This section describes how to detect failure in the SAP system services, what the effects of failure are, and how to recover from failure of SAP services.
For a different view of SAP system failures, see Component Failure.
The SAP service processes can fail due to any one of the common causes outlined “Standard Failures” in System Failure (SAP NetWeaver AS).
SAP System Service Failures: Frontend, Dispatcher, Dialog, Update, and Enqueue
SAP System Service Failures: Background, Message, Gateway, Internet Communication Manager (ICM), Spool, Database, and Java Message Service
Consider using Switchover Software
You can use switchover software for the database host and the application host on which the enqueue and message services are running. Automatic failure detection and automatic restart (on a standby host) are thereby ensured.
For more information about using switchover to protect the enqueue service, see Replicated Enqueue Server.
In a standard SAP system, the enqueue lock table is held at a single location. If the enqueue service fails – for example, if the host supporting it crashes – then the information in the table is lost. This means that, when the enqueue service is restarted, all transactions that obtained locks from the enqueue table must be restarted.
Application services constantly track the enqueue table. When a new, initialized version of the enqueue table appears following failure, all relevant transactions are automatically restarted. This feature is relevant to all standard, distributed SAP systems but especially to high availability SAP systems using switchover software. It has the following advantages:
· After enqueue service failure:
Ў You no longer need to restart the hosts supporting application services, so saving time and effort.
Ў All local SAP data caches are preserved and the performance of application servers remains at optimal levels for all users.
· Switchover administration and switchover scripts become much simpler. For more general information about switchover, see Switchover Software for High Availability.
For more information, see Replicated Enqueue Server.