Dell EMC Metro node 7.1 Administrator Guide

Detach-rule

Detach rules are a consistency group’s policy for automatically picking a winning cluster when there is an inter-cluster link outage.

For metro node Metro configurations, there are two consistency group detach rules:

• no-automatic-winner - The consistency group does not select a winning cluster.
• winner cluster-name delay seconds - The cluster specified by cluster-name is declared the winner if an inter-cluster link outage lasts more than the number of seconds specified by delay.

If a consistency group has a detach rule configured, the rule applies to all volumes in the consistency group and overrides any rule-sets applied to individual volumes.

This property is not applicable for local consistency groups.

By default, no specific detach rule is configured for a consistency group. Instead, the no-automatic-winner detach rule is set as default for a consistency group with visibility to both clusters.

Best practice is to apply detach rules to a consistency group that meet the needs of your application in terms of I/O continuance and data loss tolerance.

Use the consistency-group set-detach-rule commands to configure the detach-rule for a consistency group:

• Use the consistency-group set-detach-rule no-automatic-winner command to set the detach-rule as no-automatic-winner:

  VPlexcli:/clusters/cluster-1/consistency-groups/TestCG> set-detach-rule no-automatic-winner

• Use the consistency-group set-detach-rule winner command to specify which cluster is the winner, and the number of seconds metro node waits after a link outage before detaching the winning cluster:

  VPlexcli:/clusters/cluster-1/consistency-groups/TestCG> set-detach-rule winner --cluster cluster-1 --delay 5s

The following table describes the detach rule behavior for synchronous consistency group.

Note the following:

• Active I/O indicates active writes.
• The detach rule behavior described the previous table are based on the assumption that there is a healthy leg in the winner cluster, during the time of cluster partition.
• Use the consistency-group resume-after-rollback command to resume after roll back.
• In case of detach rule no-automatic-winner, to resume I/O, you must manually choose a cluster as the winner. Use the consistency-group choose-winner command to choose a winner.