X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=bootman.py;h=5e8b908359f025fe9f0cb1bd4e1fbbee5df5b365;hb=refs%2Fheads%2F1.0;hp=ff2a6d5608780df0ac287f30d469a8fb93b059a2;hpb=5c139827242cd7896d21a5369d5c113b0b6418ba;p=monitor.git diff --git a/bootman.py b/bootman.py index ff2a6d5..5e8b908 100755 --- a/bootman.py +++ b/bootman.py @@ -26,6 +26,8 @@ from nodeconfig import network_config_to_str import traceback import config +class ExceptionDoubleSSHError(Exception): pass + import signal class Sopen(subprocess.Popen): def kill(self, signal = signal.SIGTERM): @@ -58,14 +60,18 @@ 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) + download(self.c, "/var/log/dmesg.bm.log", "log/history/%s-dmesg.%s.log" % (t_stamp, self.node)) + os.system("cp log/history/%s-dmesg.%s.log log/dmesg.%s.log" % (t_stamp, self.node, 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)) + t_stamp = time.strftime("%Y-%m-%d-%H:%M") + download(self.c, "/tmp/bm.log", "log/history/%s-bm.%s.log" % (t_stamp, self.node)) + #os.system("zcat log/bm.%s.log.gz > log/bm.%s.log" % (self.node, self.node)) + os.system("cp log/history/%s-bm.%s.log log/bm.%s.log" % (t_stamp, self.node, self.node)) log = open("log/bm.%s.log" % self.node, 'r') return log @@ -232,7 +238,7 @@ class PlanetLabSession: 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. @@ -324,7 +330,8 @@ def reboot(hostname, config=None, forced_action=None): mailtxt.newbootcd_one[1] % args, True, db='bootcd_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) print "\tDisabling %s due to out-of-date BOOTCD" % hostname api.UpdateNode(hostname, {'boot_state' : 'disable'}) @@ -337,6 +344,8 @@ def reboot(hostname, config=None, forced_action=None): try: k = SSHKnownHosts(); k.update(node); k.write(); del k except: + from nodecommon import email_exception + email_exception() print traceback.print_exc() return False @@ -345,9 +354,16 @@ def reboot(hostname, config=None, forced_action=None): session = PlanetLabSession(node, False, True) else: session = PlanetLabSession(node, config.nosetup, config.verbose) + except ExceptionDoubleSSHError, e: + msg = "ERROR setting up session for %s" % hostname + print msg + return False except Exception, e: - print "ERROR setting up session for %s" % hostname + msg = "ERROR setting up session for %s" % hostname + print msg print traceback.print_exc() + from nodecommon import email_exception + email_exception(msg) print e return False @@ -359,12 +375,18 @@ def reboot(hostname, config=None, forced_action=None): try: time.sleep(session.timeout*4) conn = session.get_connection(config) + except EOFError: + # failed twice... no need to report this really, it's just in a + # weird state... + return False except: print traceback.print_exc() + from nodecommon import email_exception + email_exception(node) return False if forced_action == "reboot": - conn.restart_node('rins') + conn.restart_node('reinstall') return True boot_state = conn.get_boot_state() @@ -455,7 +477,8 @@ def reboot(hostname, config=None, forced_action=None): mailtxt.baddisk[1] % args, True, db='hardware_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.set_nodestate('disable') return False @@ -514,6 +537,7 @@ def reboot(hostname, config=None, forced_action=None): ('hardwarerequirefail' , 'Hardware requirements not met'), ('mkfsfail' , 'while running: Running mkfs.ext2 -q -m 0 -j /dev/planetlab/vservers failed'), ('nofilereference', "No such file or directory: '/tmp/mnt/sysimg//vservers/.vref/planetlab-f8-i386/etc/hosts'"), + ('kernelcopyfail', "cp: cannot stat `/tmp/mnt/sysimg/boot/kernel-boot': No such file or directory"), ('chrootfail' , 'Running chroot /tmp/mnt/sysimg'), ('modulefail' , 'Unable to get list of system modules'), ('writeerror' , 'write error: No space left on device'), @@ -569,7 +593,7 @@ def reboot(hostname, config=None, forced_action=None): ]: 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", @@ -583,14 +607,23 @@ def reboot(hostname, config=None, forced_action=None): "bminit-cfg-auth-getplc-update-installinit-validate-rebuildinitrd-netcfg-update3-implementerror-nofilereference-update-debug-done", "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-exception-mkfsfail-update-debug-done", "bminit-cfg-auth-getplc-installinit-validate-rebuildinitrd-exception-chrootfail-update-debug-done", + "bminit-cfg-auth-getplc-update-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done", + "bminit-cfg-auth-getplc-hardware-installinit-installdisk-installbootfs-installcfg-installstop-update-installinit-validate-rebuildinitrd-netcfg-disk-update4-update3-update3-kernelcopyfail-exception-update-debug-done", "bminit-cfg-auth-getplc-installinit-validate-exception-noinstall-update-debug-done", + # actual