Hi, I've recently added a UPS (CyberPower CP1300EPFCLCD) to one of our servers, and I'm managing it with NUT and the usbhid-ups driver. I've configured it to do what we want, and it all works fine. (upsc displays information from upsd correctly, upsmon correctly shuts down both the server and the UPS when the battery is low, messages are logged when the mains supply to the UPS is cut/restored, and so forth.)
However, in the week since installing NUT the server has suffered four kernel crashes (on stock 64-bit Ubuntu kernels, currently 3.2.0-33-generic) and has been unable to stay up for longer than 12-36 hours without crashing hard and requiring manual power cycling; this system has never previously crashed. Googling around the issue brings up various things, including http://thr3ads.net/nut-upsdev/2011/01/1475908-a-nasty-kernel-oops, which make me think this could be an existing issue with the kernel/library/USB stack that's being triggered by NUT. I've tried a couple of workarounds (including running the driver with -D), but none have prevented these lock-ups. I'd like to use NUT with usbhid-ups to manage the UPS, but also (obviously) need to keep the server stable. If there's a way to do this, I'd like to; if not, I've got some crash logs that someone might perhaps find useful. I'm happy to provide more details, but first I'd like to check a few basics since I'm new to the list: 1. Is this the right mailing list to be asking about this? I can imagine the devs would be interested in this, but I'm primarily here as a user. 2. Is the Linux USB subsystem already known to be prone to stability issues like this, or am I just having some bad luck on this machine? 3. What's the polite way to supply the list with large attachments, such as log extracts (~10kB) and photos of kernel crash screens (~900kB)? Many thanks in advance, Paul. _______________________________________________ Nut-upsuser mailing list Nut-upsuser@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsuser