Hello,
I've recently established a full-time relay running on a fast Linux router. Working nice with version 0.2.4.17-rc.
Have a daily offsite backup that was butting heads with TOR traffic, so I wrote a 'tor_bwreduce' and 'tor_bwrestore' script like these:
nc 10.92.88.1 9151 <<EOF AUTHENTICATE "xxx" SETCONF RelayBandwidthRate=60000 SETCONF RelayBandwidthBurst=60000 QUIT EOF :
nc 10.92.88.1 9151 <<EOF AUTHENTICATE "xxx" SETCONF RelayBandwidthBurst=375000 SETCONF RelayBandwidthRate=250000 QUIT EOF :
and run them at the appropriate points in the offsite backup script. The backup is rate-limited as well so the link now avoids going into significant oversubscription or contention.
This works much better than I expected. The Tor network consensus bandwidth for the node was taking a bad hit due to congestion delays when the backup ran. Now it seems unaffected and the consensus for my node has shot up above the 'RelayBandwidthRate' value.
So that's all super.
The question is, should I put in an adjustment for 'MaxAdvertisedBandwidth' during the backup window or make any other change to advise remote relays to de-prioritize the node for the duration?
Ideally I'd like to put the node into temporary hibernation, but it seems the only way to do that is indirectly by fiddling with several Accounting* parameters.
Thanks!
P.S. BTW Vidalia could use some improvement in the area of dealing with a Tor relay living on a different box in the same network. Nothing major, but it should
1) not override the config and/or warn first 2) import the config when a virgin TBB install connects to a non-local address 3) not shut down the relay casually without an "are you sure?" pop-up