Failover design clarification question
we are testing our failover design, and I am looking for some clarification.
when you use an approach like documented here https://jazz.net/library/article/722
and you have a duplicated environment for the parts in the marked box..
if A component in the production complex fails (JTS), do the OTHER components (RM, QM, CCM) also need to be restarted? (in this case, the license server fell out from under the apps).
how about when RM fails and we fallover its app server?.. do the 'friends' need to be restarted? any open connections are gone..any transactions in flight died.
So I would expect the 'system' to be in an indeterminate state.
says
Configuring high availability for both primary and backup servers
Note: Examples in this section are for the jts.war application. You can use the same procedure for the ccm.war and qm.war applications.
but it does not say what to do across applications.
|
One answer
First we need to understand the constraint that only one instance of an application can be running at any given time. This is to ensure that there are no conflicts or corruption in the DB by having two applications updating the DB at the saem time.
If rm goes down, the failover instance can take over, but it needs to be able to communicate to the primary jts server. SInce both the primary and failover servers are pointing to the same DB, all the AOF / friendships should be the same. The one thing I am not sure about is if the friendships for the /admin app wil be the same. This is because they are kept in an xml file under the /admin directory. Well this depends on how the failover servers are created....if the files will be copied from the primary, this should not be an issue. |
Your answer
Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.