Scenario-Based Troubleshooting Tips and Techniques

Edit online

Depending on your scenario, troubleshooting tips and techniques are available to possibly remedy the problem.

This topic contains a list of scenario-based troubleshooting tips and techniques. This list is not exhaustive and does not cover all scenarios.

Situation Explanation/Resolution

During Oracle RAC failover, DMI visibility takes time to recover under high volume test when default queue sizes and thread counts are used.

You can increase the number of visibility queues and event input queue capacity in customer_overrides.properties to ensure that adapters are responsive.

To increase the default queue size and the thread counts, perform the following tasks:
  1. In the customer_overrides.properties file, set the following values:
    • dmivisibility.number_visibility_threads=6
    • dmivisibility.event_input_queue_capacity=3072
  2. From the install_dir/bin directory, run one of the following commands:
    • (UNIX or Linux®) setupfiles.sh
    • (Windows) setupfiles.cmd
  3. Restart Sterling B2B Integrator.

While running Oracle 11g in non-RAC or RAC environment, you might encounter index hot spots, which may cause performance degradation.
It is recommended that you create the following indexes in a global partition to improve system performance:
Note: If you have not deployed Sterling File Gateway in your environment, you can ignore the indexes starting with the name FG.
  • PERF_IDX_1 : WORKFLOW_CONTEXT (BASIC_STATUS, START_TIME, WORKFLOW_ID, WFD_ID)
  • PERF_IDX_2 : WORKFLOW_CONTEXT (START_TIME, WORKFLOW_ID, STEP_ID, ACTIVITYINFO_ID, WFD_ID)
  • PERF_IDX_3 : WORKFLOW_CONTEXT (WFD_ID, START_TIME, WORKFLOW_ID, STEP_ID, ACTIVITYINFO_ID)
  • SCI_IDX_125 : WORKFLOW_CONTEXT (WFD_ID, WFD_VERSION)
  • SCI_IDX_195 : WORKFLOW_CONTEXT (END_TIME, WFD_ID, WFD_VERSION, STEP_ID)
  • SCI_IDX_3 : WORKFLOW_CONTEXT (WORKFLOW_ID)
  • PERF_IDX_4 : WORKFLOW_CONTEXT (WORKFLOW_ID, START_TIME, STEP_ID, ACTIVITYINFO_ID, WFD_ID)
  • SCI_IDX_120 : WORKFLOW_CONTEXT (WORKFLOW_ID, BASIC_STATUS, NEXT_AI_ID, ACTIVITYINFO_ID, ARCHIVE_FLAG)
  • SCI_IDX_109 : WORKFLOW_CONTEXT" (WORKFLOW_ID, BRANCH_ID)
  • SCI_IDX_5 : EVENT (EXPIRES_AT)
  • SCI_IDX_56_2 : TRANS_DATA (WF_ID)
  • SCI_IDX_DMI17 : ACT_SESSION (CON_END_TIME, PROTOCOL, SESSION_ID, ADAPTER_NAME)
  • FG_RTE_MODTS_IDX : FG_ROUTE_EVENT (MODIFYTS)
  • FG_TEA_EVTKEY_IDX : FG_EVENTATTR (EVENT_KEY)
  • FG_RE_DF_IDX : FG_ROUTE_EVENT (DATA_FLOW_ID)
Note: If there are any changes in a patch for the partitioned indexes, the patch installation will drop the index and create a non-partitioned index, treating it as a normal index. You should analyze the indexdrop SQL's and recreate the partitioned indexes again if they have changed.
The following example explains the procedure to drop an index and create the same in a global partition:

  1. To drop index PERF_IDX_1, run the following sql command:

    drop index PERF_IDX_1 ;

  2. Create the PERF_IDX_1 in global partitions by running the following command:

    create index PERF_IDX_1 on WORKFLOW_CONTEXT (BASIC_STATUS,START_TIME,WORKFLOW_ID,WFD_ID) global partition by hash (BASIC_STATUS,START_TIME,WORKFLOW_ID,WFD_ID) partitions 16 store in (SI);

In some cases, perimeter services may fail with the following error. The error will be recorded in the perimeter services log (perimeter.log).

You can improve the situation by reducing high timeout values in communications adapters and by increasing the delayed event queue size (delayed.event.queue.size) in perimeter.properties file.

The default value of the delayed.event.queue.size property is 30000. You can increase the value based on your requirement.