>> More generally, I'll start a discussion about the driver polling rate >> (2 sec) which was accurate with the dumb and old smart units, but >> which is really too high for recent / verbose units (that also support >> an interrupt / notification mechanism, allowing to lower the polling >> rate, while not missing events). > That is a no-brainer. We already support changing the polling rate through > the -i command line option for the drivers. Just advise people to set this > to 30 seconds or so for the drivers in question and we should be fine.
Or by adding 'pollinterval = 30' in 'ups.conf' for these drivers once it has been determined that this is the cause of the problems. Alternatively, we could remove the 'static' from static unsigned int poll_interval = 2; in 'drivers/main.c' and add static unsigned int poll_interval; to 'drivers/main.h' so that driver authors can set this value to a more sensible default for their driver if the UPS is known not to handle the two second polling interval that is default now. I wouldn't want to see the default changed to a higher value for backwards compatibility. Hammering a driver with too many polling commands will probably show up in the logs somewhere (or otherwise give unsatisfactory results) while silently increasing this to a higher value may mean that power events are missed on older systems/installations. The latter may go unnoticed until it causes a problem. Best regards, Arjen _______________________________________________ Nut-upsuser mailing list Nut-upsuser@lists.alioth.debian.org http://lists.alioth.debian.org/mailman/listinfo/nut-upsuser