[atlas] Proble rebooting several times a day.
Robert Kisteleki robert at ripe.net
Wed Feb 22 15:04:03 CET 2012
On 2012.02.22. 14:47, Lukasz Trabinski wrote: > 2012/2/22 David Precious <davidp at preshweb.co.uk>: >> On Wed, 22 Feb 2012 14:08:31 +0100 >> Lukasz Trabinski <lukasz at wsisiz.edu.pl> wrote: > >>> Why my probe (id 2598, Firmware Version: 4280) rebooting several >>> times a day? My routers and uplinks connections works correctly and >>> stable. >> >> Only a random stab in the dark, but what USB port is it powered by? >> Could the device supplying the power be going into a power-saving mode >> and shutting down power to that port? > > It's connected to USB port in PC router (Linux, FC16, with Quagga). > I know that probe send and recive packets all the time. (sniffing > packets by tcpdump) > In my opinion usb port shouldn't go into sleep or power-saving mode. I > dont's see any > action related with that usb port in my logs files but on switch where > probe is connected I > observe flap port (down/up). The reference to USB power source issues is an interesting one -- we haven't seen that as a "structural" reason for disconnection before. Having said that, the fact that the probe is up for hours at a time suggests that the local network and power conditions are fine. Please keep in mind that our definition of "up" is when the probe is connected to us, ie. a permanent TCP connection to one of our servers is maintained. This connection may be broken for a number of reasons, including local (transient) network problems at either end. Also, the probe continues its measurements even if it's disconnected, and submits all the data once it reconnects. In short: it's not a big deal if a probe disconnects, as long as it can reconnect in a reasonable time. Regards, Robert