My experience with finding ground faults using the voltage method with
a dvm was 100% until this week.

I modified my usual debug pattern and found the ground fault. With the
pack completely assembled, measuring along until I found the fake
ground fault, I would take out one battery connector. Then the ground
fault would go away. Then I’d put back that first connector and take
out another one about 20 cells away. So only one connector was removed
at a time. Removing multiple connectors does not work as I discovered
a long time ago. After a while the real ground fault was at hand. But
only because there was a single break in the pack right before the
real ground fault.

The culprit may have been of my own doing. When I pulled that bad cell
out, there was a major puddle of KOH on the box floor. I had to take
out about 12 cells to clean it all up. And in doing so found an old
dried up koh path that needed attention just to be thorough. Something
like 40 cells were pulled today, wiped off with diluted vinegar and
had the tops brushed off.

Earlier in the week when I had to remove the 27 bluetops for what
appeared to be a ground fault, I must have over stressed a cell, as it
was cracked down the corner for about an inch and had been recently
toped off with water. So it drained alot of electrolyte into the
battery box. As it turns out I must have been careless as there were
two other cells with cracks on the top that I had to leave out. So the
pack is running with 3 less cells.