That was maybe a year or two ago. Recently, it developed a new quirk, where it would leak water all over the counter. It wouldn't do this every time, and some of the water ended up in the pitcher where it belongs, but it was enough of a leak to make a thorough mess of things. When I had enough time to watch it closely in operation (the cycle time is a few hours, and it didn't always leak), I was able to tell it wasn't any of the external seals, it seemed to be leaking from within the mechanism. So I finally took it apart and carefully studied the steam/water path. I figured it was sealed all the way down and a seal had failed. Nope, it's not sealed. There's a riser tube, an L-shaped plastic connector tube, the spiral heat exchanger, which drains into a plastic assembly integrated with the pitcher-detector plunger/switch. At first, I suspected the connector tube, which felt as if it may have cracked along a seam. I snipped off the tie-wraps and had a close look, and realized two things. One was that it wasn't cracked. The other was that there's a neat hole drilled in the top of the heat exchanger. It isn't pressurized, and is designed not to be. A good idea – see the Mythbusters "exploding water heater" episode for a graphic depiction of why (the high speed footage of the water heater pretty much destroying a whole house really drives the point home). I opened up the plastic assembly and saw that the lower end of the heat exchanger is simply positioned above the top of the output assembly; it's open to the air and drains into it by gravity.
Hmm, not obvious what's going wrong. The design isn't what I expected, but everything seems to be in place. So reassembled it, except for the top cover, plugged it in, and watched it carefully while it was running. Eventually, I could see steam in the L-tube, and then water coming out of the heat exchanger, but there was a longish delay before it actually came out into the pitcher. Curious. But no obvious leaks. I carefully jiggled and poked things (mindful of both the hot steam and exposed mains voltage), to no avail. So I kept watching, and sure enough, I saw some water start to dribble out of the ring vents around the output assembly. That isn't supposed to happen. So I opened it back up (which deactivated the pitcher-detect switch), hot-wired the switch, and waited for it to heat back up.
The water started to dribble into the output cup, as expected, but started to pool up. Odd. Why would the water pool up? It should just go to the lowest point and run out.
Then I finally realized what was actually going on. There's a little gasket at the bottom of the output cup that the pitcher detect plunger engages to shut off the water if you remove the pitcher while the unit is running. What I had missed was that the gasket was supposed to stay with the plunger, not sit on the output opening! In that position, it blocked the output, and the pip on the plunger (that's supposed to hold the gasket) merely plugged the hole in the gasket (the plunger doesn't rise very far). To put it back to the original configuration, I'd attach the gasket back to the plunger. But it's made of one of those "low surface energy" plastics that's difficult to adhere to, which likely led to the problem in the first place. What if I just removed the gasket? The water would have a clear path to drain, the pitcher detection switch would still work, it would just leak water if I removed the pitcher while it was running. No problem, neither Andrea nor I do that anyway. So I left the gasket out, reassembled the thing, and fired it back up.
It shut down after maybe a third of a gallon. NOW what was wrong? Oh right, we had shut it down partway through because it was leaking, and emptied the pitcher. The boiler only had part of a gallon to start with. So I refilled it, set it running, and observed it until it started dispensing water into the pitcher like it's supposed to. Seems to be working again, yay!