In “Recovery from loss of a multiplex server”, we specified a write server failover procedure when all the necessary disk resources could be made available on a new machine. If some of those resources are unavailable following a failure, and you choose not to use the replace write server procedure on §, it is necessary to fall back on standard database restore procedures. For details on restoring databases, see Chapter 14, “Backup and Data Recovery”. For information about how to diagnose database startup problems, verify the consistency of databases, and repair databases, see the Sybase IQ Troubleshooting and Error Messages Guide.
You may choose to protect your multiplex database using either database or OS-level backups. Certain disk-mirroring schemes available with shared-disk subsystems (such as the Timefinder or SRDF facilities from EMC) provide a very fast OS-level disk backup/restore service.
If restoring the database from IQ-level backups entails moving one or more dbspaces of the shared IQ store to new device(s), you may need to use the RENAME clause of the RESTORE command. Following this, you may then have to delete and re-add the query servers in order for them to be able to access the new device(s).
