Hello everyone, I hope you can help me with this. I tried to install CM but after a few turning points at the User Credentials for the cluster stage it just halted. So I killed all the processes, and because I didn't know how to start the server again I simply unistalled everything via the uninstall script in the /usr/share/cmf directory. The problem is that after reinstalling, the server won't start at all, so I can't access to the Cloudera Manager to finish my reinstall.
After searching my logs, the error happnes right after the following lines
2015-02-13 11:48:17,422 INFO [main:core.QuartzScheduler@238] Quartz Scheduler v.2.0.2 created. 2015-02-13 11:48:17,424 INFO [main:simpl.RAMJobStore@154] RAMJobStore initialized. 2015-02-13 11:48:17,427 INFO [main:core.QuartzScheduler@259] Scheduler meta-data: Quartz Scheduler (v2.0.2) 'com.cloudera.cmf.scheduler-1' with instanceId 'NON_CLUSTERED' Scheduler class: 'org.quartz.core.QuartzScheduler' - running locally. NOT STARTED. Currently in standby mode. Number of jobs executed: 0 Using thread pool 'org.quartz.simpl.SimpleThreadPool' - with 1 threads. Using job-store 'org.quartz.simpl.RAMJobStore' - which does not support persistence. and is not clustered. 2015-02-13 11:48:17,428 INFO [main:impl.StdSchedulerFactory@1288] Quartz scheduler 'com.cloudera.cmf.scheduler-1' initialized from an externally provided properties instance. 2015-02-13 11:48:17,428 INFO [main:impl.StdSchedulerFactory@1292] Quartz scheduler version: 2.0.2 2015-02-13 11:48:17,481 INFO [main:scheduler.CmfScheduler@301] Added command schedule '1' to the scheduler 2015-02-13 11:48:17,521 INFO [main:cdhclient.CdhExecutorFactory@110] Adding options to bad options list: 2015-02-13 11:48:19,798 INFO [main:components.PythonInterpreterFactory@38] Constructing Python interpreter 2015-02-13 11:48:22,398 INFO [main:components.PythonInterpreterFactory@45] Finished constructing Python interpreter 2015-02-13 11:48:22,420 INFO [main:components.JythonObjectFactoryImpl@53] Constructing Jython object factory 2015-02-13 11:48:28,670 INFO [MainThread:components.JythonObjectFactoryImpl@66] Finished constructing Jython object factory
In my error log shows the following:
JAVA_HOME=/usr/lib/jvm/java-7-oracle java.lang.OutOfMemoryError: PermGen space Dumping heap to java_pid26146.hprof ... Unable to create java_pid26146.hprof: Permission denied # # java.lang.OutOfMemoryError: PermGen space # -XX:OnOutOfMemoryError="kill -9 %p" # Executing /bin/sh -c "kill -9 26146"...
I've increased the PermGen memory by adding the following line at the begining of the file /usr/share/cmf/bin/cmf-server
CMF_OPTS = "$CMF_OPTS -XX:MaxPermSize=2048M"
And although the server wouldn't die with an OutOfMemoryError, it never budged from that point nor the http worked.
I've uploaded the heapdump here.
Thanks for replying!
This is what happens after every restart.
fmartinez-Lenovo-G480 fmartinez # service cloudera-scm-server restart Stopping cloudera-scm-server: /sbin/start-stop-daemon: warning: failed to kill 26704: No such process * cloudera-scm-server stopped Starting cloudera-scm-server: * cloudera-scm-server started
I restart whenever I see I cannot access through the port. Also, my logs get very big. I have uploaded the log file here. But the last part of it is just what it says in my first post. It ends there.
Thanks, in advance.
I've tried to point Cloudera to that directory, but it just doesn't exists. The first time I installed CM, it installed it's own Java. But since I deleted that directory the first time to try to reinstall it, it just doesn't install it anymore.
I think the reason why CM dies without any message is because it's JVM that is dieing with an OutOfMemoryException.
Is there any hidden file that flags to make a partial installation or something like that? Any temporary file? Or is there any command to force CM to try to make a fresh install and overwirte anything in its path?