def blockcheck(self):
info("ACTION: Validating database for corruptions")
# The following command will introduce some corruption to test database validation
# check_call(['dd','if=/dev/urandom','of=/nfs/autorestore/mnt/data_D-ORCL_I-1373437895_TS-SOE_FNO-5_0sqov4pv','bs=8192','count=10','seek=200','conv=notrunc' ])
try:
self._exec_rman(self._restoretemplate.get('validateblocks'))
self._validatecorruption = True
finally:
self._exec_sqlplus(self._restoretemplate.get('showcorruptblocks'))
def delete_expired_datafilecopy():
output = exec_sqlplus(rmantemplateconfig.get('deletedatafilecopy'),
silent=True)
rmanscript = ""
for line in output.splitlines():
if line.startswith('DELETECOPY: '):
rmanscript += "%s\n" % line.strip()[12:]
if rmanscript:
info("- Deleting expired datafile copies")
exec_rman(rmanscript)
def backup_missing_archlog():
output = exec_sqlplus(rmantemplateconfig.get('archivelogmissing'),
silent=True)
archlogscript = ""
for line in output.splitlines():
if line.startswith('BACKUP force as copy'):
archlogscript += "%s\n" % line.strip()
if archlogscript:
info("- Copying missing archivelogs")
exec_rman(
"run {\n%s\n%s\n}" %
(rmantemplateconfig.get('allocatearchlogchannel'), archlogscript))
def generate_restore():
print "\n============="
info(rmantemplateconfig.get('headerrestore'))
if registercatalog:
info(rmantemplateconfig.get('headercatalog'))
info(rmantemplateconfig.get('fullrestore'))
info(rmantemplateconfig.get('restorefooter'))
def run(self):
self.starttime = datetime.now()
info("Starting to restore")
#
success = False
self.clone()
try:
self._mount()
except:
self.cleanup()
raise Exception('restore', 'Mount failed')
self._set_parameters()
self._createinitora()
self._exec = OracleExec(oraclehome=Configuration.get('oraclehome', 'generic'),
tnspath=os.path.join(scriptpath(), Configuration.get('tnsadmin', 'generic')),
sid=self._dbparams['dbname'])
#
self._run_restore()
def configure():
rmanscript = ''
# Create directory for archive logs
if not os.path.exists(archdir):
os.makedirs(archdir)
# Register database in catalog if needed
if registercatalog:
alreadyregistered = False
info("Checking from RMAN catalog if database is already registered")
output = exec_sqlplus(
rmantemplateconfig.get('isdbregisteredincatalog'),
silent=True,
header=False)
for line in output.splitlines():
if line.startswith('DATABASE IS REGISTERED IN RC'):
alreadyregistered = True
if not alreadyregistered:
rmanscript += rmantemplateconfig.get('registerdatabase')
# Configure archivelog deletion policy
if hasdataguard:
rmanscript += "\n%s" % rmantemplateconfig.get('configdelaldg')
else:
rmanscript += "\n%s" % rmantemplateconfig.get('configdelalnodg')
# configures rman default settings
rmanscript += "\n%s" % rmantemplateconfig.get('config')
info("Running RMAN configuration")
exec_rman(rmanscript)
info("Running additional configuration from SQL*Plus")
exec_sqlplus(rmantemplateconfig.get('configfromsqlplus'))
def exec_restore():
global exitvalue
restore.clone(False)
print "Please execute the following command as root to mount the backup volume:"
print ""
print "mount -t nfs -o rw,bg,hard,nointr,rsize=32768,wsize=32768,tcp,vers=3,timeo=600 %s %s" % (restore.mountstring, restoreparams['mountpath'])
print ""
while ask_yn("Did you execute it") == "N":
print "Please execute it then."
