gpsd tcp NMEA feed

[Nick Taylor]

Hey Gary

You may recall in the dim and distant past we hassled you for a change 
to make the tcp NMEA feed option more reliable and resilient... It has 
taken us a long long time to get approval to test this on our customer 
installations but we are now able to see how the option performs in the 
real world.

For some reason the feed works fine for a while and then gpsd stops 
receiving feed - even though we can verify feed using telnet. I have 
managed to setup a fail situation which may or may not be similar to 
what's happening in the field, however I think this is still a valid 
case where it would be nice if gpsd could reconnect to the feed.

My test setup is as follows:

Machine A running std gpsd with feed from hardware device

Create tcp feed from machine A as per your docs:
socat EXEC:'gpspipe -r' TCP-LISTEN:2948,reuseaddr,fork

Machine B run gpsd in debug mode:
gpsd -D5 -n -N tcp://192.168.2.53:2948

All good so farnow interrupt the feed from machine A as follows:
iptables -I OUTPUT -d 192.168.2.52 -j DROP; sleep 1200; iptables -D 
OUTPUT -d 192.168.2.52 -j DROP

Machine A we see this:
2022/09/26 09:55:07 socat[16311] E write(7, 0x7f674db0, 8192): 
Connection timed out

gpsd log on machine B loops as follows:
gpsd:PROG: CORE: pselect: timeout
gpsd:PROG: checking client(0)
gpsd:CLIENT: <= client(0): ?WATCH={"enable":true,"json":true};

gpsd:PROG: awaken(0) fd 7, path tcp://192.168.2.53:2948
gpsd:PROG: device 0 (fd=7, path tcp://192.168.2.53:2948) already active.
gpsd:CLIENT: => client(0) len 278: 
{"class":"DEVICES","devices":[{"class":"DEVICE","path":"tcp://192.168.2.53:2948","driver":"NMEA0183","a
ctivated":"2022-09-26T09:37:29.345Z","flags":1}]}
{"class":"WATCH","enable":true,"json":true,"nmea":false,"raw":0,"scaled":false,"timing":false,"split24":false,"pps":false} 


gpsd:PROG: CORE: pselect: timeout
gpsd:PROG: gpsd_multipoll(7) DEVICE_UNCHANGED for 5
gpsd:PROG: checking client(0)
gpsd:CLIENT: <= client(0): ?WATCH={"enable":true,"json":true};

gpsd:PROG: awaken(0) fd 7, path tcp://192.168.2.53:2948
gpsd:PROG: device 0 (fd=7, path tcp://192.168.2.53:2948) already active.
gpsd:CLIENT: => client(0) len 278: 
{"class":"DEVICES","devices":[{"class":"DEVICE","path":"tcp://192.168.2.53:2948","driver":"NMEA0183","a
ctivated":"2022-09-26T09:37:29.345Z","flags":1}]}
{"class":"WATCH","enable":true,"json":true,"nmea":false,"raw":0,"scaled":false,"timing":false,"split24":false,"pps":false} 


gpsd:PROG: CORE: pselect: timeout
gpsd:PROG: checking client(0)
gpsd:CLIENT: <= client(0): ?WATCH={"enable":true,"json":true};

gpsd:PROG: awaken(0) fd 7, path tcp://192.168.2.53:2948
gpsd:PROG: device 0 (fd=7, path tcp://192.168.2.53:2948) already active.
gpsd:CLIENT: => client(0) len 278: 
{"class":"DEVICES","devices":[{"class":"DEVICE","path":"tcp://192.168.2.53:2948","driver":"NMEA0183","a
ctivated":"2022-09-26T09:37:29.345Z","flags":1}]}
{"class":"WATCH","enable":true,"json":true,"nmea":false,"raw":0,"scaled":false,"timing":false,"split24":false,"pps":false}

Feed can be seen on machine B using telnet, but it seems gpsd doesn't 
notice the failed connection and the only way I found to recover is to 
restart gpsd

Your input would be much appreciated

Thanks and regards

Nick