VCS ERROR V-16-1-50921 CONCURRENCY VIOLATION:Group <service group> is online on the following clusters [prod_clus, dr_clus]

book

Article ID: 100011254

calendar_today

Updated On:

Description

Error Message

VCS INFO V-16-1-10298 Resource SG1_Oracle_FD (Owner: Unspecified, Group: SG1_FD) is online on drnode01 (VCS initiated)
VCS INFO V-16-1-10298 Resource SG1_Netlsnr_FD (Owner: Unspecified, Group: SG1_FD) is online on drnode01 (VCS initiated)
VCS NOTICE V-16-1-10447 Group SG1_FD is online on system drnode01
VCS INFO V-16-1-10299 Resource SG1_Oracle (Owner: Unspecified, Group: SG1) is online on drnode01 (Not initiated by VCS)
VCS INFO V-16-1-10299 Resource SG1_Netlsnr (Owner: Unspecified, Group: SG1) is online on drnode01 (Not initiated by VCS)
VCS NOTICE V-16-1-10233 Clearing Restart attribute for group SG1 on all nodes
VCS ERROR V-16-1-50921 CONCURRENCY VIOLATION:Group SG1 is online on the following clusters [prod_clus, dr_clus]

Cause

Resources in disaster recovery fire drill service groups are usually created by duplicating the equivalent resources in application service groups. They may share identical values for the resource attributes that are used to determine resource state (Sid / Owner / Home attributes for Oracle resources, for example).

When such resources in the fire drill service group become online, it will appear (to the VCS agents) as if the corresponding resources in the application service group had also become online, outside of VCS control, making the application service group partially (or fully, depending on the configuration) online on the secondary cluster. If the application service group on the primary site was also online at the time, this would trigger concurrency violation error.

Note: Persistent resource types, such as NIC, do not cause the same problem. Their states are ignored when determining the state of a service group.

 

Resolution

On the DR cluster, enable the "FireDrill" attribute for non-persistent resource types that are configured in both the application service group and the fire drill group with identical resource attributes:

 


Run the following commands from any node in the secondary cluster:

haconf -makerw
hatype -modify [resource_type] FireDrill 1
haconf -dump -makero

 

For example:

# haconf -makerw
# hatype -modify Oracle FireDrill 1
# haconf -dump -makero

 

 

When the FireDrill attribute is set for a specific resource type, the VCS agent framework keeps track of the resource being brought online (via VCS) and prevents other resources of the same type (with identical configuration) from being detected as online.

 

Notes:

  1. FireDrill is a type attribute; modifying its value affects all resources of the same type in the cluster, effective immediately (does not require a cluster restart) and maintained across future restarts.
  2. The FireDrill attribute should not be enabled for persistent resource types (CoordPoint, Disk, ElifNone, FileNone, MultiNICA/B, MultiPrivNIC, NIC, PrivNIC, Proxy). Enabling it for persistent resource types may result in incorrect resource states and/or unexpected resource faults.
  3. While the FireDrill attribute is set to 1 for a resource type, VCS will be unable to detect when a resource of said type becomes online outside of VCS (manual startup by user, for example). If this behavior is undesirable during normal operation, turn on FireDrill only during fire drill activities. Set the attribute to 1 just before bringing fire drill service groups online. Then, set it back to 0 once fire drill service groups are taken offline at the end of the activity.

Issue/Introduction

Concurrency violation error is triggered when fire drill service group is brought online on secondary site.
In the engine log, some resources in the application service group (previously offline) are reported as online ("not initiated by VCS") on the secondary cluster node on which the fire drill service group has been brought online.
 

Additional Information

UMI: V-16-1-50921
Powered by Wolken Software