Kbase P110704: Why may clients get disconnected during fail-over of a Windows cluster?
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
18/03/2010 |
|
Status: Unverified
GOAL:
Why may clients get disconnected during fail-over of a Windows cluster?
FACT(s) (Environment):
Windows
Progress 9.x
OpenEdge 10.x
FIX:
In case a failover condition occurs within Microsoft Cluster Server (MSCS), resources within the failing group(s) will be moved to another available node within the cluster. The time required to bring resources online on another node (failover transition time) may vary based on resource types and other factors, such as the type of resource, it's properties and dependent resources etc. MSCS needs to detect that the primary active node has failed before commencing the failover transition process to the next available node of the cluster, this is done in less then 10 seconds.
In the process of starting a Progress database resource on the failed-over cluster node, a Progress database need to ensure integrity, i.e. roll forward the transactions within the before image file etc, this will of course increase the failover transition time.
The AppServer / WebSpeed etc. brokers will need to connect to the Progress database so any connections to these brokers may be delayed until the database(s) are up and running.
The clients may need to re-connect due to the transition time of the particular failover scenario.