# Verify that clone is mounted
autorestorefile = os.path.join(restoreparams['mountpath'], 'autorestore.cfg')
if not os.path.isfile(autorestorefile):
print "The mounted path does not look correct, file %s not found" % autorestorefile
exitvalue = 1
return
#
debug("Oracle home: %s" % Configuration.get("oraclehome", "generic"))
debug("Clone mount path: %s" % restoreparams['mountpath'])
debug("Target instance SID: %s" % restoreparams['sid'])
debug("Restore target time UTC: %s" % restoreparams['timepoint'].astimezone(pytz.utc))
debug("Restore target time local: %s" % restoreparams['timepoint'].astimezone(get_localzone()))
debug("Restored from snapshot: %s" % restore.sourcesnapid)
info("Starting database restore")
#
try:
restore.pit_restore(restoreparams['mountpath'], restoreparams['sid'])
restore.verify(False)
info("Database restore complete")
info("SID: %s" % restoreparams['sid'])
info("Requested target time: %s" % restoreparams['timepoint'].astimezone(get_localzone()))
info("Verified restored database time: %s" % restore.verifytime)
info("Difference from target: %s" % restore.verifydiff)
except:
exception("Database restore failed")
exitvalue = 1
print ""
print "Commands to clean up:"
print "1. Shut down database instance %s" % restoreparams['sid']
print "2. Execute as root: umount %s" % restoreparams['mountpath']
print "3. Drop clone: BACKUPCONFIG=%s %s %s dropclone %s" % (os.path.basename(Configuration.configfilename),
os.path.join(scriptpath(), 'zsnapper.py'), configname, restore.clonename)
def runrestore(database):
global exitstatus
#
Configuration.defaultsection = database
# Reinitialize logging
BackupLogger.init(
os.path.join(
logdir, "%s-%s.log" %
(datetime.now().strftime('%Y%m%dT%H%M%S'), database)), database)
BackupLogger.clean()
#
restore = RestoreDB(database)
restore.set_mount_path(mountdest)
restore.set_restore_path(restoredest)
#
info("Logfile: %s" % BackupLogger.logfile)
Configuration.substitutions.update({'logfile': BackupLogger.logfile})
cleantarget()
#
success = False
#
if validatemodulus > 0:
# Validation based on modulus
validationinfo = validationdate(database)
validatecorruption = validationinfo[0]
if not validatecorruption:
debug("Next database validation in %d days: %s" %
(validationinfo[1], validationinfo[2]))
else:
# Validation based on random
validatecorruption = (validatechance > 0) and (randint(
1, validatechance) == validatechance)
if validatecorruption:
debug("Database will be validated during this restore session")
# Start restore
try:
restore.run()
restore.verify()
if validatecorruption:
restore.blockcheck()
success = True
except:
exitstatus = 1
exception(
"Error happened, but we can continue with the next database.")
finally:
restore.cleanup()
# Log result to catalog
Configuration.substitutions.update({
'log_dbname':
database,
'log_start':
restore.starttime.strftime('%Y-%m-%d %H-%M-%S'),
'log_stop':
restore.endtime.strftime('%Y-%m-%d %H-%M-%S'),
'log_success':
'1' if success else '0',
'log_diff':
restore.verifyseconds,
'log_snapid':
restore.sourcesnapid,
'log_validated':
'1' if validatecorruption else '0'
})
debug('Logging the result to catalog.')
try:
oexec.sqlldr(Configuration.get('autorestorecatalog', 'autorestore'),
restoretemplate.get('sqlldrlog'))
except:
debug("Sending the logfile to catalog failed.")
try:
oexec.sqlplus(restoretemplate.get('insertlog'), silent=False)
except:
debug("Logging the result to catalog failed.")
# Finish up
info("Restore %s, elapsed time: %s" %
('successful' if success else 'failed',
restore.endtime - restore.starttime))
BackupLogger.close(True)
print "Validation is based on chance, probability 1/%d" % validatechance
else:
print "Database validation is not turned on"
else:
# Actions that need a lock
lock = BackupLock(Configuration.get('autorestorelogdir', 'autorestore'))
try:
if action is not None:
if action.startswith('--'):
if action == '--createcatalog':
BackupLogger.init(
os.path.join(
logdir, "%s-config.log" %
(datetime.now().strftime('%Y%m%dT%H%M%S'))),
'config')
info("Logfile: %s" % BackupLogger.logfile)
Configuration.substitutions.update(
{'logfile': BackupLogger.logfile})
oexec.sqlplus(restoretemplate.get('createcatalog'),
silent=False)
else:
runrestore(action)
else:
# Loop through all sections
for configname in loopdatabases():
