You have 2 different CPIDs. In general this can be caused by three things. The first is joining a new project, the second is wrong emails, and the third is no path from client to server to client to server... that updates all of the CPIDs to the same.
In this case it appears that the path from host to project server is broken since Predictor has been down for a month. This should fix itself within a week or so after Predictor is back up and running. Until that time there is no way to fix it.
You can check are that the email accounts are the same on all of the projects including case (foo@bar.com is different from Foo@bar.com). You should also make certain that there is some path to get the CPID to Predictor when it comes back. For example if Machine A runs Predictor only, and Machine B runs S@H and CPDN only and there are no machines that run both Predictor and one of S@H or CPDN, there is no path. If you have one machine that runs all three, there is a path. If you have one machine that runs Predictor and CPDN, and another that runs CPDN and S@H, there is a path.