[Nut-upsdev] Autodetecting running drivers by the 'upsd' server

[Arnaud Quette]

2007/6/6, Arjen de Korte <nut+devel at de-korte.org>:
> >> 1) Drivers must have created their socket before the server is
> >> started. This is not different from what we have now, but the server
> >> wouldn't be able to detect that a driver which should be running
> >> (according to 'ups.conf') is not available. Personally, I don't think
> >> this is not much of a problem as long as 'upsdrvctl' complains about
> >> this (which it usually does).
> > The server could periodically look for new sockets (and/or on SIGHUP).
>
> I'm not in favor of periodically looking for new sockets. If you don't
> start additional drivers, it is pointless to check for new sockets
> regularly. A better option would be for the drivers to send a SIGHUP to
> the server on startup (if a 'upsd.pid' file is found), which we already
> honor. The prerequisite that drivers and server are running under the same
> UID is satisfied most of the time (if not always) anyway, since they will
> have to communicate to one another. The bonus here is, that we might send
> a SIGHUP to the server on intentional termination of the driver too, so
> that it wouldn't be checking for stale sockets forever. This opens the way
> for true hot(un)plugging of drivers.
>
> >> 2) The STATEPATH should be dedicated for the driver, since the server
> >> can't know in advance if a socket is for a NUT driver or some other
> >> application that happens to be using the same directory for its
> >> sockets. I don't think this is a big problem though.
> > The nut directory in /var should be for the exclusive use of NUT anyway,
> > for security (that way you can chown the directory to the nut user)
>
> Most likely, yes.

fine by me.

Arnaud
-- 
Linux / Unix Expert - MGE UPS SYSTEMS - R&D Dpt
Network UPS Tools (NUT) Project Leader - http://www.networkupstools.org/
Debian Developer - http://people.debian.org/~aquette/
OpenSource Developer - http://arnaud.quette.free.fr/