runrestore(configname)
# Run ADRCI to clean up diag
adrage = int(Configuration.get('logretention', 'generic')) * 1440
f1 = mkstemp(suffix=".adi")
ftmp = os.fdopen(f1[0], "w")
ftmp.write("set base %s\n" % logdir)
def runrestore(database):
global exitstatus
#
Configuration.defaultsection = database
#
restoredest = Configuration.get('autorestoredestination', 'autorestore')
mountdest = Configuration.get('autorestoremountpoint', 'autorestore')
logdir = Configuration.get('autorestorelogdir', 'autorestore')
Configuration.substitutions.update({
'logdir':
logdir,
'autorestorecatalog':
Configuration.get('autorestorecatalog', 'autorestore')
})
if restoredest is None or not os.path.exists(
restoredest) or not os.path.isdir(restoredest):
print "Restore directory %s not found or is not a proper directory" % restoredest
sys.exit(2)
if mountdest is None or not os.path.exists(mountdest) or not os.path.isdir(
mountdest):
print "Clone mount directory %s not found or is not a proper directory" % mountdest
sys.exit(2)
#
validatechance = int(
Configuration.get('autorestorevalidatechance', 'autorestore'))
validatemodulus = int(
Configuration.get('autorestoremodulus', 'autorestore'))
# Reinitialize logging
BackupLogger.init(
os.path.join(
logdir, "%s-%s.log" %
(datetime.now().strftime('%Y%m%dT%H%M%S'), database)), database)
BackupLogger.clean()
#
restore = RestoreDB(database)
restore.set_mount_path(mountdest)
restore.set_restore_path(restoredest)
#
info("Logfile: %s" % BackupLogger.logfile)
Configuration.substitutions.update({'logfile': BackupLogger.logfile})
cleantarget(restoredest)
#
success = False
#
if validatemodulus > 0:
# Validation based on modulus
validationinfo = validationdate(database)
validatecorruption = validationinfo[0]
if not validatecorruption:
debug("Next database validation in %d days: %s" %
(validationinfo[1], validationinfo[2]))
else:
# Validation based on random
validatecorruption = (validatechance > 0) and (randint(
1, validatechance) == validatechance)
if validatecorruption:
debug("Database will be validated during this restore session")
# Start restore
try:
restore.run()
restore.verify()
if validatecorruption:
restore.blockcheck()
success = True
except:
exitstatus = 1
exception(
"Error happened, but we can continue with the next database.")
finally:
restore.cleanup()
# Log result to catalog
Configuration.substitutions.update({
'log_dbname':
database,
'log_start':
restore.starttime.strftime('%Y-%m-%d %H-%M-%S'),
'log_stop':
restore.endtime.strftime('%Y-%m-%d %H-%M-%S'),
'log_success':
'1' if success else '0',
'log_diff':
restore.verifyseconds,
'log_snapid':
restore.sourcesnapid,
'log_validated':
'1' if validatecorruption else '0'
})
debug('Logging the result to catalog.')
try:
oexec.sqlldr(Configuration.get('autorestorecatalog', 'autorestore'),
restoretemplate.get('sqlldrlog'))
except:
debug("Sending the logfile to catalog failed.")
try:
oexec.sqlplus(restoretemplate.get('insertlog'), silent=False)
except:
debug("Logging the result to catalog failed.")
# Finish up
info("Restore %s, elapsed time: %s" %
('successful' if success else 'failed',
restore.endtime - restore.starttime))
# Run ADRCI to clean up diag
adrage = int(Configuration.get('logretention', 'generic')) * 1440
f1 = mkstemp(suffix=".adi")
ftmp = os.fdopen(f1[0], "w")
ftmp.write("set base %s\n" % logdir)
ftmp.write("show homes\n")
ftmp.close()
f2 = mkstemp(suffix=".adi")
ftmp2 = os.fdopen(f2[0], "w")
ftmp2.write("set base %s\n" % logdir)
with TemporaryFile() as f:
try:
oexec.adrci(f1[1], f)
f.seek(0, 0)
output = f.read()
startreading = False
for line in output.splitlines():
if line.startswith('ADR Homes:'):
startreading = True
elif startreading:
ftmp2.write("set home %s\n" % line.strip())
ftmp2.write("purge -age %d\n" % adrage)
ftmp2.close()
oexec.adrci(f2[1], f)
except:
print "Executing ADRCI failed."
finally:
os.unlink(f1[1])
os.unlink(f2[1])
#
BackupLogger.close(True)
def exec_restore():
global exitvalue
restore.clone(False)
print "Please execute the following command as root to mount the backup volume:"
print ""
print "mount -t nfs -o rw,bg,hard,nointr,rsize=32768,wsize=32768,tcp,vers=3,timeo=600 %s %s" % (restore.mountstring, restoreparams['mountpath'])
print ""
while ask_yn("Did you execute it") == "N":
print "Please execute it then."
