Common Faults in Installing OMS

1. A Message Indicating a Decryption Failure Is Displayed When OMS Is Installed

Symptom:
java.security.InvalidKeyException: Illegal key size or default parameters
        at javax.crypto.Cipher.checkCryptoPerm(Cipher.java:1011)
        at javax.crypto.Cipher.implInit(Cipher.java:786)
        at javax.crypto.Cipher.chooseProvider(Cipher.java:849)
        at javax.crypto.Cipher.init(Cipher.java:1213)
        at javax.crypto.Cipher.init(Cipher.java:1153)
        at org.wcc.framework.util.encrypt.PasswordUtil.decryptByAes256(PasswordUtil.java:180)
        at org.wcc.framework.util.encrypt.PasswordUtil.decryptByAes256(PasswordUtil.java:146)
        at com.omm.encrypt.DecryptCommandMain.decryptCommandLine(DecryptCommandMain.java:71)
        at com.omm.encrypt.DecryptCommandMain.main(DecryptCommandMain.java:130)
[2014-12-09 02:58] ERROR Failed to get omm db password. [installcontroller.sh(main):788](12851)

Fault Locating:
The commonest cause of a decryption failure is that JDK versions do not match. The OMS system installs JDK under user omm. If JDK has been installed or JAVA_HOME environment variables have already existed under user root, OMS installation may fail.
Solution:
a.Uninstall JDK and remove JAVA_HOME environment variables under user root.

b.Run the env |grep JAVA_HOME command as user root to check for JAVA-related environment variables.

c.Delete any existed JAVA-related environment variables from the /root/.bash_profile file or JAVA_HOME-related variables from the /etc/profile file.

2. OMS Installation Fails Because LogMan Installation Fails

Symptom:

Fault Locating:
The cron configuration file has been modified. As a result, the crontab file cannot be modified and LogMan installation fails.
Solution:
Modify cron.deny.bak to cron.deny. Uninstall and then reinstall OMS.

3. GaussDB Installation Failure
Symptom:

 
Fault Locating:

No resources are available.

Solution:

Perform the following operations to increase the value of the parameter for restricting the OS semaphore resources based on requirements:

a. Add the following parameter to the /etc/sysctl.conf file:

kernel.sem = 2500 32000 1000 1280

b. For a SUSE system, run the following command to enable the SUSE Linux to automatically read kernel parameters during startup: (Skip this step in a Red Hat system.)

/sbin/chkconfig boot.sysctl on

c. Run the following command for the kernel parameters to take effect:

/sbin/sysctl -p

d. Reinstall OMS.

4. GaussDB Installation Failure
Symptom:

Fault Locating:
User root is not assigned to a correct group.

Solution:
Change the ID of group to which user root user assigns to 0 and reinstall OMS.

5. GaussDB Installation Failure
Symptom:

Fault Locating:
The value of shared_buffers is greater than the value of kernel.shmmax.
Solution:
a. Change the value of kernel.shmmax in the /etc/sysctl.conf file.
kernel.shmmax = value

You are advised to set the value of kernel.shmmaxto50% of the physical memory space.

Run the following command to query the physical memory space: grep MemTotal /proc/meminfo

Information similar to the following is displayed:

MeMTotal: 6088240 kB

In this example, set the value of kernel.shmmax to 3117178880 (6088240 x 1024/2 = 3117178880).

kernel.shmmax = 3117178880
b. Run the following command to enable the SUSE Linux to automatically read kernel parameters during startup:

/sbin/chkconfig boot.sysctl on
c. Run the following command for the kernel parameters to take effect:

/sbin/sysctl –p

6. OMS Installation Failure Due to Incorrect LADAP Installation

Symptom:
2014-12-09 02:58 566 linux-126 SlapdServer ERROR ldap server rpm is not installed!

2014-12-09 02:58 569 linux-126 SlapdServer ERROR ldap server rpm is not installed!

Fault Locating:
OMS Installation Failure Due to Incorrect LADAP Installation
Solution:

Run the preinstall script to add LADP-related RPM packages.

7. OMS Installation Failure Due to an Incorrect Floating IP Address

Symptom:  

Fault Locating:  
The NIC IP address does not map to the host name.

Processing logic of the installation script: If OMS is installed in single-node mode, values of om_float_ip_interface and ws_float_ip_interface can be obtained from the output of the ifconfig command by running the sed command by default.

local defaultInterface=$(ifconfig | sed -n "/${g_server1IP1}/{x;p};h" | awk '{print $1}')

The g_server1IP1 is obtained by running the hostname-i command. Therefore, error "The float_ip_interface is empty" is displayed because the IP addresses obtained by running the ifconfig command and the hostname –i command are inconsistent.
Solution:
a.Run the hostname -i command to check the host IP address.

b.Run the ifconfig command to check whether the corresponding NIC IP address is the same as the host IP address. If they are different, change them to the same value.

c.Verify that the host name in /etc/hosts maps the NIC IP address and reinstall OMS.

