Hello to all.
I have a IntegrationServer 7.1.2 working normally,
my doubt is, if it’s possible to connect another IntegrationServer
of the same version into the same data base of the product and work both together?
Yes you can share DB but you would have to cluster them.
I have seen a lot of people do this without clustering the IS. I am not sure what the Advantages and Disadvantages are to connecting multiple ISs to the same DB.
Can anyone shed some lights on this? I think lot of people did this in 6.5 and the reason was schedulers and issues with them. Im so not sure.
Our systems are configured same database across three IS Servers, we do not any problems but we noticed following things.
if you create any scheduler(public queue/private queue), all these jobs will appear in all servers and try to execute from all servers. you might have create wrapper services in order to restrict to run on one server incase if you need. Some time we have seen scheduler page was hung but not sure this was happend because of sharing same database.
I am also trying to see there any other issues? Is this is good architecture?
We had same dilama to use Clustering or not…
Software AG suggested to consider clustering only if session state is needed in case of clnats that are cluster aware…
IS 7.1.2
i have configured 2 IS same way I would do for the cluste but without cluster enavbled and perforemd 4 tests:
- communication with BROKER
We had same dilema to use Clustering or not…
Software AG suggested to consider clustering only if session state is needed in case of clients that are cluster aware…
IS 7.1.2
I have configured 2 IS same way I would do for the cluster but without cluster option enabled (check clustering guide for 7.1.2) and performed 4 tests:
- communication with BROKER
ALL OK as 2 IS have same CP - resubmitting of procesess
Also ALL OK, resubmit is working like a charm… once on 1 node other time on other node… also when one node is down it is working fine. - Schedulers
With 7.1.2 there are new options for schedulers $any $all
default is $any. so the scheduler will run onlu ONCE
you can als specify local node and in that case it will run localy on the specified server. $all is not available as there is no cluster yet… - Guaranteed delivery…
it is still not clear how to really connect and at this time Core fix 11 is needed… we have lots of problems but here some of the test results:
- you need to specify the Retry server to be 2nd node of the cluster but there we have problem with 20s delay in case of server down/failover. SR open with Software AG… failover as such is working… it is always connecting to the First server in Remote Alias. (same issue when cluster is enabled GD not so good at the moment with 7.1.2)
So you don’t need cluster if you only want these 4 features to work in HA environment ! ! !
br
Nenad