So I had a bit of an experience with the PW microgrid restart Saturday morning, figured I'd post my findings.
I had my first 'real' outage Friday night (with issues, those documented
here), my only previous one was less than 30 seconds plus some simulated outages I did very early on (over a year ago). I didn't do a great job of managing my home power draw overnight, I left the portable AC running for 2 hours (until the outside temp was below 90F) which drew the PW down to 70%, ran the whole-house-fan as it continued to cool outside until around midnight, then realized that I was already below 40% and not in good shape to last until sun-up, so I powered off some electronics and went to bed. Turns out right around 7am the PW hit 10% (Gateway reported SoE) and turned off the microgrid (solar was probably about 15 minutes from coming online, and maybe about 30 minutes from being able to cover the house load). I woke up around 7:30, so that's when I started looking at things.
So of course with no line voltage, the inverters were stuck offline (blank displays). I don't know for certain that the GW wouldn't have eventually (when?) tried to bring things back online to see if the solar inverters would come up, but that certainly had not happened by 8am, so I opened up the GW & breaker panels and started poking at things. The Gateway was still reporting to Tesla's app, and showed the battery at 5% but everything black. I verified on the Gateway's 4-pin PW connector that it was still receiving 12V from the PW, and that the 'lower box' in the Gateway was still powered (the LEDs visible through the reset hole were still doing their thing). In addition, the TEG-xxx WiFi network was still present. I joined it and verified that I could read the GW REST API. It seems like the Neurio was probably not powered, as the GW reported a last communication time of 0001-01-01 from everything but the battery. Having confirmed that ground was common between the PW connector's ground pin and the JUMP connector's - pin, I tried shorting the JUMP connector's + pin to the 12V supply from the PW. I connected through a multimeter so that I could read the DC current. The Gateway drew about 1mA through this connection, but its behavior did not change at all (didn't bring up the microgrid, etc). Next, I figured I'd hit the GW's reset button (I still had the JUMP connection in place). After a minute or so the GW made it's startup chirps, and the microgrid started up again. So the one Solivia Tesla has on the PW side of my Gateway went into Synchronization, and after 5 minutes was online and the PW started charging. I was trying to play it safe on loads, so I had flipped all of my load breakers before I had reset the GW.
I flipped all of the breakers back on (house draw around 400W, solar was running around 500-700W), and within a few minutes the sun went behind some clouds and production dropped below consumption (<300W). The PW hit its 10% level again, and the system shut down again. Suspecting that the JUMP connection hadn't really done anything, I removed it, threw off all the house breakers again, then pushed the GW reset button again. Once again as soon as the Gateway chirped the microgrid came up and solar came back after 5 minutes as expected. This time I left off the breaker that has my office, where most of the 'always-on' power goes to reduce the load in case the solar production dropped again.
I had to pick up some 10/3 romex and another 30A breaker to properly move the second inverter to the PW side of the Gateway (I decided to do this Friday night since the west-facing array was on this inverter, so as the sun was setting I was basically getting nothing from the inverter that was left powering the PW, and didn't want to have to repeat that again, so I wanted the option to put the second inverter on either panel, but they use different breakers (my original panel is a D-Square, Tesla's added panel is Eaton).
Two things I was looking to avoid on Saturday was reaching full-charge on the PW (to avoid more microgrid frequency shifts, which didn't go well Friday night as I mentioned on the other post), and ending up with more solar output than my PW+house could absorb (with the portable AC on again the house was running around 2kW, and at solar noon the array should be generating around 7.7kW peak in this heat, which would still be above 5+2. So I left the second inverter switched-off until late afternoon once I knew I'd be past 7kW production, to help the PWs charge a bit more. They made it up to ~85% by the time the sun could no longer cover the house consumption of around 500W, since I'd switched the AC and some other loads back off again to manage my load for this evening and ensure things will still be working Sunday morning when the sun comes up.
1pm Sunday here and the grid is still down (it came up a few hours Saturday evening, but only a few hours). I had planned to leave both inverters on, to let the PW get fully-charged and start playing with the microgrid frequency, hoping it would hit full-charge before the inverters were in danger of exceeding the ~7kW, but it was a cloudy morning so charging was slower and I was seeing cloud-edge-style peaks which would likely go beyond 8kW, so I switched back to a single inverter, I'll still fill the battery today, it's at 88% with 2.4kW flowing into it right now. I also made some fixes to my RPi logging setup during the day Saturday, so hopefully I'll be a in a better place to keep logging today.
Circling back to the OP, now I will admit that I never actually threw the switch on the side of the PW, I would expect that probably cuts the 12V to the Gateway when it's off (maybe not), but hopefully that returns when you throw the switch on again. But my GW was without AC voltage on either the grid side or backup side (verified with my meter), and just pushing the GW's reset button (and waiting a minute or so for it to finish rebooting) without doing anything else on the GW was sufficient to get the microgrid running long enough for the solar to get going again. Obviously trying to reduce home loads during this period would be wise at least when the PW is already at its backup-reserve (10% GW SoE, 5% in the app) level, since you need at least 5 minutes of good power for the solar to start, so you don't want to risk drawing the battery down too much during that time, not sure if there's another threshold below which it won't even try to start the microgrid anymore, or couldn't sustain it for at least 5 minutes.
On the JUMP header, I suspect that's probably for when the PW is no longer supplying its 12V. Not sure if it can back-feed power to the PW or if it's just for the GW electronics. But it doesn't appear to be needed or change things when the PW is already powering the GW, which can be seen by the LEDs in the reset hole.