I had four audio inputs to mix. Three were synthesizer based: from the Fantom, the XV5080 and the Muse Receptor. I also took a DI out from the bass amp and fed that into the mixer as well. This way the bass player could use his amp as his own monitor, keeping it at a low volume, and the bass amp feed could be mixed appropriately with the rest of the instruments for a better balanced sound for the audience to hear. I use a Mackie 1202-VLZ3 as it is lightweight, has enough inputs and outputs, and has decent audio quality and durability.
Our theater runs its sound in mono. They have two Front-Of-House (FOH) speakers set up on speaker poles as well as two smaller speakers which are used as monitors for the actors to hear the music while they are on-stage. I sent the main output of the Mackie mixer to our soundboard. This signal is then routed at the soundboard to both FOH and monitors. The audio for the FOH is sent on the left channel and the monitor feed is sent on the right channel, then each channel strip on the soundboard can be panned for the appropriate mix of FOH to monitor signal. Generally the microphones for the actors are panned hard left so they only come through the FOH (too much sent through the monitors creates feedback). The music mix from the Mackie is sent to its own channel strip, and that generally runs on center pan, feeding the audience and the actors on stage equally.
I used the AUX outputs on the Mackie to send feeds to small amplifiers (15W) which our drummer and bass player had positioned on the floor near them. They also needed to be able to monitor the sound and have their levels set independently of the FOH mix.
For myself, I used a pair of AKG K240 Studio headphones. I generally kept the headphones on one ear (to be sure I was able to capture the tracks if the FOH level was lowered by the sound board operator) and kept my other ear open to monitor the ambient sound in the room (as well as the FOH output of all the actors' microphones). This proved sufficient in our environment, but an alternative would have been to send a submix of the actors' microphones back to my Mackie and use that feed only in my monitoring mix, not in the main mix, of course (which would have created a feedback loop).