The most serious concern when recovering from an abnormal system shutdown is whether or not all write-required frames have been written back to disk from memory. The amount of time it takes to write these frames back to disk varies depending on the amount of memory the system has.
On large systems it could take as long as a few minutes before all write-required frames are written back to disk. If there is even a chance that they have not been written to disk, then you must take care to reboot in such a way as to ensure that system integrity is preserved.
This section discusses the different kinds of abnormal shutdown and methods of recovery.
Power Failure
If power is cut off, processing will be interrupted and will not continue when the power comes back on.
If the mvBase Server is stopped by a power failure, and if your system is equipped with properly installed SPM (Standby Power Module) or UPS (Uninterruptible Power Supply) power-fail equipment, the system will perform an orderly shutdown with no data lost, provided that the battery backup is functioning properly.
Using the Powerfail Monitor
mvBase 2.0 patch set 6 introduces a new strategy for handling powerfail monitoring. Prior to this release, the powerfail software running under the Operating System would call the mvpfail.exe program in the mvBase root directory and both the mvBase server and workstation processes would be closed without waiting for virtual. Under patch set 6, the mvpfail.exe program issues a :SHUTDOWN to properly shut the mvBase virtual down from within virtual. This clean shutdown prevents GFEs from occurring during a power failure (provided there is sufficient battery power to complete the shutdown). For more information on issuing a :SHUTDOWN, see the topic :SHUTDOWN Proc in the mvBase Guide to mvBase Files and Accounts.
Complete the following general instructions to configure the Powerfail Monitor.
Configure the Windows power monitoring software to run the mvpfail.exe program located in the mvBase root directory.
Activate the mvBase Powerfail Monitor. delete the POWERFAIL item from the ERRSMG file.
Execute the following command:
|
RUN SPL POWER.FAIL.SETUP |
The Powerfail Monitor is activated. If a power failure occurs, the program uses the process prior to the Spooler process to issue a :SHUTDOWN. For example, if the highest process number running on mvBase is 16, the Powerfail monitor will run on process number 15.
If you need to configure the Powerfail program to use a specific mvBase process (for example, in a case where the process prior to the Spooler is being used by a background process), set the value of attribute 1 in the POWERFAIL item in the ERRSMG file to the process number to be used. In such situations, a new DWORD value named POWERFAILPROCESS must be created under the mvBase/ Server/ Boot registry key and then set to the same process number specified in the POWERFAIL item in the ERRSMG file.
|
CAUTION |
Do not make changes to the registry if the default mvBase process is used (last process before the spooler). This configuration pertains only to systems where the last process before the spooler is already in use. |
Once the patch set containing the POWERFAIL feature is loaded, an Information (type 'I') report is made to the SYSTEM-ERRORS file. The report indicates whether or not the Powerfail monitor has successfully started and, if started, the process on which it is running.
|
NOTE |
To de-activate the Powerfail Monitor, re-create the POWERFAIL item in the ERRMSG file and set its value to the default value of 0 (zero). |
Checking for System Failure
In order to recover from an mvBase system failure, you must first determine that it has really stopped processing.
Check to see whether the mvBase system will respond to input from the keyboard. Enter the following in succession:
BREAK
CTRL+BREAK
CTRL+B or CTRL+Q.
If the mvBase system does not respond, check the Windows console and make a note of any messages that may be displayed there. (A message may or may not display on the system console when the system fails.)
If the mvBase system does respond, no failure has occurred, and you should continue looking for the cause of the processing halt elsewhere.
Windows Application Event Log
The Application Event Log provides a means to observe information, warnings and messages. This is a helpful method for displaying records of and tracking down problems.
See your Windows documentation for additional information on using the Application Event Log.
Recovery Procedures
After restarting the mvBase system, several procedures may be used to recover after an abnormal shutdown:
|
If your database is severely corrupted and cannot be repaired, you may need to reload a clean database by performing a fileload from a filesave tape. |
|
|
Allows the system to recover from a failure in which it was not possible to write memory frames to disk by removing frames from the Overflow Table so they do not cause GFEs. |
|
|
Should files become corrupted as a result of any system anomalies, an ABS/Kernel restore procedure must be performed from mvBase source files located on either CD-ROM or floppy diskette media. |
See Also
Starting and Stopping the mvBase Server
Starting the Server for the First Time
Starting the Server as an Application
Starting the Server as a Service
Restarting from Disk: The Coldstart Process
Starting the Server in the Recovery Mode
Stopping the Server via the Server Tab