- tor-relays - lists.torproject.org

optimize performance of a relay running on a VM
by s7r 02 Jul '14

02 Jul '14

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I have multiple relays running on the following systems: - - vmware vsphere virtualization technology - - 100 mbps port - - 1GB dedicated RAM - - 2.6 Ghz 1 core CPU dedicated - - OS: FreeBSD 10.0 Release amd64 or Debian Stable - - DO NOT KNOW IF I HAVE AES-NI SUPPORT OR HOW TO ACTIVATE IT (?) Currently atlas shows 3-4-5 MB/s advertised bandwidth for these relays. Arm shows between 600 and 1200 concurrent circuits (total of inbound, outbound and exit) and average traffic consumption is 5-6 TB per month (total both download and upload). I think this can be improved, but how? The servers have only purpose Tor, there is nothing else running on them and there is no bandwidth cap or throttling. Suggestions? Thanks in advance. - -- s7r PGP Fingerprint: 7C36 9232 5ABD FB0B 3021 03F1 837F A52C 8126 5B11 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (MingW32) iQEcBAEBAgAGBQJTswH9AAoJEIN/pSyBJlsRtTMH/jFV2FdfSksYNalSn9fmHcRk sKoWfNuRL/FQAQ1R0faN0F3lemOfDq80vosmd95h5EIZG7UmmN/xs9s8K9PQAzxf j3XHwCtYVkKkEA1djdAbFUB5s3WN6ieNm0qnnqZeBk16A9vJeRI2wkYFPRaDn6c5 n9u+p5QUjRmBKMkY/TOZ12TWSdeD683SoGdx4O5FeP90B+Q61SFgDtkvX6wtWKZr flWM/Bf0gpQL6Qowl9E49J5dJj+RgSYqPZ943x37nutBYzGmn0pbqXU1ulg0WFuz 3tu4bjwLITg1BB64jnF9hwn91pnd8ECay8OMqnZxbI1QycvPTPcNkwkasqARfTs= =MJBs -----END PGP SIGNATURE-----

3 4

Re: [tor-relays] One IPv4 address, 1Gbit connection
by Tim 01 Jul '14

01 Jul '14

Tom, Why not run multiple tor relays on different ports on the same IPv4 address? For example, you could run 6 relays on 6 different ports on your IPv4 address (6 x 180 Mpbs > 1 Gbps). This would also utilise your 4 cores much more efficiently than running 2 relays (each relay will only ever use 1 core for most operations). Tim > Mon, 30 Jun 2014 14:48:59 +0200 Christian Dietrich <christian.d.dietrich(a)gmail.com>: > > Hi there, > > You will never be able to utilize the whole gigabit connection, at least > with the current tor version. > I'm also running 2 tor nodes (00000000000myTOR) on a single machine (3rd > gen Core i, Quad Core /w AES-NI) on an single IP, > reaching ~1.3 Gbit/s on an 1000 Base-T FD connection. Since you do not use > hardware accelerated crypto, > your transfer speed should be much lower. Anway my relays are also cpu > limited due to the fact that tor isn't > really utilizing much more than one cpu core. > > - Christian > > > 2014-06-30 13:05 GMT+02:00 Tom van der Woerdt <info(a)tvdw.eu>: > >> Hi, >> >> I'm running a Tor exit node on a 1gbit connection. Currently it's maxing >> at about 180Mbit/s (both ways, so 360Mbit/s) per instance, and I'm running >> two instances. >> >> That's not really using the connection well. The box has 4 cores (no >> AES-NI) and I'm looking for ways to utilize the other 640Mbit/s. >> >> Sadly it's not possible to get more IPv4 addresses on this box. I do >> however have access to a big range of IPv6 addresses. >> >> Can I somehow run more Tor instances on this box by utilizing those IPv6 >> addresses? Or are there other ways to optimize the throughput and get >> closer to that 1Gbit? >> >> Tom

3 3

New Tor node operator hibernation question
by pandemicrelay＠hushmail.com 01 Jul '14

01 Jul '14

Hello, I am a new Tor relay operator, and have setup my first relay. It's been running as a Tor middle relay for about 10 days now, and it's been running solid. It has the stable, guard, fast, running, and valid flags. I have my AccountingMax quota set to start at the first of each month at midnight. However, just past midnight on 07-01 Tor said it was going to hibernate until 07-05. This had me confused as I was only about half way through my monthly quota setup in AccountingMax for June, and my hosting provider reset the data on the first at midnight. I read the manual page about AccountingMax again, and it appears this may have been normal behavior. However before I read the manual page, I forced my relay to wake up out of that hibernation. It appears to be operating normally right now. However, I am wondering if this will cause any issues this month, or in future months? Also, can someone explain if the hibernation of 5 days was normal behavior to begin with. Thank you

1 0

Hits against GFC hex-string fingerprint in IPTables yesterday
by tor＠t-3.net 30 Jun '14

