SharePlex 9.2.6 - Administration Guide

View and kill SharePlex processes

These instructions show you how to forcefully terminate SharePlex processes in cases where replication must be shut down immediately. In normal circumstances, follow the shutdown instructions in the following topics:

Run SharePlex on Unix and Linux

Run SharePlex on Windows

View and terminate processes on Unix and Linux

On Unix and Linux systems, you can use the ps -ef | grep sp_ command to view the SharePlex processes that are running.

• The sp_cop process is the root process.

• The following child processes are spawned by sp_cop on a source system:

  • Command and Control process (sp_cnc)
  • Capture (sp_ocap)
  • Read (sp_ordr)
  • Export (sp_xport)

• The following child processes are spawned by sp_cop on a target system:

  1. Command and Control process (sp_cnc)

  2. Import (sp_mport)
  3. Post (sp_opst_mt if the database is Oracle or sp_xpst if the database is Open Target)

Each child process has the same -uidentifier as its parent sp_cop process. This makes it easier to identify related processes when multiple session of sp_cop are running.

To terminate a SharePlex process on Unix and Linux

Or...

View and terminate SharePlex processes on Windows

On Windows systems, SharePlex is managed by the NuTCRACKER operating environment, rather than the Windows operating system. However, you can view and terminate SharePlex processes in the normal manner, as you would other Windows programs.

View SharePlex processes on Windows

You can view the SharePlex processes on Windows systems in the following ways:

• From the Command Prompt console using the tlist program provided with the SharePlex software.
• From the Windows Task Manager.

In the Windows Task Manager, SharePlex appears as Sp_Copsrv.exe, representing the SharePlex sp_cop process. The operating system controls the parent Sp_Copsrv.exe service. The parent Sp_Copsrv.exe process spawns child Sp_Copsrv.exe processes — one for each replication process (Capture, Read, Export, Import, Post, sp_ctrl, and so forth.).

For a standard uni-directional configuration replicating through default queues to one target system, there are following processes on a Windows system:

On the source system:

• One parent Sp_Copsrv.exe process.
• One Sp_Ocap or Sp_capture (Capture) process plus one child Sp_Copsrv.exe process.
• One Sp_Ordr (Read) process plus one child Sp_Copsrv.exe process.
• One Sp_Xport (Export) process plus one child Sp_Copsrv.exe process.
• If there are any additional SharePlex processes running, such as sp_ctrl, there is an additional Sp_Copsrv.exe process for each one.

On the target system:

• One parent Sp_Copsrv.exe process.
• One Sp_Mport (Import) process plus one child Sp_Copsrv.exe process.
• One Sp_Opst_Mt (Oracle Post) or Sp_Xpst (Open Target Post) process plus one child Sp_Copsrv.exe process.
• If there are any additional SharePlex processes running, such as sp_ctrl, there is an additional Sp_Copsrv.exe process for each one.

If there are no active replication configurations, the SharePlex processes do not start when you start the service, and just the parent Sp_Copsrv.exe will be running.

To identify the parent Sp_Copsrv.exe process in the Windows Task Manager, look for the one that is using the largest amount of memory. The child Sp_Copsrv.exe processes consume less memory than the parent process.

To identify which replication process is associated with a child Sp_Copsrv.exe process, look in the SharePlex Event Log for the message stating when the replication process started. This entry provides the PID for that process and the PID of the associated Sp_copsvr.exe process.

Terminate SharePlex processes on Windows

You can terminate SharePlex processes on Windows systems as follows:

• Windows Task Manager: Terminate one SharePlex process at a time
• KillSharePlex.bat: Terminate all SharePlex processes at once with a batch file.

To terminate a process with Task Manager

1. Press Control | Alt | Delete.
2. Select Start Task Manager or Task Manager, depending on your Windows version.
3. Select the Processes tab.
4. (Optional) Sort the processes by name.
5. Select the process that you want to kill.
6. Click End Process.

To terminate all processes with KillSharePlex.bat

Run KillSharePlex.bat from the bin sub-directory of the SharePlex product directory.

Note: This tool terminates all SharePlex processes without displaying them. There are no user prompts or confirmation messages.

View events and errors

SharePlex reports errors and other abnormal conditions in the following ways.

Event Log

SharePlex reports operational errors, notices and warning conditions to the Event Log. This log provides a perpetual step-by-step record of replication activities, errors, and events. The Event Log can help you replay the sequence of events that led up to a problem.

Examples of replication events include:

• Start or stop of sp_cop or a replication process

• Execution of a command in sp_ctrl. User-issued commands are recorded for every SharePlex command that is issued.

  Note: A user-issued command appears in the Event Log as a notice, as in the following example:

  Notice 08-07-02 16:13:24.641582 23696 1 User command: rjones activate config 1route (from mycomp14)

• Database errors
• Failure of a network connection or SharePlex process
• Start or stop of a utility or script

• Login or logout of a user

Each entry in the Event Log includes:

• The date and time of the event.
• A description of the event and any related messages (error or non-error).
• The event’s process ID number, if it is associated with a SharePlex process.

To view the Event Log

Use the show log command in sp_ctrl or open the file named event_log in the log sub-directory of the SharePlex variable-data directory.

SharePlex provides a script for unattended monitoring of this log. For more information, see Monitor events with sp_eventmon.

Note: To control the number of out-of-sync messages that Post logs when a target table is very out-of-sync, use the SP_OPO_SYNC_LOG_FREQUENCY parameter. For more information, see the SharePlex Reference Guide.

