def _parseEquals(self, list):
return parseEquals(list)
def _parseEquals(self, list): return parseEquals(list)
def __check_jobtracker(self, desc, id, pkgdir):
# Check jobtracker status. Return properly if it is ready to accept jobs.
# Currently Checks for Jetty to come up, the last thing that can be checked
# before JT completes initialisation. To be perfectly reliable, we need
# hadoop support
name = desc.getName()
if name == 'jobtracker':
# Yes I am the Jobtracker
self.log.debug("Waiting for jobtracker to initialise")
version = desc.getVersion()
self.log.debug("jobtracker version : %s" % version)
hadoopCmd = self.getRunningValues()[id]
attrs = hadoopCmd.getFilledInKeyValues()
attrs = parseEquals(attrs)
jobTrackerAddr = attrs['mapred.job.tracker']
self.log.debug("jobtracker rpc server : %s" % jobTrackerAddr)
if version < 16:
jettyAddr = jobTrackerAddr.split(':')[0] + ':' + \
attrs['mapred.job.tracker.info.port']
else:
jettyAddr = attrs['mapred.job.tracker.http.address']
self.log.debug("Jobtracker jetty : %s" % jettyAddr)
# Check for Jetty to come up
# For this do a http head, and then look at the status
defaultTimeout = socket.getdefaulttimeout()
# socket timeout isn`t exposed at httplib level. Setting explicitly.
socket.setdefaulttimeout(1)
sleepTime = 0.5
jettyStatus = False
jettyStatusmsg = ""
while sleepTime <= 32:
# There is a possibility that the command might fail after a while.
# This code will check if the command failed so that a better
# error message can be returned to the user.
if not hadoopCmd.getCommandStatus():
self.log.critical('Hadoop command found to have failed when ' \
'checking for jobtracker status')
hadoopCmd.handleFailedCommand()
addnInfo = ""
if hadoopCmd.stdErrContents is not "":
addnInfo = " Information from stderr of the command:\n%s" \
% (hadoopCmd.stdErrContents)
raise Exception("Could not launch the %s using %s/bin/hadoop.%s" \
% (desc.getName(), pkgdir, addnInfo))
try:
jettyConn = httplib.HTTPConnection(jettyAddr)
jettyConn.request("HEAD", "/jobtracker.jsp")
# httplib inherently retries the following till socket timeout
resp = jettyConn.getresponse()
if resp.status != 200:
# Some problem?
jettyStatus = False
jettyStatusmsg = "Jetty gave a non-200 response to a HTTP-HEAD" +\
" request. HTTP Status (Code, Msg): (%s, %s)" % \
( resp.status, resp.reason )
break
else:
self.log.info("Jetty returned a 200 status (%s)" % resp.reason)
self.log.info("JobTracker successfully initialised")
return
except socket.error:
self.log.debug("Jetty gave a socket error. Sleeping for %s" \
% sleepTime)
time.sleep(sleepTime)
sleepTime = sleepTime * 2
except Exception, e:
jettyStatus = False
jettyStatusmsg = ("Process(possibly other than jetty) running on" + \
" port assigned to jetty is returning invalid http response")
break
socket.setdefaulttimeout(defaultTimeout)
if not jettyStatus:
self.log.critical("Jobtracker failed to initialise.")
if jettyStatusmsg:
self.log.critical( "Reason: %s" % jettyStatusmsg )
else: self.log.critical( "Reason: Jetty failed to give response")
raise Exception("JobTracker failed to initialise")
def __check_jobtracker(self, desc, id, pkgdir):
# Check jobtracker status. Return properly if it is ready to accept jobs.
# Currently Checks for Jetty to come up, the last thing that can be checked
# before JT completes initialisation. To be perfectly reliable, we need
# hadoop support
name = desc.getName()
if name == 'jobtracker':
# Yes I am the Jobtracker
self.log.debug("Waiting for jobtracker to initialise")
version = desc.getVersion()
self.log.debug("jobtracker version : %s" % version)
hadoopCmd = self.getRunningValues()[id]
attrs = hadoopCmd.getFilledInKeyValues()
attrs = parseEquals(attrs)
jobTrackerAddr = attrs['mapred.job.tracker']
self.log.debug("jobtracker rpc server : %s" % jobTrackerAddr)
if version < 16:
jettyAddr = jobTrackerAddr.split(':')[0] + ':' + \
attrs['mapred.job.tracker.info.port']
else:
jettyAddr = attrs['mapred.job.tracker.http.address']
self.log.debug("Jobtracker jetty : %s" % jettyAddr)
# Check for Jetty to come up
# For this do a http head, and then look at the status
defaultTimeout = socket.getdefaulttimeout()
# socket timeout isn`t exposed at httplib level. Setting explicitly.
socket.setdefaulttimeout(1)
sleepTime = 0.5
jettyStatus = False
jettyStatusmsg = ""
while sleepTime <= 32:
# There is a possibility that the command might fail after a while.
# This code will check if the command failed so that a better
# error message can be returned to the user.
if not hadoopCmd.getCommandStatus():
self.log.critical('Hadoop command found to have failed when ' \
'checking for jobtracker status')
hadoopCmd.handleFailedCommand()
addnInfo = ""
if hadoopCmd.stdErrContents is not "":
addnInfo = " Information from stderr of the command:\n%s" \
% (hadoopCmd.stdErrContents)
raise Exception("Could not launch the %s using %s/bin/hadoop.%s" \
% (desc.getName(), pkgdir, addnInfo))
try:
jettyConn = httplib.HTTPConnection(jettyAddr)
jettyConn.request("HEAD", "/jobtracker.jsp")
# httplib inherently retries the following till socket timeout
resp = jettyConn.getresponse()
if resp.status != 200:
# Some problem?
jettyStatus = False
jettyStatusmsg = "Jetty gave a non-200 response to a HTTP-HEAD" +\
" request. HTTP Status (Code, Msg): (%s, %s)" % \
( resp.status, resp.reason )
break
else:
self.log.info("Jetty returned a 200 status (%s)" %
resp.reason)
self.log.info("JobTracker successfully initialised")
return
except socket.error:
self.log.debug("Jetty gave a socket error. Sleeping for %s" \
% sleepTime)
time.sleep(sleepTime)
sleepTime = sleepTime * 2
except Exception, e:
jettyStatus = False
jettyStatusmsg = ("Process(possibly other than jetty) running on" + \
" port assigned to jetty is returning invalid http response")
break
socket.setdefaulttimeout(defaultTimeout)
if not jettyStatus:
self.log.critical("Jobtracker failed to initialise.")
if jettyStatusmsg:
self.log.critical("Reason: %s" % jettyStatusmsg)
else:
self.log.critical("Reason: Jetty failed to give response")
raise Exception("JobTracker failed to initialise")