Tor node was doing more traffic than its bandwidth is configured for
I updated our node to the RC version some days ago. Earlier today, it started to do a traffic amount that was higher than it had been configured to do in torrc. Torrc was configured for 35M use and 40M burst, but today it went to 50M and stayed there for an hour. Killing it HUP didn't help. I did a service restart on it to bring it back down.
On Sat, Sep 07, 2013 at 10:16:51PM -0400, tor@t-3.net wrote:
I updated our node to the RC version some days ago. Earlier today, it started to do a traffic amount that was higher than it had been configured to do in torrc. Torrc was configured for 35M use and 40M burst, but today it went to 50M and stayed there for an hour. Killing it HUP didn't help. I did a service restart on it to bring it back down.
Are you sure you didn't confuse bits and bytes? Tor counts in bytes. (The arm monitor, if that's what you're using, counts in bits by default.) --Roger
On 9/8/13, Roger Dingledine <arma@mit.edu> wrote:
Are you sure you didn't confuse bits and bytes? Tor counts in bytes.
(The arm monitor, if that's what you're using, counts in bits by default.)
As with real networks and operators, if this is so, then big thank you to arm people for correctly counting network bandwidth in bps. No thanks on webhosters and isp's who convert their upstrream bandwidth contracts into transfer bytes and pass that on to their customers for their apache logs, thereby spoofing hosted network *bandwidth* apps like Tor into feeling some silly need to count bytes. Do wish Tor would speak properly in bits by default.
participants (3)
-
grarpamp -
Roger Dingledine -
tor＠t-3.net