SavWareHA Checkout

After SavWareHA has been installed and configured, a trial fallback transition should be tested.

Use the SavWareHA program to verify that the system is in normal mode (upper right hand corner of screen). Next, go into the ‘Utilities’ menu and display the monitor status. The monitor status should be normal. Automatic fallback will only occur if the monitor status is normal! The monitor will be PAUSED during a regeneration and should return to NORMAL when the regeneration is complete. The status of the mirror should be ACTIVE / REGEN (regeneration is in process).

Use the mount command on the Primary System and insure that the user filesystems are mounted.

Error Handling & Notification

Whenever any error condition occurs, mail is sent to each user listed in the ‘/etc/sentinel.d/administrators’ file. Users can be at a remote location if the e-mail system is correctly configured for the system to pass messages in this manner. In addition, pagers may be activated automatically if ‘mmdf ’ and ‘uucp ’ have been properly configured.

The administration message handling should be tested once SavWareHA configuration is complete. This can be done by using the ‘Utilities: Force Fallback Mode’ selection to manually force the system into fallback mode . Once the switch to fallback mode is complete, the system administrator(s) should receive mail describing the transition to fallback mode.

Final Testing

Terminals, printers, and network devices on the IO link should be tested on both the Primary System in normal mode and the Standby System in fallback mode to verify operation under normal and fallback modes.

To perform a fallback and recovery test, wait for the regeneration to complete. On the Standby System , log on to ‘root’, enter the ‘sentinel’ command and observe the status of the system; it should state that the mirror is ‘Controlled by Primary System ’. Issue two ‘sync’ commands on the Primary System, then press the ‘reset’ button on the Primary System (This is the preferred method in place of powering off the Primary System).

Within 30 seconds, the reader should see that the status of the Standby System has changed to ‘Fallback Mode’; the mirrored unit will be undergoing an ‘fsck ’; and finally the mirror should be mounted on the Standby System, and all devices and logins should be available from the Standby System. Correct any login, printer, or other device errors at this time.

To recover, reboot the Primary System . The operating system will automatically prompt the user for an ‘fsck ’ of the root filesystem ; perform this and clear the possibly many zero length files that ‘fsck’ prompts for clearance. These zero length files are temporary ‘pipe’ files that TCP/IP uses during operation, and may be safely discarded.

Enter ‘sentinel’ on the Primary System . When ready to recover, select ‘Utilities: Recover from Fallback to Normal Mode’. This will prompt the user with another question to authorize the recovery; enter ‘y’. The Primary System then:

• Sends a message to the Standby System
• Instructs it to stop all users on the mirrored unit
• Unmounts the mirror from the Standby System
• Remounts the mirror on the Primary System
• Starts up daemons and users on the Primary System

The SavWareHA system will be regenerating data from the Standby over to the Primary system; all reads will be done from the Standby System until regeneration is complete.

When regeneration is complete, the read flag is turned ‘off’ on the Standby and ‘on’ on the Primary System . The SavWareHA system is back in compliance.