Before updating dbspaces, see the overview of dbspaces and dbfiles in “Configurable tablespaces” in Chapter 1, “Overview of System Administration,” in System Administration Guide: Volume 1.
In multiplex, CREATE DBSPACE, ALTER DBSPACE and DROP DBSPACE commands for main and catalog dbspaces are only permitted on the coordinator. Due to delays in TLV propagation and version maintenance, there may be a timing window of up to a few minutes before you can drop an empty dbfile or dbspace. A “Command not replayed” error will occur during this timing window. In order to perform the DROP DBSPACE or ALTER DBSPACE DROP FILE commands, the OkToDrop column reported by the sp_iqdbspace and sp_iqfile procedures must indicate “Y.”
The following rules affect multiplex dbspace updates:
Only the coordinator node is allowed to manipulate shared IQ dbspaces.
Secondary servers must be in synch before you run ALTER DBSPACE ALTER FILE RENAME PATH. Synchronization happens automatically through TLV log replay, but there may be a time delay of up to two minutes between the previous dbspace operation on this dbspace (create or alter) and the time the rename path can be run. If secondary servers are out of synch, you may see a “Command not replayed” error.
Temporary dbspace files must be created/altered/dropped from the node where they reside. Syntax is identical to simplex temporary dbspace syntax.
If any shared IQ main dbspace files have paths that are not accessible from a secondary node, that secondary node cannot access the file or any contents of that file until the path is corrected. See “Accessing the shared disk array”.
