Our enterprise geodatabase has two feature classes relating to a single table (1:M, simple, None messages, GlobalID to GUID). We published a single service from both feature classes and the related table. For some reason, when we collect related records on feature class A, the globalID from feature class B is getting populated in the related table. It seems to be 'grabbing' the related records from surrounding features from feature A. This is causing many issues. In some cases, as many as 40 records relate back to a single point in feature class B which is not correct. We noticed it only happens when running ArcCollector on Andriod OS.
See the attached image. I know this is a confusing question, but we need to get this squared away as soon as possible.
The images shows 24 table records relating to 1 point feature in feature class B, and 8 in feature class A. Our relationships are not set up to where a GUID would be populated for BOTH FC A guid and FC B guid. It should be either one or the other.