If you select REPOS(YES), you should check that all cluster sender channels
in the relevant cluster are inactive or stopped before you issue the REFRESH
CLUSTER command. If there are cluster sender channels running at the time
the REFRESH CLUSTER command is processed, and they are used exclusively by
the clusters being refreshed, and REPOS(YES) is used, the channels are stopped,
by using the STOP CHANNEL command with MODE(FORCE) if necessary. This ensures
that the refresh can remove the channel state, and that the channel runs with
the refreshed version after the refresh has completed. If a channel's state
cannot be deleted, for example because it is in doubt, or because it is also
running as part of a another cluster, its state will not be new after the
refresh, and it does not automatically restart if it was stopped.
Selecting
REPOS(YES) on the sole working full repository queue manager in a cluster
requires you to alter it to be a partial repository, with the result that
there would be no full repository left. After the queue manager is refreshed
and then restored to its status of a full repository, the other partial repositories
would also have to be refreshed to restore a working cluster. If there is
another working full repository in the cluster, it informs the other members
of the cluster that the full repository executing the REFRESH CLUSTER command
has resumed its role as a full repository, and no further action is needed.