The Pro Mini is cute. I could probably use a bare chip almost as easily but these are a bit neater. At $18au the price is right.
I have standardised on using 8 pin DIN connectors for my TWI devices. I use a 12V power feed, ground and two wires for TWI. I was careful to avoid using connections on the connector which has the "in circuit programming" signals.
I don't use the arduino bootstrap or development environment so I need program in my bootstrap program. I made an adaptor the convert the 10 pin kanda connector to the pro mini. I found it is quite reliable without need to solder pins or socket to the PCB. Ideally it only has to be programmed once and it only takes a minute or two.
As usual I protected the wiring in polymorph after taking the photo.
I also polymorped the DIN connector wiring. I flatted out a couple of wings which I can potentially use for mounting screws. The DIN connector are tight enough that they can easily support the weight of the pro mini without extra mounting.
My bootstrap allows me to upload application code via the TWI. With the addition of a solid state relay I'm ready to use the micro to do some real control work.
It was then used to make a pressure controller.
The symptoms for both mini were the same - more and more unreliable over time then failing to boot. I was able to tickle them to life sometime by touching the xtal pins with a probe lead or similar - cooling it also helped. It seemed like a crystal/resonator problem and I unsoldered the 16 meg surface mounted resonator on one and replaced it with a 16 meg through hole version I had in stock. The board then worked again and seemed reliable.
I suspect a bad batch on resonators was used build "pro-minis" and probably "pros" as well. The USB-arduino has been trouble free and I think they crystals not resonators.
I haven't heard anyone else report this and hopefully the newer boards are OK.
The mini-pro I have has fuses set for "low power" oscillator - this might be part of the problem.