How to recover from dual active state for a backup broker de

Posted by smahanta on 28-Jun-2010 08:13

We are using SDM in our production environment.

Earlier we did not enecounter any problem in starting the active & back up broker. But nowadays though the acive broker starts alright, the backup broker never starts.The container log - the container housing the back up broker - says "Dual active detected, local saved state STANDALONE & peer saved state STANDALONE".

Without running a clean installation script, is there any way to solve this problem?

Pl. help.

Regards,

Subhendu

Technology Partner - Forum

All Replies

Posted by jerem on 01-Jul-2010 10:33

hello Subhendu,

You can find your answer in the knowledge Base. hope this help

Be very carefull because you may lost some messages  when you will reinitialize broker's storage.

you can also read

ID: P137832

Title: "The Sonic tools do not allow easy recovery from a dual active scenario."

Regards

Chan Jeremy

ID: S267
Title: ""Dual active detected, local saved state STANDALONE and peer saved state STANDALONE." error in container log file"
Created: 02/02/2007Last Modified: 05/24/2010
Status: Verified


Symptoms: 
  • The container log file shows the following error:
  • Dual active detected, local saved state STANDALONE and peer saved state STANDALONE.


Facts: 
  • Using a Fault Tolerant pair of brokers
  • SonicMQ
  • All Supported Operating Systems


Cause: 
  • If a Fault Tolerant pair of brokers (primary and backup brokers) is configured, then if a network failure occurs the following error is thrown when the brokers are reconnected: (severe) Dual active detected, local saved state STANDALONE and peer saved state STANDALONE.
    (info) Closing down all replication connection(s) What happens is that when there is a network failure, and after "Failure Detect Timeout" (see note below) expires on the standby (and after all replication connections have been lost) then the standby fails over and both brokers transition to the STANDALONE state. In this case, both brokers are referred to as dual-active, dual-standalone, or partitioned brokers, a condition where the two brokers are isolated from each other yet still running and accepting the active role. Broker negotiation will force both brokers to stop. When both brokers are accessible to clients, two clients can perform inconsistent operations on separate brokers that cannot be reconciled after the brokers reconnect.


Fixes: 

The broker disparity can only be resolved by stopping both brokers and initializing one of the databases.

Please refer S1213,"How to create or reinitialize a broker's storage" on how to initialize the broker's database.

The note from the SonicMQ Deploym

This thread is closed