Imagine we want to do a "poor man's" HA because the weblogic clustering license is too expensive. On ServerA we have one OBIEE stack (MWHome), and on ServerB we have a second OBIEE stack. The 2 weblogic servers are not aware of eachother. There is no "Scale-Out". Is it possible to point the 2 environments to the same WebCat? Only one environment is active at a time, and then we switch over to the second if there is a problem with the first. Or will this cause problems in 11g because of the whole user/role GUID nightmare?
Thanks RM. I'm fine with the part about creating a share that both machines can mount. What I'm more worried about is corruption of the webcat or issues with GUIDs being out of sync. There are 2 non-clustered WebLogic/BI instances that are unaware of eachother. Right now what I'm seeing is that I can log in with "weblogic' to one of the environments, but not the other. I followed the "regenerate Guids" procedure with the tags in InstanceConfig.xml and NQSConfig.ini.
it has got nothing to do with the corruption webcat or refresh the GUID's. See you have 2 env on 2 servers. So both will have each id as weblogic So on first machines if it works and on 2nd server u r not able to login using weblogic so check thru the logs as why the authentication fails for weblogic.
I guess I'm giving bad details. Sorry. Here is a bit more background:
on Environment1, weblogic user can log into FusionMiddleWareControl, AdminConsole and OBIEE application
on Environment2, weblogic user can log into FusionMiddleWareControl, AdminConsole, but cannot login to OBIEE application
Originally it, could log into obiee application on Environment2. It's only after I deployed the WebCat (shared with env1) in environment2 that I started to have the login problem. The error in the log is invalid username/password. The reason I suspect the WebCat is because the problem only started occurring when I deployed the webcat.
I got you. Conside Env1 as Dev and Env 2 as Prod/QA/test So when you migrate the webcatalog you just take the catalog from Dev and put it in QA/Prod/Test So that doesnt effect login into OBIEE application using weblogic ID. So there might be something else like the bisystem user got out of sync or something else. So go through the logs related to presentation services(/MiddlewareHome/instances/instance1/diagnostics/logs/OracleBIPresentationServicesComponent/coreapplication_obips1) and see what is the error and provide me the same will see as whats the issue.
I restarted the BI System (all components), a few more times and now weblogic login is working again on all three apps (FMC, Console, OBIEE) on both environments. Still, I don't have a "warm fuzzy" about this setup.
In our setting we're trying to prove out, we are looking for an active-passive solution where Env1 is active and Env2 is passive (but it's a hot standby in the sense that the services are started, it's just no traffic is going to it) and the shared webCat is on it's own clustered file system. If Env1 fails, the load balancer starts pointing to Env2. My hope was I could point both environments to the same webcat to avoid having to do any syncing or redeployment of webcat when there is a failover, but now I'm having doubts that this is possible, and it seems unstable to have 2 different environments pointing to the same webcat.
to make it active-passive you need to setup the veritas cluster for the SAN mount so if env 1 goes down it switches over to secondary or else you need to setup cluster env using scale out option so even if primary goes down the secondary should be able to take all requests accordingly.