solution appears to involve removing the bad files, and + # continually trying to boot the node. + "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", ]: 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"}) - # 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", @@ -612,6 +645,7 @@ def reboot(hostname, config=None, forced_action=None): "bminit-cfg-auth-implementerror-bootcheckfail-update-implementerror-bootupdatefail-done", "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", ]: sequences.update({n: "restart_node_boot"}) @@ -668,7 +702,7 @@ def reboot(hostname, config=None, forced_action=None): m = PersistMessage(hostname, mailtxt.unknownsequence[0] % args, mailtxt.unknownsequence[1] % args, False, db='unknown_persistmessages') m.reset() - m.send(['monitor-list@lists.planet-lab.org']) + m.send([config.cc_email]) conn.restart_bootmanager('boot') @@ -683,16 +717,16 @@ def reboot(hostname, config=None, forced_action=None): conn.restart_bootmanager('boot') elif sequences[s] == "restart_bootmanager_rins": if config and not config.quiet: print "...Restarting BootManager.py on %s "% node - 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] == "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') + conn.restart_bootmanager('reinstall') pass else: # there was some failure to synchronize the keys. @@ -706,7 +740,7 @@ def reboot(hostname, config=None, forced_action=None): m = PersistMessage(hostname, "Suspicous error from BootManager on %s" % args, mailtxt.unknownsequence[1] % args, False, db='suspect_persistmessages') m.reset() - m.send(['monitor-list@lists.planet-lab.org']) + m.send([config.cc_email]) conn.restart_bootmanager('boot') @@ -717,7 +751,8 @@ def reboot(hostname, config=None, forced_action=None): m = PersistMessage(hostname, mailtxt.plnode_cfg[0] % args, mailtxt.plnode_cfg[1] % args, True, db='nodeid_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.dump_plconf_file() conn.set_nodestate('disable') @@ -726,10 +761,11 @@ def reboot(hostname, config=None, forced_action=None): args = {} args['hostname'] = hostname args['bmlog'] = conn.get_bootmanager_log().read() - m = PersistMessage(hostname, mailtxt.plnode_network[0] % args, mailtxt.plnode_cfg[1] % args, + m = PersistMessage(hostname, mailtxt.plnode_cfg[0] % args, mailtxt.plnode_cfg[1] % args, True, db='nodenet_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.dump_plconf_file() conn.set_nodestate('disable') @@ -744,7 +780,8 @@ def reboot(hostname, config=None, forced_action=None): mailtxt.newalphacd_one[1] % args, True, db='bootcd_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) print "\tDisabling %s due to out-of-date BOOTCD" % hostname conn.set_nodestate('disable') @@ -762,7 +799,8 @@ def reboot(hostname, config=None, forced_action=None): mailtxt.baddisk[1] % args, True, db='hardware_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.set_nodestate('disable') elif sequences[s] == "update_hardware_email": @@ -774,7 +812,8 @@ def reboot(hostname, config=None, forced_action=None): mailtxt.minimalhardware[1] % args, True, db='minhardware_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.set_nodestate('disable') elif sequences[s] == "bad_dns_email": @@ -782,8 +821,10 @@ def reboot(hostname, config=None, forced_action=None): args = {} try: node = api.GetNodes(hostname)[0] - net = api.GetNodeNetworks(node['nodenetwork_ids'])[0] + net = api.GetInterfaces(node['interface_ids'])[0] except: + from nodecommon import email_exception + email_exception() print traceback.print_exc() # TODO: api error. skip email, b/c all info is not available, # flag_set will not be recorded. @@ -792,12 +833,13 @@ def reboot(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'] m = PersistMessage(hostname, mailtxt.baddns[0] % args, mailtxt.baddns[1] % args, True, db='baddns_persistmessages') loginbase = plc.siteId(hostname) - m.send([const.PIEMAIL % loginbase, const.TECHEMAIL % loginbase]) + emails = plc.getTechEmails(loginbase) + m.send(emails) conn.set_nodestate('disable') if flag_set: