this is just a horizontal timing issue. (monitor is turned 90 degrees, so an up and down direction is actually left and right as the monitor scans the lines) maybe a cap, maybe a dirty adjustment pot. noise in the power supply. hard to really say without knowing exactly which monitor is in that cabaret.
timing is basically waiting for a capacitor to charge from a power supply... whose level and durations is adjusted by a potentiometer. a capacitor will charge at a set rate... 0 being the start and charging up to whatever, say 20 volts) would be the end of the scan line. as a result, allowing more voltage would charge the capacitor faster to that 20 volts before being reset. the scanline gets longer and your picture width reduces.
this is of course grossly simplified. you could tell where the issue is by looking at like 10 different oscilloscope traces the makers of the monitor... from all over the board... and every one of them influences the final power supply feed rate for that timing thing I mentioned above.
but easy things to check is how old the caps are and if they might have actually been changed previously...they may have skipped a few ( if it WAS done....or whatever.)
and just cycling or cleaning the pot controls responsible for the horizontal timing size/hold/etc