PMB disconnect on mesh change
I have two ESP8266 running the sample code from https://randomnerdtutorials.com/esp-mesh-esp32-esp8266-painlessmesh/ (but I set one of them to say node1 and the other to say node2, in their messages), with painlessMesh 1.4.7 . I've cloned painlessMeshBoost (master; tag v0.9.8) to my Ubuntu laptop, and compiled it. I plug in the two ESP8266, and connect the computer to the resulting wifi. I figure out the gateway IP from ifconfig. I start PMB with e.g.
./painlessMeshBoost --client 10.104.114.1
and for a while, everything works fine. However, if I turn off or reset one of the ESP8266 (it doesn't seem to matter which one), PMB often disconnects - like, >50% of the time. It stops printing messages for a few seconds, then prints a disconnect event. Here's the tail end of a log:
{"event":"receive","nodeTime":717452417,"time":"2021-10-08T22:49:30","nodeId":2888001650,"msg":"Hi from node2 2888001650"}
{"event":"receive","nodeTime":719245068,"time":"2021-10-08T22:49:32","nodeId":3194154387,"msg":"Hi from node1 3194154387"}
{"event":"receive","nodeTime":719337295,"time":"2021-10-08T22:49:32","nodeId":2888001650,"msg":"Hi from node2 2888001650"}
{"event":"receive","nodeTime":721301792,"time":"2021-10-08T22:49:34","nodeId":2888001650,"msg":"Hi from node2 2888001650"}
{"event":"receive","nodeTime":726029728,"time":"2021-10-08T22:49:39","nodeId":2888001650,"msg":"Hi from node2 2888001650"}
{"event":"offset","nodeTime":726552106,"time":"2021-10-08T22:49:39","offset":-3677}
{"event":"offset","nodeTime":726559352,"time":"2021-10-08T22:49:40","offset":-479}
{"event":"receive","nodeTime":729986355,"time":"2021-10-08T22:49:43","nodeId":2888001650,"msg":"Hi from node2 2888001650"}
{"event":"change","nodeTime":741313077,"time":"2021-10-08T22:49:54", "layout":{"nodeId":429934944}}
{"event":"disconnect","nodeTime":741313172,"time":"2021-10-08T22:49:54","nodeId":2888001650, "layout":{"nodeId":429934944}}
I'm pretty sure I reset the one not directly connected to the computer - I connected PMB while only one node was on, then turned on the other one, and that was the one I later reset.
Is this expected behavior? The ESPs themselves seem to be robust, so far; it's just the laptop that's having trouble staying connected. I'd have expected it to retain the connection - preferably even reconnecting after resets of the node the computer is connected to, but at least it should be immune to resets of unrelated nodes.