However does anybody have an idea why if I connect one servo to my arduino and use the arduino power to supply it its working.
When I use my multimeter to meassure the current the problems are starting again?
I use relative good multimeter so the resistance should be really small and shouldn't change anything on the circuit
If I'm reading you correctly, the servo is still connected to the Arduino's power when you are trying to measure it and the problem starts again. It could be the single servo is getting just barely enough current from the Arduino's power supply, and even the slight additional load of the multimeter is enough to change that. I admit it's a really unlikely scenario, but the only other thing I can think of, the extra length of the multimeter's cables adding more noise, seems even less probable.
I don't know how many +5 VDC
outputs your ATX PSU has, but could you instead test individual servos on an otherwise unused +5 VDC