Status Database

The Status Database contains a summary of the conditions reported in the Event Log, including events that did not generate an error message or warning at the sp_ctrl user interface. This information alerts you to potential problems and helps you resolve existing ones. The Status Database may refer you to the Event Log for a more detailed explanation of a warning, notice or event.

To view the Status Database

Use the show statusdb command in sp_ctrl or open the file in the data sub-directory of the SharePlex variable-data directory.

Error Log

When the Post process detects that source and target tables are out of synchronization, it logs the first 100 SQL statements and data for the out-of-sync transactions to an error file on the target system. You can use this log to determine the extent of the out-of-sync condition, and you can use the SQL statements to repair target tables if the condition is not too severe, after first correcting the cause of the problem.

To view the Error Log

Open the ID_errlog.sql file in the log sub-directory of the SharePlex variable-data directory (where ID is the identifier of the SharePlex target, for example a target database).

Process logs

When a SharePlex process cannot process a record, the process not only logs the record to the Event Log, but also to its process log file. The process logs are primarily for use in debugging.

The name of a process log consists of the datasource identifier (such as the ORACLE_SID), the short name of the process (such as ocap, ord, opo, rcl), the file number, and the file extension (.log).

Examples:

Capture: ora10_ocap02.log

Read: ora10_ord01.log

Post: ora10_opo03.log

Reconcile: ora10_rcl01.log

The aging of old log files is performed in a circular pattern. The numbering begins with 01 and ends with 03. Up to three logs can exist at any time, including the current one. When all three logs are full (50 MB), the process starts overwriting them, beginning with the oldest one.

To view a process log

Open the file in the log sub-directory of the SharePlex variable-data directory.

Activation log

When you activate a configuration, it generates a log.

To view the activation log

Open the file named SID_oconf##.log in the log sub-directory of the SharePlex variable data directory.

Compare/repair log

The compare and repair commands log errors, messages and warnings to a log. For more information about these logs, see the compare commands in the SharePlex Reference Guide.

Monitor with sp_ctrl commands

The information commands in sp_ctrl help you monitor different aspects of replication. Issue them frequently to:

• Monitor for out-of-sync tables.
• Verify that replication processes are running.
• View the number of replicated messages in the queues.
• View the Event Log to view warnings, errors and other notifications.
• View process statistics that are helpful for tuning and problem solving.
• Detect tables or operations that are slowing down the replication process.

List of information commands

See the SharePlex Reference Guide for details about these commads.

Run monitor scripts on UNIX

The SharePlex monitoring scripts notify you of events and conditions that can adversely affect replication on Unix or Linux systems. These scripts provide a monitoring mechanism without the need for frequent status checks through sp_ctrl. They can be run independently or through scheduled jobs.

SharePlex provides the following scripts:

• sp_eventmon monitors the SharePlex Event Log and reports errors that you specify in a special file.
• sp_logmon monitors how well Capture is keeping pace with the changes entering the redo logs. If Capture loses pace by a specified number of logs, sp_logmon alerts you before the logs wrap so that you can take corrective action.
• sp_ps monitors the SharePlex processes and notifies you if one or more are stopped so that you can correct the problem before the logs wrap or the queues exceed available disk space.
• sp_qstatmon monitors the status of the SharePlex queues and sends a warning if the backlog exceeds a threshold (limit) that you define. This enables you to take corrective action before the queues exceed available disk space and replication is adversely affected.

Important!

• These scripts run on Unix or Linux systems only.
• The monitoring scripts are overwritten with new scripts during patches and upgrades of SharePlex. Before you install the patch or upgrade, rename your existing scripts so that your customizing is retained. After applying the patches, update the new scripts with your customizing. Do not rename the existing scripts to replace the updated scripts, or you could lose important improvements or fixes.

Requirements for using the monitoring scripts

• These scripts must be run in the ksh shell.

• All monitoring scripts must remain in the directory where they were installed. All but sp_ps are in the .app-modules directory of the SharePlex installation directory. The sp_ps script is in the util directory of the installation directory.

• The scripts must be customized to reflect your environment, such as the type of e-mail or the paging available.

• To use the monitoring scripts, start sp_cop with the -uname name option, where name can be an identifier of your choice. Suggestions are:

  • the SharePlex port number
  • the ORACLE_SID of the instance for which replication is being monitored

  • the SharePlex Administrator’s name

• SharePlex must be running prior to executing a monitoring script.

• Verify the ORACLE_HOME (the path to the ORACLE_HOME directory) for each Oracle instance being monitored.

• The monitoring scripts make use of sp_ctrl commands. Before you use the scripts, make a link in the util sub-directory to the sp_ctrl binary in the bin sub-directory of the SharePlex product directory. Do not copy the binary itself, because that makes it difficult to maintain patches to sp_ctrl.

• Users of the monitoring utilities must have the following rights:

  • Local access to sp_ctrl and permission to run the script on the system on which the sp_cop to be monitored is running.
  • Korn (ksh) shell access and ps permission from the Unix or Linux command line.
  • Read, write and execute permissions to the directory where the scripts reside.
  • The permission on the iwgrep utility must be 755.
• The monitoring utilities use the mailx program to send e-mail notifications. Before using a script, make sure mailx is configured to send e-mail on all systems where the monitoring scripts will be deployed.
• Paging requires that your service provider supports receiving e-mails on your paging device.
• To kill any of the processes generated by these scripts, use the kill -9 command. The kill command alone does not kill all of the processes.