If they (the basics) are squared away then I always peek into any hot components (anything that intrinsicly warms up during operation) and all classic relays. (active -- also prone to gradual deterioration in the contacts)
High impedance DMMs etc have led me on a merry path more than once. They're passing very little current. This means that active relays that are in failure mode come off 'reading' okay.
One particularly maddening control circuit (spent half a day on this puppy) tested perfect time and time again. It only went crappy under load. Corrosion/ pitting had fouled up the contact surfaces of the relay enough to throttle the voltage -- but only under load. No readable resistance was shown when independently tested.
If you can do so, latch the relay into 'closed' then meg the resistance across the contact. At the low voltages used, even small bumps in resistance can scramble the logical brains of the controller. (Alternately, just swap out a fresh ice cube and retest.)
You'd hate to find out that you've bought a new $2000 beast -- and it still doesn't fix the problem!
As you might imagine, ice cubes are built with the same volume and attention to care that Edison lamps are! Of course they crap out all of the time. If that failure mode is in a semi-open semi-closed state -- they become absolute puzzlers.
For my trouble call all that was required was a file job on the contacts. (!) Suddenly everything was cured. ( It was a snakes nest of control cabling running with motor feeders. A terrible practice, IMHO.)
I strongly recommended that the client totally rewire his motor controllers ASAP -- if reliability means anything to him. (Industrial environment was attacking his gear with acid vapors. It was astonishing to see the damage done in less than three years. He needed NEMA 4X in the worst way.)