# Verify that clone is mounted
autorestorefile = os.path.join(restoreparams['mountpath'], 'autorestore.cfg')
if not os.path.isfile(autorestorefile):
print "The mounted path does not look correct, file %s not found" % autorestorefile
exitvalue = 1
return
#
BackupLogger.init('/tmp/restore_%s_%s.log' % (datetime.now().strftime('%Y%m%dT%H%M%S'), configname))
BackupLogger.clean()
Configuration.substitutions.update({
'logdir': '/tmp',
'logfile': BackupLogger.logfile
})
print "Session log file: %s" % BackupLogger.logfile
info("Starting database restore")
#
try:
restore.pit_restore(restoreparams['mountpath'], restoreparams['sid'])
restore.verify(False)
info("Database restore complete")
info("SID: %s" % restoreparams['sid'])
info("Requested target time: %s" % restoreparams['timepoint'].astimezone(get_localzone()))
info("Verified restored database time: %s" % restore.verifytime)
info("Difference from target: %s" % restore.verifydiff)
except:
exception("Database restore failed")
exitvalue = 1
print ""
print "Commands to clean up:"
print "1. Shut down database instance %s" % restoreparams['sid']
print "2. Execute as root: umount %s" % restoreparams['mountpath']
print "3. Drop clone: BACKUPCONFIG=%s %s %s dropclone %s" % (os.path.basename(Configuration.configfilename),
os.path.join(scriptpath(), 'zsnapper.py'), configname, restore.clonename)
def imagecopywithsnap():
starttime = datetime.now()
restoreparamfile = os.path.join(backupdest, 'autorestore.cfg')
#
info("Check if there are missing archivelogs")
backup_missing_archlog()
#
info("Switch current log")
output = exec_sqlplus(rmantemplateconfig.get('archivecurrentlogs'),
silent=True)
if os.path.isfile(restoreparamfile):
with open(restoreparamfile, 'a') as f:
for line in output.splitlines():
if line.startswith('CURRENT DATABASE SCN:'):
f.write("lastscn: %s\n" % line.strip()[22:])
elif line.startswith('CURRENT DATABASE TIME:'):
f.write("lasttime: %s\n" % line.strip()[23:])
elif line.startswith('BCT FILE:'):
f.write("bctfile: %s\n" % line.strip()[10:])
#
if dosnapshot:
info("Snap the current backup area")
snapid = snap.snap()
debug("Created snapshot: %s" % snapid)
#
info("Checking for expired datafile copies")
delete_expired_datafilecopy()
#
info("Refresh imagecopy")
backup('imagecopy')
exec_sqlplus(rmantemplateconfig.get('archivecurrentlogs'))
#
if dosnapshot:
info("Clean expired snapshots")
cleaningresult = snap.clean()
for r in cleaningresult:
debug(r['infostring'])
#
info("Dump additional information about the environment to the log file")
if gimanaged:
p = Popen([os.path.join(scriptpath, 'dbinfo.py'), configsection],
stdout=PIPE,
stderr=None,
stdin=None)
output, outerr = p.communicate()
debug(output)
# Write ORACLE_HOME patch information to log file
p = Popen([
os.path.join(Configuration.get('oraclehome', 'generic'), 'OPatch',
'opatch'), 'lsinventory'
],
stdout=PIPE,
stderr=None,
stdin=None)
output, outerr = p.communicate()
debug(output)
#
info("Write database parameters for autorestore")
with open(restoreparamfile, 'w') as f:
f.write("[dbparams]\n")
output = exec_sqlplus(rmantemplateconfig.get('autorestoreparameters'),
silent=True)
for line in output.splitlines():
if line.startswith('dbconfig-'):
f.write("%s\n" % line[9:])
#
endtime = datetime.now()
info("------------ TOTAL ------------")
info("Total execution time: %s" % (endtime - starttime))
info("Execution started: %s" % starttime)
info("Execution finished: %s" % endtime)
for line in output.splitlines():
if line.startswith('CDB-DETECT:') and line.strip() <> 'CDB-DETECT: NO':
commonprefix = line.strip()[12:]
Configuration.substitutions.update({'scheduleuserprefix': commonprefix})
#
script = "%s\n" % rmantemplateconfig.get('createuser')
script += "%s\n" % rmantemplateconfig.get('dropschedule')
script += "%s\n" % rmantemplateconfig.get('createschedule')
exec_sqlplus(script)
################################################
### Main section
################################################
info("Configuration file: %s" % Configuration.configfilename)
lock = BackupLock(lockdir=backupdest,
maxlockwait=int(Configuration.get('maxlockwait', 'generic')))
try:
# User interface action execution
if scriptaction == 'config':
configure()
elif scriptaction == 'generaterestore':
generate_restore()
elif scriptaction == 'imagecopywithsnap':
imagecopywithsnap()
elif scriptaction == 'setschedule':
setschedule()
elif scriptaction == 'missingarchlog':
