Hey all,
Just got off a call this morning and it looks like there's been some changes with the upgrade to HCO that aren't quite common knowledge, even among the support techs. Figured I'd share what we had to do to work around it and get it (partially) working.
We had to do the following:
- Download the ODAC package (19c x64 was what we used)--Pay attention to the client version the .exe is going to install, you're going to use the corresponding .zip later
- Unpack it to an arbitrary folder (Let's just say F:\OracleClient)
- Run the installer (I pointed it to install in the above created folder, retaining the structure)
- Unzip the .zip archive that corresponds to the ODAC version on the .exe (remember that client version we noted above?)
- Here's the fun part: We had to go into the Microsoft .NET root in C:\Windows and search for our DLL to copy into the SolarWinds directory
- Search the subfolders for Oracle.ManagedDataAccess.dll
- Copy that into your SolarWinds application root folder (Ours is on a different drive than system, so example: F:\SolarWinds\ )
- And now, in an elevated command prompt, navigate back to the folder you unzipped, and then get to where OraProvCfg.exe lives
- In our case, F:\OracleClient\V1032762-01\odp.net4\odp.net\bin\4\
- Then run the following command (substitute your own directory, ours is probably different)
- OraProvCfg.exe /action:gac /providerpath:"F:\SolarWinds\Oracle.ManagedDataAccess.dll"
- Expected output is "INFO: F:\solarwinds\oracle.manageddataaccess.dll is registered successfully in GAC."
Hoping it goes *that* simply for you, not sure if default C: drive install has you putting the DLL in \SolarWinds or \SolarWinds\Orion.
* * * And now, the kicker: HCO can't run Oracle queries from an APE. * * *
If you're tuning in early to mid February 2023, there's a bug in HCO, which I'm hoping gets patched with a HotFix quickly. They say they have a workaround developed for it, to be released in a later version update, but this bug breaks Oracle UX monitoring functionality in a big way and doesn't need to wait around for a new release:
As of right now, you can only run an Oracle UX monitor from the primary poller--Not an Alternate Polling Engine. This is very *very* bad if you have a distributed environment, automated load balancing scripts, a lot of Oracle queries, like to keep your core server clear, or just don't like single points of failure.
Hoping we can get some quick turnaround on that fix they've already developed and they can push it as a HotFix instead of just part of the new release roadmap. If you know the right people to tag and make that happen, please tag them.
