Automated restart

In a Db2® pureScale® environment, Db2 cluster services automatically detects software and hardware failures, and initiates either a member restart or a group restart, depending on the type of failure that has occurred.

Automated restart helps ensure that host, member, or cluster caching facility failures have a minimal affect on the database. A member failure (and the subsequent restart) is transparent to applications and only temporarily affects uncommitted transactions running on the failed member. Applications can continue to access databases despite multiple host or member failures.

In situations where Db2 cluster services cannot restart the failed member on its original host, the failed member is restarted on another host in a process known as restart light. Some extreme failures, such as the loss of all cluster caching facilities in the Db2 pureScale instance, do result in a database outage. In these cases, Db2 cluster services initiates a group restart of the cluster caching facilities and members.