Using OATS and OVR - The TLA edition

OATs is very good for load testing JD Edwards.  I think it's handy to use at all phases of the software lifecycle, but especially for testing a tools release, an upgrade or platform migration.

I've recently been involved in getting OATs working with OVR, this was a bit more of a challenge, as the standard correlation library that comes for JD Edwards does not do any of the OAT's session correlations - therefore, it'll never work "out of the box".

There are a bunch of internal header values used for communicating to OATS that you need to manually keep track of and substitute in your scripts, namely:

• tid
• xtoken
• xdm
• xdo
• resourceCacheID

Many of these can be set up in a new correlation library, but some cannot.  It's a little more difficult when the substitutions need to be done on _xdo and also xdo when using correlation libraries.

I'd generally show you a pile of code snippets and screen shots, but I'm struggling with blogging from windows 10 at the moment to blogger, so these are not going to happen.

The process is to search the incoming HTML for something like xdo, and create a variable out of that.  then replace all other instances of xdo in the script, like I said in a perfect world this would work by adding this rule to the correlation library - but the world is not perfect.  The other thing to remember is that this substitution must be done in RAW.

Ensure that your variable rule is something like "&_xdo=(.+?)&_xt=", it needs to be nice and generic to ensure that you are getting the xdo variable.  Note also that if you call a subsequent OV report in the same JDE login session you'll need to refresh this and the xtoken (yes, painful).

Make sure you then prescribe the variable substitution with the following:

xdo=((.+?))(&|$)

Resource cache is done with

cache:oracle\.xdo\.common\.io\.Cache(.+?)\.tmp#

I then say replace in all locations.