Re: [gpsd-dev] ntpd/gpsd tangle

[Hal Murray]

address@hidden said:
> OK, I want to be sure I know what's going on here.  You were restarting ntpd
>  *while gpsd was running*?

Sure.  I've been doing that for years.  It works with SHM.

I assume I can start/stop/restart them in any order.


> It would not at all surprise me if this fails; in fact, if you're using the
> new JSON driver I would expect it to fail.  If that's what's going on I need
> to add some documentation that you can't do that, and  explain why.

Yes, I'm using the JSON driver.  (2 instances)

Restarting ntpd is the same as control-Cing a random program JSONing to gpsd. 
 (or having a program crash)  I expect gpsd to recover from that.

It may be one step more complicated than that.  Maybe it has to sleep then 
crash so the socket is in a non-empty state on one end or the other.


> There is still a possible gpsd bug here.  When a client disappears without
> closing its socket, gpsd is supposed to close down its end of the connection
>  gracefully after a timeout interval.  If that is failing to happen we need
> to understand why. 

I can make it happen reasonably easily.  (now that I know the recipe)

If that stack trace wasn't enough, tell me what to type and/or I'll try to 
work up a simple recipe you can run.


-- 
These are my opinions.  I hate spam.