Contents

Index

Troubleshooting the N1 SPS Oracle 10g R2 RAC Plug-In

When you experience problems with a component installation or removal you need to check the standard output and standard error logs held by N1 SPS. If the output does not provide sufficient detail to enable you to determine the root cause, you will need to enable the shell script debugging flag and retry the plan. To enable debugging, you should set the debug_shell_code session variable to true. This results in the -x flag being set prior to shell commands being executed, thereby enabling you to see the results of the commands run and the shell variables that are used for substitution.

For components that do not use a synchronized installation, the root cause of a problem will usually be given by error messages at the end of the standard error output.

For components that use a synchronized installation, the root cause of the problem could be due to a fault on any one of the nodes targeted. You should start by examining the standard output and standard error messages on the targeted system with the lowest cluster node id. (The cluster node ID for a node can be obtained by running /usr/sbin/clinfo -n). If the output indicates contains Detected that a remote plan failed,then you need to consult the output from the other nodes to see which of them encountered the problem,. Otherwise, you will find the cause of the problem in the log you are currently viewing.

Under some circumstances you will find messages such as the following:

However, when you create a rac-framework-rg resource group, this message would be expected because the resource group might not exist. Alternatively, the message might indicate that you have mistyped the resource group name and as a result a new one has been created with the wrong name.

When you debug problems with CRS, RAC, or database installation, it will often be necessary to consult the log files that Oracle generates during the procedure. You should look through the plan standard output to find the names of any log files that Oracle might have created and then use those log files to investigate further.

Example: Troubleshooting a Plan Failure

The RAC framework resource group exists on nodes pgyruss1, pgyruss2 and pgyruss3 and you want to expand it to pgyruss4. You target the plan at all four nodes but all of them fail. The standard error for the plan on pgyruss2, pgyruss3 and pgyruss4 contains the following in the output of the final "native procedure ksh" step:

However, when you look at the standard error for pgyruss1, you find the following:

The root cause is therefore a result of the underlying Sun Cluster clresourcegroup command failing. You rectify the problem and complete the expansion of the resource group successfully.