I did a search on this, and surprisingly it hasnt come up yet. Basicallly, what seems to be happening is that we have an initial development server of BOEXI R2 SP1 installed with the CMS hosted on a Solaris 32-bit oracle 10g server. I have been advised that we are required to move to another server, which is a 64-bit Redhat Oracle 10G server.
The problem arises when I try and change the datasource. If I simply point it to the new server's TNSnames entry (which I can successfully tnsping, plus connect through SQL plus.) The account also has all necessary rights- verified through creating/dropping tables, procedures, and views.
When the server is stopped, I point it to the new TNSnames entry and it fails to resolve the cluster name of the CMS, but it takes the entry. I then try to "recreate the current datasource" to build a blank environment in the new database. This hangs and the whole CCM crashes. Restart, and the CMS service is still stopped. Event viewer indicates "Database failure. Reason ." (real helpful, huh?) The CMS log indicates an OCI call error. That tells me something is screwy with the scripting.
I also tried migrating from oracle server to oracle server to no avail- just hangs. However, migrating from the old (32-bit) to a local (BOEXI server installed) MySQL database, it works w/o a hitch.
After going over this every way I can possibly think of, I thought of adding the "-trace" switch on the cms command line. I then started with the old CMS, shut it down, switched datasources, and attempted to recreate the new CMS from scratch. I let it hang, then went to the CMS log, which showed several lines of SQL being executed with no problems, until it gets to a certain line and returns a datatype mismatch error- got BLOB info for a numeric, or vise versa.
With all that said, BO support is saying that they dont think 64 bit is tested or supported- even though the platforms.txt file says that 10G is supported, without regard to 32 or 64-bit.
So now, I ask- has anyone run into this before, or is anyone else running a 64-bit datasource for their CMS?
These aren't the droids you're looking for.
He can go about his business.
Move along.
The problem arises when I try and change the datasource. If I simply point it to the new server's TNSnames entry (which I can successfully tnsping, plus connect through SQL plus.) The account also has all necessary rights- verified through creating/dropping tables, procedures, and views.
When the server is stopped, I point it to the new TNSnames entry and it fails to resolve the cluster name of the CMS, but it takes the entry. I then try to "recreate the current datasource" to build a blank environment in the new database. This hangs and the whole CCM crashes. Restart, and the CMS service is still stopped. Event viewer indicates "Database failure. Reason ." (real helpful, huh?) The CMS log indicates an OCI call error. That tells me something is screwy with the scripting.
I also tried migrating from oracle server to oracle server to no avail- just hangs. However, migrating from the old (32-bit) to a local (BOEXI server installed) MySQL database, it works w/o a hitch.
After going over this every way I can possibly think of, I thought of adding the "-trace" switch on the cms command line. I then started with the old CMS, shut it down, switched datasources, and attempted to recreate the new CMS from scratch. I let it hang, then went to the CMS log, which showed several lines of SQL being executed with no problems, until it gets to a certain line and returns a datatype mismatch error- got BLOB info for a numeric, or vise versa.
With all that said, BO support is saying that they dont think 64 bit is tested or supported- even though the platforms.txt file says that 10G is supported, without regard to 32 or 64-bit.
So now, I ask- has anyone run into this before, or is anyone else running a 64-bit datasource for their CMS?
These aren't the droids you're looking for.
He can go about his business.
Move along.