Hi, I'll fix/mitigate the race using the needed Wants/After against udev-settle.service.
At the same time I don't think it's really coherent that: When the driver is initially started and no device is present, the driver immediately exits. But when the driver is running and that the device is unplugged, the driver is still running, waiting for the device to return. Cheers Laurent Bigonville Le Sat, 11 Aug 2012 20:31:10 +0200, Arnaud Quette <[email protected]> a écrit : > 2012/8/11 Laurent Bigonville <[email protected]> > > > retitle 677143 nut-server: driver not always started when using > > systemd thanks > > > > Hello, > > > > Actually it seems that it's the driver which is not started > > properly: > > > > Aug 11 17:37:39 fornost upsd[1967]: Can't connect to UPS [ellipse] > > (usbhid-ups-ellipse): No such file or directory > > > > Is it possible that /dev/bus/usb/ is not fully populated. This seems > > random on my system, so I guess there is still a race somewhere. > > > > a classic issue with USB devs, and a thing I've never had time to > check toroughly, is our udev calls in postinst WRT the device being > already plugged in. > in that case, I'm not sure the perms are refreshed, without > unplugging / plugging back the UPS USB cord... > and as you told, a race in systemd is still possible. > > cheers, > Arnaud -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
