The following table summarizes the issues that were resolved in WebLogic SIP Server 2.2.
When using a replicated WebLogic SIP Server installation, the default execute queue configuration for data tier servers poses a risk that garbage collection pauses in engine tier servers will cause delays in servicing other engine tier servers.
To minimize this risk, on each data tier server set the thread count for the weblogic.kernel.Default queue to twice the number of engine tier servers in your deployment. You can set the thread count in the Administration Console by following these instructions:
|
|
The Administration Console lists the
sipserver implementation application as a standard J2EE application, and allows a Console user to redeploy or even remove the application from a running WebLogic SIP Server installation. The sipserver application must never be undeployed or redeployed except indirectly via the ConfigManagerRuntimeMBean . Redeploying the application yields several nested exceptions starting with InstanceAlreadyExistsException , and forces running data tier server instances to shut down.
To avoid these problems, never redeploy or undeploy the
sipserver application using the Administration Console or weblogic.Deployer utility. Perform all engine tier configuration changes using the SIP Servers node in the Console or the WLST command-line utility, as described in
Configuring Engine Tier Container Properties.
|
|
Beginning with WebLogic SIP Server 2.2, the "replicated" domain deploys the
sipserver implementation application using the default "stage" mode, rather than the "nostage" mode used in previous releases. With stage mode deployment, if you manually edit a configuration file (sipserver.xml , datatier.xml , or diameter.xml ), you must explicitly redeploy the config Web Application component in sipserver to all target server instances. See the instructions under
Default SIP Servlet Configuration.
Stage mode deployment also changes the procedure for applying patches to WebLogic SIP Server. After applying a patch as described in
Applying Patches Using InstallPatch, you must either:
|
|
The Diameter Sh client application included in WebLogic SIP Server 2.2 uses threads from the
sip.transport.Default execute queue. Because this queue is also used for general SIP message processing, applications that use the Sh interface may experience poor performance with the default execute queue settings. To work around this problem, increase the number of threads available in the sip.transport.Default queue to a large number (for example, 200 threads).
The Sh client application may also consume additional threads in
sip.transport.Default if the HSS is unavailable. This problem occurs because the Sh application uses a large default timeout value (30 seconds) when waiting for a response from the HSS. Using a smaller timeout value (for example, 1 second) ensures that available threads are not quickly consumed when the HSS is unavailable.
To change the
timeout value, edit the diameter.xml configuration file for engine tier servers to configure the timeout parameter. For example:
... |
|
In the profile service API, the
SipApplicationSessionAdapter is not recreated after a call state has been restored. This means that if a Servlet on a particular engine creates a profile service subscription and the server subsequently fails, another engine tier server that recreates the necessary call state cannot obtain the session with ProfileSubscription.getApplicationSession() . Attempting to recreate the session throws:
|