebugInterface: ('protoerror' , 'XML RPC protocol error'), ('nodehostname' , 'Configured node hostname does not resolve'), ('implementerror', 'Implementation Error'), - ('readonlyfs' , '[Errno 30] Read-only file system'), - ('baddisk' , "IOError: [Errno 13] Permission denied: '/tmp/mnt/sysimg//vservers/\w+/etc/hosts'"), + ('fsckabort' , 'is mounted. e2fsck: Cannot continue, aborting'), + ('fsckfail' , 'Running e2fsck -v -p /dev/planetlab/root failed'), + ('readonlyfs' , '\[Errno 30\] Read-only file system'), + ('baddisk' , "IOError: \[Errno 13\] Permission denied: '/tmp/mnt/sysimg//vservers/\w+/etc/hosts'"), ('noinstall' , 'notinstalled'), ('bziperror' , 'bzip2: Data integrity error when decompressing.'), ('noblockdev' , "No block devices detected."), ('dnserror' , 'Name or service not known'), + ('noconfig' , "Unable to find and read a node configuration file"), ('downloadfail' , 'Unable to download main tarball /boot/bootstrapfs-planetlab-i386.tar.bz2 from server.'), ('disktoosmall' , 'The total usable disk size of all disks is insufficient to be usable as a PlanetLab node.'), ('hardwarerequirefail' , 'Hardware requirements not met'), @@ -612,8 +693,12 @@ class DebugInterface: return sequence - def restore(sitehist, hostname, config=None, forced_action=None): + ret = restore_basic(sitehist, hostname, config, forced_action) + session.flush() + return ret + +def restore_basic(sitehist, hostname, config=None, forced_action=None): # NOTE: Nothing works if the bootcd is REALLY old. # So, this is the first step. @@ -624,27 +709,19 @@ def restore(sitehist, hostname, config=None, forced_action=None): if fbnode['observed_category'] == "OLDBOOTCD": print "\t...Notify owner to update BootImage!!!" - if not found_within(recent_actions, 'newbootcd_notice', 3): + if not found_within(recent_actions, 'newbootcd_notice', 3.5): sitehist.sendMessage('newbootcd_notice', hostname=hostname) print "\tDisabling %s due to out-of-date BootImage" % hostname - api.UpdateNode(hostname, {'boot_state' : 'disable'}) + api.UpdateNode(hostname, {'boot_state' : 'disabled'}) # NOTE: nothing else is possible. return True debugnode = DebugInterface(hostname) conn = debugnode.getConnection() - #print "conn: %s" % conn - #print "trying to use conn after returning it." - #print conn.c.modules.sys.path - #print conn.c.modules.os.path.exists('/tmp/source') if type(conn) == type(False): return False - #if forced_action == "reboot": - # conn.restart_node('rins') - # return True - boot_state = conn.get_boot_state() if boot_state != "debug": print "... %s in %s state: skipping..." % (hostname , boot_state) @@ -674,11 +751,12 @@ def restore(sitehist, hostname, config=None, forced_action=None): print "...Should investigate. Skipping node." # TODO: send message related to these errors. - if not found_within(recent_actions, 'newbootcd_notice', 3): + if not found_within(recent_actions, 'baddisk_notice', 7): + print "baddisk_notice not found recently" log=conn.get_dmesg().read() sitehist.sendMessage('baddisk_notice', hostname=hostname, log=log) - conn.set_nodestate('disable') + conn.set_nodestate('disabled') return False @@ -715,6 +793,8 @@ def restore(sitehist, hostname, config=None, forced_action=None): args['sequence'] = s args['bmlog'] = conn.get_bootmanager_log().read() args['viart'] = False + args['saveact'] = True + args['ccemail'] = True sitehist.sendMessage('unknownsequence_notice', **args) @@ -731,16 +811,18 @@ def restore(sitehist, hostname, config=None, forced_action=None): conn.restart_bootmanager('boot') elif sequences[s] == "restart_bootmanager_rins": print "...Restarting BootManager.py on %s "%hostname - conn.restart_bootmanager('rins') + conn.restart_bootmanager('reinstall') elif sequences[s] == "restart_node_rins": - conn.restart_node('rins') + conn.restart_node('reinstall') elif sequences[s] == "restart_node_boot": conn.restart_node('boot') + elif sequences[s] == "fsck_repair": + conn.fsck_repair_node() elif sequences[s] == "repair_node_keys": if conn.compare_and_repair_nodekeys(): # the keys either are in sync or were forced in sync. - # so try to reboot the node again. - conn.restart_bootmanager('rins') + # so try to start BM again. + conn.restart_bootmanager(conn.get_nodestate()) pass else: # there was some failure to synchronize the keys. @@ -752,6 +834,8 @@ def restore(sitehist, hostname, config=None, forced_action=None): args['sequence'] = s args['bmlog'] = conn.get_bootmanager_log().read() args['viart'] = False + args['saveact'] = True + args['ccemail'] = True sitehist.sendMessage('unknownsequence_notice', **args) conn.restart_bootmanager('boot') @@ -759,7 +843,7 @@ def restore(sitehist, hostname, config=None, forced_action=None): # TODO: differentiate this and the 'nodenetwork_email' actions. elif sequences[s] == "update_node_config_email": - if not found_within(recent_actions, 'nodeconfig_notice', 3): + if not found_within(recent_actions, 'nodeconfig_notice', 3.5): args = {} args['hostname'] = hostname sitehist.sendMessage('nodeconfig_notice', **args) @@ -767,7 +851,7 @@ def restore(sitehist, hostname, config=None, forced_action=None): elif sequences[s] == "nodenetwork_email": - if not found_within(recent_actions, 'nodeconfig_notice', 3): + if not found_within(recent_actions, 'nodeconfig_notice', 3.5): args = {} args['hostname'] = hostname args['bmlog'] = conn.get_bootmanager_log().read() @@ -776,7 +860,7 @@ def restore(sitehist, hostname, config=None, forced_action=None): elif sequences[s] == "update_bootcd_email": - if not found_within(recent_actions, 'newalphacd_notice', 3): + if not found_within(recent_actions, 'newalphacd_notice', 3.5): args = {} args.update(getconf.getconf(hostname)) # NOTE: Generates boot images for the user: args['hostname'] = hostname @@ -790,17 +874,17 @@ def restore(sitehist, hostname, config=None, forced_action=None): # require either an exception "/minhw" or other manual intervention. # Definitely need to send out some more EMAIL. # TODO: email notice of broken hardware - if not found_within(recent_actions, 'baddisk_notice', 1): + if not found_within(recent_actions, 'baddisk_notice', 7): print "...NOTIFYING OWNERS OF BROKEN HARDWARE on %s!!!" % hostname args = {} args['hostname'] = hostname args['log'] = conn.get_dmesg().read() sitehist.sendMessage('baddisk_notice', **args) - conn.set_nodestate('disable') + conn.set_nodestate('disabled') elif sequences[s] == "update_hardware_email": - if not found_within(recent_actions, 'minimalhardware_notice', 1): + if not found_within(recent_actions, 'minimalhardware_notice', 7): print "...NOTIFYING OWNERS OF MINIMAL HARDWARE FAILURE on %s!!!" % hostname args = {} args['hostname'] = hostname @@ -824,7 +908,7 @@ def restore(sitehist, hostname, config=None, forced_action=None): args['hostname'] = hostname args['network_config'] = nodenet_str - args['nodenetwork_id'] = net['nodenetwork_id'] + args['interface_id'] = net['interface_id'] sitehist.sendMessage('baddns_notice', **args)