A second point of failure for a high availability system occurs when the primary companion is already in failover mode, and another point in the system fails. Sybase Failover includes dbcc ha_admin, which addresses second points of failure.
See “dbcc options for high availability systems” for information about dbcc ha_admin syntax and a complete list of options.
After you run installhasvss on a companion server, you should re-run this script only if the stored procedures it creates are corrupted, or to install a newer version of installhasvss. dbcc ha_admin (' ', state_machine) temporarily moves the companion to single-server mode so installhasvss can safely reinstall or update the stored procedures. If you attempt to run installhasvss without running dbcc ha_admin, the companion issues the following error message:
Server is not in single-server mode.
Please run dbcc ha_admin (' ', 'state_machine', 'halt') and try again
Because dbcc ha_admin moves
the companion to single-server mode, run this command only when
there is no concurrent activity.
Rerunning installhasvss:
Make a note of the srvnetname for the SYB_HACMP entry in sysservers. When it is configured for Sybase Failover, SYB_HACMP points to the companion server’s svrnetname (for example, the srvnetname for the SYB_HACMP entry on companion server MONEY1 is PERSONNEL1).
Run dbcc ha_admin to move the companion to single-server mode:
dbcc ha_admin (' ', 'state_machine', 'halt')
Where ' ' is used as an empty placeholder.
Re-run installhasvss. After installhasvss finishes, the companion server reverts to its original mode.
If the node crashes after you perform step 2, above, the srvnetname of the remote server is removed from sysservers. If this occurs, add the name of the remote server to sysservers by issuing:
sp_addserver SYB_HACMP, null, 'remote_server_svrnetname'
Run dbcc ha_admin to return the companion server to its original mode:
dbcc ha_admin (' ', 'state_machine', 'restart')
| Copyright © 2005. Sybase Inc. All rights reserved. |
|
|
