IP Address Caching

,
1

Hi,
Need help in resolving IP Address caching issue in webMethods 7.1.3 version.

Environment - webMethods 7.1.3
Integration Type - webService/SOAP URL/HTTP Call
Background: Downstream application runs either on PROD server and Recovery Server.
Our application (webMethods) is invoking downstream application Virtual IP URL (HTTP Call). Request will be directed to active server (PROD or Recovery). Redirecting is taken care by external LoadBalancer.

Issue: Whenever downstream application switches from PROD Server to Recover Server and vice versa, webMethods is getting SOCKETEXCEPTION error. However, connectivity works fine after restarting webMethods server.

Additional Information: CacheResults property at FlowService level has been set to FALSE.
Before restarting the webMethods server, Able to ping the downstream application servers IPAddress/port from webMethods UNIX box.
Before restarting the webMethods server, Able to telnet the URL from webMethods UNIX box.

Please let me know the possible solution if anyone has come across the similar situation.

2

My guess is that, since you use virtual IP, the WM will try to reuse the existing connections that are still open pointing to this downstream system.

you can try to not reuse outbound sessions by setting
newSession to true when call pub.client:http service.
or reduce the server’s outbound session timeout at: Settings > Resources page.

3

Yes the IP cache is a known issue but if you really want to turn of cache you can edit file jre/lib/security/java.security

the param is networkaddress.cache.ttl to greater than 0 (60 or 120) and restart all the affected IS’s

HTH,
RMG

4

Hi RMG,
Thanks for the response!
I would like to turn-off cache. So, I need to set networkaddress.cache.ttl=0.
If I want to refresh the cache after certain time, I need to set networkaddress.cache.ttl > 0 (60 or 120).

Hope my understanding is correct.

5

How is the detection of the outtage and the switching to the recovery system setup.

In my project we have installed the servers in SAN and when the outtage occurs the server is tried to shutdown.
Then the SAN-Mountpoint will be re-mounted on the recovery side and the server startsup.

The server(s) are configured to use the virtual addresses where applicable, as these will switch to the recovery along with the mountpoint.

Only the outbound connections will use the direct address of the respective machine.

We never had problems with this setup in the last 7 years.

Regards,
Holger

6

No to turn off the cache set networkaddress.cache.ttl > 0 (positive number)

To remain in cache then networkaddress.cache.ttl=-1 (negative number)

NOTE: setting this to anything other than the default value can have

serious security implications. Do not set it unless

you are sure you are not exposed to DNS spoofing attack.

HTH,
RMG