Troubleshooting
Problem
How can we deploy a fixpack to a cluster whilst keep down time to a minimum.
Symptom
Typically customers bring down all the application servers in the cluster and
apply Fix packs. This may have extended down time.
Environment
A Clustered deployment of Sterling Order Management
Resolving The Problem
1) Install Fix Packs on the nodes of the cluster (without stopping the application server) after setting the below properties in <install>/properties/sandbox.cfg :
NO_DBVERIFY=true
REINIT_DB=false
LOAD_FACTORY_SETUP=false
Note : Any change in sandbox.cfg should be followed by running 'setupfiles'.
This will ensure that the database is 'not touched' during FP installation. They can build new EARs on each node but don't deploy them yet. It is advised to revert these properties after Fix Pack installation has completed.
2) Now, when you reach the last node of the cluster, bring
down application servers of all the nodes including the last node. Install Fix
Pack on this last node, without the above properties in sandbox.cfg.
This will make all the necessary changes in the common database.
3) Now start the application servers in all the nodes with 'new EARs'. These new
EARs will continue to point to the database, which has now been updated.
With this approach, the downtime of the clusters is reduced to the time
taken in 'Fix Pack installation and build EAR on the last node' only,
and all the nodes are in sync
Was this topic helpful?
Document Information
Modified date:
10 May 2022
UID
swg21883218