30 Jun '14

Wanted to point out to the list that we got 3 hits to our "Libero2" relay on port 9001 from chinese IP address 210.72.9.44 to a "GFC" reject line in the relay's iptables. I think this is the first time we've seen hits to this type of rule. The packets were dropped via this iptables detection: -A INPUT -p tcp -m string --hex-string "|00002800390038008800870035008400160013000a00330032009a009900450044002f00960041000500ff020100000400230000|" --algo kmp The packets came through at approximately these times EST: Jun 29 15:36:10 Jun 29 15:36:38 Jun 29 15:37:15

1 0

One IPv4 address, 1Gbit connection
by Tom van der Woerdt 30 Jun '14

30 Jun '14

Hi, I'm running a Tor exit node on a 1gbit connection. Currently it's maxing at about 180Mbit/s (both ways, so 360Mbit/s) per instance, and I'm running two instances. That's not really using the connection well. The box has 4 cores (no AES-NI) and I'm looking for ways to utilize the other 640Mbit/s. Sadly it's not possible to get more IPv4 addresses on this box. I do however have access to a big range of IPv6 addresses. Can I somehow run more Tor instances on this box by utilizing those IPv6 addresses? Or are there other ways to optimize the throughput and get closer to that 1Gbit? Tom

2 1

Spam
by kingqueen 29 Jun '14

29 Jun '14

hi, well having run a relay for just under 2 weeks, I've got my first spam on this email address. As you will no doubt have guessed, this is not my usual email address, it is only used for the Tor relay contact details, this email list and a very few other projects that will not have published my email address anywhere. So I guess that this spam is as a result of my email address appearing on Atlas or Globe or similar as a result of it being the contact address for my Tor relay. I know this possibility is mentioned in the Tor documentation; I read it somewhere. I guess there's not much to be done about it; I wouldn't want to remove contact details - the address is already in the public domain anyway, and I want to be contactable with any problems. I guess I just have to use the standard spam filtering and combating measures. Could those with experience of running a relay for longer, please advise if this will become a major problem? Will spam increase substantially? Also, any suggestions very gratefully received. Thank you! KQ --- This email is free from viruses and malware because avast! Antivirus protection is active. http://www.avast.com

8 13

Circuit purposes
by Joel Cretan 29 Jun '14

29 Jun '14

I'm running a relay that I do not intend to use for anything else, so I set SocksPort to 0. I usually have two or three circuits established anyway, though, so I guess I haven't managed to disable creating those. I'm not sure what they are for. They are always labeled "Purpose: Ags=is_internal,need_capacity,". Several of them seem to get created around the same time, stay open for a while, and then get closed around the same time. I see some other circuit purposes documented here: https://stem.torproject.org/api/control.html But that documentation indicates that "Tor may provide purposes not in this enum", which is the case here. What are these circuits for, and do I need them? Here's my torrc: http://pastebin.com/fuQv8B2m Here's my ifconfig output: http://pastebin.com/mxMkhbMj I'm running tor 0.2.4.22 compiled from source (did not specify any options on make) on raspbian wheezy June 2014 with kernel version 3.12.20+. Thanks, Joel

2 3

Directory Server and bandwidth accounting
by Kali Tor 28 Jun '14

28 Jun '14

Hi, Is my understanding correct that if I set AccountingMax, the relay will never be used as a DR? It kind of feels odd because in my situation I can donate 500GB (and maybe even more) but I do want to keep a max limit and at the same time let the relay be a DR as well. Any way to achieve this? -kali-

4 5

BBC iPlayer blocked for UK middle relay operators?
by Chris Whittleston 28 Jun '14

28 Jun '14

Is anyone else running a relay in the UK having issues accessing BBC iPlayer from the same connection? For the last 6 weeks or so now I've been getting messages telling me I'm 'outside the UK' when I'm in London, and my IP address clearly resolves to show that. Just wondering if the provider BBC use for iPlayer might have started just blocking all Tor IPs recently - including middle relays. Chris

3 6

Bridge still advertising removed pluggable transport?
by Rick Huebner 28 Jun '14

28 Jun '14

Hi. I removed the recently-deprecated obfs2 transport from my (0.2.4.22) bridge's torrc, but after restarting, it seems to still be advertising support for it. It correctly doesn't appear as a "Registered server transport obfs2" line in the tor log file, and it's not listed in the TOR_PT_SERVER_TRANSPORTS or TOR_PT_SERVER_BINDADDR env vars fed to obfsproxy, nor is obfsproxy listening at the port that was formerly used for obfs2 any more. But... it's still listed in a "TransportProxy obfs2" line in the tor state file after several hours, and is still showing up as a supported transport in my bridge's details page on globe.torproject.org. It seems like that would mean it's still being handed out as a working bridge by BridgeDB for those requesting obfs2 bridges, though that port and transport will never be up again. And it might trigger mistaken "bridge is down" or "destination is censored" signals from those attempting to monitor such things. Is there something else I need to do besides dropping obfs2 from the torrc ServerTransportPlugin line, or is this just a bug? If so, it'll be showing up a lot more as obfs2 is phased out. Thanks.

2 1