[Nut-upsdev] Some thoughts on loss of communication
clepple at gmail.com
Sat Nov 21 23:54:22 UTC 2015
On Nov 21, 2015, at 2:06 PM, john hart <jsamcr at yahoo.com> wrote:
> 1. When communications are lost, instead of shutting down immediately why not set a flag and then stop the driver and then restart the driver. If communication is not re-established, and the flag was set, then shut down the system. Perhaps this is already being done.
This is not likely to be implemented any time soon. It is too dependent on distribution-related things, like how the driver is launched.
Our preference is to fix the driver so that it reconnects itself, but the maintainer of the blazer_* drivers has not worked on NUT for several years (hence the move to nutdrv_qx).
> 2. In my case, when the shutdown occurs due to loss of communication, the NUT software tries to turn off the UPS. It can't do that because there is no communication. This cause the shutdown process to hang up for some period of time. In my opinion this is not good because the system may crash during that period when it is hung. The software should skip trying to shut off the UPS and just execute a normal shutdown.
NUT cannot foresee every kind of communication failure, and most of the drivers already have timeouts built in. As a backup, the distribution startup/shutdown scripts should probably have timeouts as well - this used to be implemented as a delay between sending SIGTERM and SIGKILL.
clepple at gmail
More information about the Nut-upsdev