Two melting peaks usually means that you have two PCR products. Have you run these specific samples, that gave you 2 melt peaks, on an agarose gel? This should be your first troubleshooting step. If you have 2 bands, then you could try Sanger sequencing them to find out what they are. There are lots of possibilities (a nonspecific product, an unidentified isoform, an indel, a contaminant...)

If you are getting two melting peaks but only one band on a high resolution agarose gel, I would still suggest sequencing the band, because it's possible (although very rare) that you have a big substitution, resulting in two PCR products similar enough in size to run as one band on a gel, but that melt at different temperatures. I don't think that a point mutation or other small polymorphism would explain a 7 degree difference in melt temp. It's also possible that this is just an idiosyncracy of your primers. I have 2 primer pairs that produce weird melt results (double peaks, or big shoulders on peaks) randomly at least a quarter of the time, and I have never been able to work out why. These primer sets each produce a single product which sequences clean, tight cT values between technical replicates (including single peak vs double peak replicates) and repeatable results between runs. My best guess is that there are multiple possible secondary structures for the PCR product which melt at different temperatures.