Operating SavWareHA

Booting SavWareHA

Standby System Bootup

When a SavWareHA system is started from a power-off or complete shutdown status, the Standby System should be started first. The Primary System will automatically look for a response from the Standby System telling in what mode (‘normal’ or ‘fallback’ ) the Primary System should be immediately after the Primary System comes into multi-user mode . If the Standby System is not in full multi-user mode when the Primary System comes up, the SavWareHA system will assume (correctly) that the Standby System is unable to handle mirroring requests, and therefore mirroring over to the Standby System (the slave side) will be disabled (the slave side is turned ‘OFF ’ ).

Primary System Bootup

After the Standby System has come up into multi-user mode, the Primary System should then be started The booting sequence for the Primary System will proceed as usual. The Primary System will check the Standby System's mode during the transition to multi-user mode, and compare it with its own mode. If the Standby System is in fallback mode and the Primary System is in normal mode , the Primary System will automatically switch itself to fallback mode.

Access of Filesystems

Access to the mirror , and to the master and slave sides of the mirror, must be done through the mirror device node (OpenServer 5 -/dev/mirror00, UnixWare 7 - /dev/mirror000). Access of either side directly while the mirror is mounted will immediately compromise the mirrored filesystem (s).

If access to one side only is desired for some reason, make certain to turn that side ‘OFF ’ with the ‘Fix Broken Mirror: Turn Off ..... Element’ selection. After the desired side’s status has been set to ‘OFF’, the side may have ‘fsck ’ run against it, and then may be mounted. Be aware, though, that once the mirror is brought back into full operation, the side marked as ‘OFF’ will be overwritten with the other side’s data.

Shutdown & Maintenance

Normal Shutdown

The proper procedure to shutdown a SavWareHA system is to issue the ‘shutdown’ command from the Primary System . During the shutdown, the Primary System will turn off the monitor program on the Standby System so it doesn't think there is a Primary System failure . The Standby System will continue to run but it will not take control of the users or filesystems once the Primary System has been shutdown. The Standby System can then be shutdown with the normal shutdown command on its console.

Taking the Primary System Off-Line

The Primary System can be taken off-line for maintenance, updates or diagnostics by manually issuing the ‘Utilities: Force Fallback’ command from the SavWareHA ‘Utilities’ sub-menu on either the Primary or Standby System. This will cause the Standby System to take control of the users and the filesystems. Once control has passed to the Standby System, then the Primary System may be shutdown gracefully, and taken off-line. See the ‘Utilities Sub-Menu’ section.

Alternatively, if the installation has ‘hot swap ’ disk drives, the system administrator may desire to perform maintenance on the drive or drives without interrupting processing with a fallback transition . To accomplish this, use the ‘Fix Broken Mirror: Turn Off Primary Element’ option on the mirror in question. This will mark the mirror as ‘OFF ’, and allow removal / insertion of drives. Bring the mirror back into compliance with the ‘Fix Broken Mirror: Regenerate Mirror’ option on the mirror in question.

Placing the Primary System Back On-Line

When all off-line maintenance is complete on the Primary System , boot it into multi-user mode. The Primary System will access the Standby System , see that it should be in fallback mode , and set its mode accordingly. Log in to ‘root’ on the Primary System and enter the command sentinel. Use the ‘Utilities: Recover from Fallback to normal’ option. Since this is "planned downtime ", the system administrator should first notify all active users that a transition back to normal mode will take place shortly. All users will be evicted from the Standby System during this procedure; once the transition is complete and a regeneration started, the users may log in to the Primary System and continue without further interruption.

Taking the Standby System Off-Line

The Standby System can be taken off-line to perform maintenance, updates or diagnostics by turning off the Standby System mirror element with the ‘Fix Broken Mirror: Turn Off Standby Element’ option, and then issuing the ‘shutdown ’ command on the Standby System. Data mirroring will be stopped and the users will continue uninterrupted on the Primary System . Applications running on the Primary System will not be affected in any manner.

Placing the Standby System Back On-Line

When the off-line maintenance is complete on the Standby System , boot it in to multi-user mode. Log in to ‘root’ on the Primary System and run the command sentinel. Select the appropriate mirror number, and use the "Fix Broken Mirror: Regenerate Mirror " selection from the mirror sub-menu. This will start a regeneration of the data to the Standby System. Users will continue uninterrupted on the Primary System during the regeneration process.