The only time you really can get the peaks and the rms value to be the same would be a DC (0 Hz) signal. Any other signal will have peaks that are higher than the RMS value. A pure sine wave should have an RMS value that is about .7 times lower than the peak value (-3dB). So I would expect that any signal that contains more than one frequency would have even a higher difference between RMS and peak. Even if you limit or compress the signal the RMS and Peak values will always show some difference.
In addition unless you use a fast brickwall type limiter some signal can get through at the beginning of a transient. In your picture you have a compressor with 3ms attack time. This attack time allows some of the initial sound to get through the compressor and can actually make the peak to RMS difference larger because part of the peak gets through but then the rest of the sound is compressed.
There is nothing wrong with having peaks, that is part of what gives sound its timbre and dynamics. You just want the peaks not to clip when leaving your master buss.