8. OMS Installation Failure Due to Incorrect File Permissions
Symptom:
Config_cmd:[/opt/huawei/Bigdata/OMSV100R001C00x8664/gaussdb/bin/gs_guc reload -c "replconninfo1='localhost=10.75.201.227 localport=12211 remotehost=10.75.201.218 remoteport=12211'"]sudo: must be setuid root
[/opt/huawei/Bigdata/OMSV100R001C00x8664/tools/omsconfig.sh:488] Gauss config failure, return:[1].[/opt/huawei/Bigdata/OMSV100R001C00x8664/tools/omsconfig.sh:626] ERROR: config Database system failure!
Fault Locating:
The sudo command permissions in /usr are changed.
Solution:
ls -l  /usr/bin/sudo
chown root:root /usr/bin/sudo
chmod 4755 /usr/bin/sudo
reboot

9. OMS Installation Fails Because the Installation Directory Is Not Empty

Symptom:

Fault Locating:
The OMS installation directory must not store any other files or directories.
Solution:

a. If the installation directory contains the uninstall script, run the script to clear the directory.

b. If the installation directory contains files irrelevant to FusionInsight HD, delete the files and reinstall OMS.

10. OMS Installation Fails Because /etc/hosts Is Not Correctly Configured
Symptom:

Fault Locating:

localhost 127.0.0.1 is not configured in /etc/hosts.
Solution:
Add localhost 127.0.0.1 in /etc/hosts. Uninstall and then reinstall OMS.

11. OMS Installation Failure Due to Incorrect Permission on the /etc/sudoers Directory
Symptom:

Fault Locating:

The permission on the /etc/sudoers file is incorrect. As a result, the sudo command fails to be executed and OMS fails to be installed.

Solution:

Change the permission on the /etc/sudoers directory to 440. Uninstall and then reinstall OMS.

12. NTP Startup Failure During Standby OMS Installation
Symptom：:

Solution:
a.Uninstall the standby OMS.

b.Disable the firewall.

c.Reinstall the standby OMS.

Common Faults in Installing a Cluster
1. NodeAgent Installation Failure
Symptom:
2014-12-09 02:58,164 WARN  [pool-33-thread-1] 160.138.1.19: Failed Command execution CommandResponse [success=false, completed=true, exitCode=12, stdOut=Installing node agent...
keytool error: java.lang.Exception: Alias <tomcat> does not exist
ERROR:Failed to start nodeagent Start nodeagent process.
ERROR:Failed to install node agent.
, stdErr=chmod: cannot access `/opt/huawei/Bigdata/nodeagent/setup/web/*': No such file or directory, command=Command [executableScript=su - omm -c "/opt/huawei/Bigdata/nodeagent/setup/installnodeagent.sh {OM_NODEAGENT_NODEID} 160.157.0.99 20025 /var/log/Bigdata {IS_START_AGENT}", capture=EXECUTE_ONLY, script=su - {OMM_USER} -c "{NODEAGENT_HOME}/setup/installnodeagent.sh {OM_NODEAGENT_NODEID} {OM_NODEAGENT_CONTROLLER_IP} {OM_NODEAGENT_CONTROLLER_PORT} {BIGDATA_LOG_HOME} {IS_START_AGENT}", success=null, ifFail=null]]  com.huawei.hadoop.om.controller.agent.nodesetup.SshActionChannelHandler.processFail(SshActionChannelHandler.java:515)
Fault Locating:

The permission on /etc/hosts is 600. As a result, the IP address cannot be found after the hostname -i command is run under user omm.
Solution:
Change the permission on /etc/hosts to a value greater than 644 and click Retry.
2. Cluster Installation Fails Due to a Package Download Failure
Symptom:

Possible Causes:

Cause 1: The disk space is exhausted.

Cause 2: The mutual trust fails between the node that fails to download packages and the controller node.

Check whether configurations about user omm (wheel group) are correct in DenyUsers, DenyGroups, AllowUsers, and AllowGroups in the /etc/ssh/sshd_config file.

If security hardening has been performed, the /etc/ssh/sshd_config file may contain the following configurations:

These configurations affect SSH connections, resulting in a package download failure.

Log in to the active OMS node as user omm. Connect to the other agent nodes using SSH. If the password is required, mutual trust has been damaged. Check permissions on /home/omm/.ssh and /home/omm/.ssh/*. Their permissions are 750, which is different from the normal values 700 and 600. It is concluded that the permission on /home/omm/.ssh has been changed, resulting in a mutual trust failure and finally an installation package download failure.

Run the expect command on the involved node. If this command cannot be found, package download fails due to the lack of the expect command.

Solution:

a.For cause 1

Delete unused files in the environment to release space.

b.For cause 2

Comment out ListenAddress and AllowGroups.

Change permissions on /home/omm/.ssh and /home/omm/.ssh/* to700 and 600.

Run the preinstall script on the involved node or install the expect software.

4. Node Discovery Failure
Symptom:

Fault Locating:

Because some encryption algorithms that are not supported are configured in /etc/ssh/sshd_config.

Solution:
a.Modify /etc/ssh/sshd_config.

b.Restart the SSH service.

service sshd restart

c.Re-discover nodes.

5. Multiple Service IP Addresses Are Displayed in the Node Discovery Result

Symptom: :

Possible Cause:

The same host name corresponds to multiple IP addresses in the /etc/hosts file.

Solution:

Delete the unnecessary IP addresses from the /etc/hosts file and re-discover nodes.