Re: Tinc performance on a Dir-300
We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. What about packets per second ? Are you sending a lot of small packets ? That could be a problem ! my 2 cents. Saverio ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
Am Dienstag 21 September 2010, 10:13:30 schrieb ZioPRoTo (Saverio Proto): We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. What about packets per second ? Are you sending a lot of small packets ? That could be a problem ! I´m shure. I´m not that familar with networks. Our tinc device tap0 has an MTU of 1500 but it is in a bridge (br-mesh) wich has an MTU of 1476. Maybe you can have a look at this? -- r...@00240117b755:~# brctl show br-mesh bridge name bridge id STP enabled interfaces br-mesh 8000.00240117b755 yes ath0 bat0 tap0 r...@00240117b755:~# ifconfig ath0 Link encap:Ethernet HWaddr 00:24:01:17:B7:55 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:13119 errors:0 dropped:55 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 B) TX bytes:998226 (974.8 KiB) ath1 Link encap:Ethernet HWaddr 06:24:01:17:B7:55 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:701 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 B) TX bytes:163844 (160.0 KiB) br-mesh Link encap:Ethernet HWaddr 00:24:01:17:B7:55 inet6 addr: fe80::224:1ff:fe17:b755/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1476 Metric:1 RX packets:8588 errors:0 dropped:0 overruns:0 frame:0 TX packets:5705 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:1058896 (1.0 MiB) TX bytes:817276 (798.1 KiB) tap0 Link encap:Ethernet HWaddr 8E:33:AE:29:5B:16 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:12472 errors:0 dropped:0 overruns:0 frame:0 TX packets:18438 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:500 RX bytes:1776840 (1.6 MiB) TX bytes:1794668 (1.7 MiB) wifi0 Link encap:UNSPEC HWaddr 00-24-01-17- B7-55-00-00-00-00-00-00-00-00-00-00 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:514267 errors:0 dropped:0 overruns:0 frame:1242 TX packets:91806 errors:1439 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:195 RX bytes:7579 (67.8 MiB) TX bytes:9736303 (9.2 MiB) Interrupt:3 Memory:b000-b000 --- Greetings Clemens signature.asc Description: This is a digitally signed message part. ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
Our tinc device tap0 has an MTU of 1500 but it is in a bridge (br-mesh) wich has an MTU of 1476. Maybe you can have a look at this? OK, maybe you have a problem with packet fragmentation and you waste a lot of CPU. Try to put the MTU of your tap device to a lower value. Make this test MTU 1280 and add the following rule to your iptables firewall: iptables -A FORWARD -p tcp -m tcp --tcp-flags SYN,RST SYN -j TCPMSS --clamp-mss-to-pmtu this will cause new TCP connections to use segments that fit your interface MTU. Note that 1280 is not the optimal value, you can fine tune later if you see you get more speed. Saverio ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
On Tue, Sep 21, 2010 at 7:49 AM, Clemens John clemens-j...@gmx.de wrote: Am Dienstag 21 September 2010, 13:51:26 schrieb ZioPRoTo (Saverio Proto): this will cause new TCP connections to use segments that fit your interface MTU. I have had this problem too! Can you do the same for UDP connections? ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
moin, Am Montag, 20. September 2010, 16:37:03 schrieb Clemens John: we are using Tinc in our Freifunk Network in Oldenburg for internode connections over the internet. So Tinc is running on OpenWrt 10.03 on Dlink Dir-300 Routers. We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. We faced the same issues on our routers. They are used as mesh-nodes in a Freifunk network, too ;) Is it possible to get more Speed with tinc on this machines? I think we have compression and encryption already turned off so what is using the cpu? We tried to profile this, but it wasn’t easy to do that on the hardware, so we gave up on this :/ We also couldn’t see any difference when turning off encryption, IIRC. I remember running tinc using valgrind on my desktop and mentioned, that most time was spent with xor operations, so in the crypto part. I can’t remember many details but I concluded, that everything seems fine on the desktop. May be, you could try to reproduce this or get any better results than me. If you’re interested, I might have the data still laying around somewhere. Our Tinc configuration looks like this: - Name = 0014224074A7 Mode = Switch Port = 655 #PingTimeout = 30 Hostnames=yes PMTUDiscovery=yes Cipher = none Compress = 0 Digest = none IndirectData = yes ConnectTo=0021912CF309 ConnectTo=00240117B755 ConnectTo=batgw ConnectTo=0022B0967CD7 ConnectTo=0014224074A7 -- Seems, you are using our scripts, if I think about it ;) or you just had the same idea to use the routers’ MAC addresses as names in tinc ;) At least this seems very familiar to me. If there is no way to get more speed, do you know another VPN-Solution which is better concerning speed? We dont need security because the network is completely open, but we need speed. I expect, you want to use tinc to connect several mesh-clouds using tinc over some private Internet connections. That’s exactly, what we had in mind when we decided that we want to use tinc for that. It just seemed to be _the_ tool to do this job. Please keep me posted if you make any progress. (contact me directly or just post to our ML: freifunk.luebeck AT asta.uni-luebeck.de) bye then julian PS: At the moment, our group is a little inactive, because we didn’t have much time recently. But I think, at least I will be able to discuss these issues, may be we can find a solution for this. If you want us to test something or want to know details about our setup, just ask. signature.asc Description: This is a digitally signed message part. ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
That device uses the Atheros AR2317 processor which isn't exactly robust at 180Mhz. Have you considered alternative hardware? On Mon, Sep 20, 2010 at 10:37 AM, Clemens John clemens-j...@gmx.de wrote: Hi, we are using Tinc in our Freifunk Network in Oldenburg for internode connections over the internet. So Tinc is running on OpenWrt 10.03 on Dlink Dir-300 Routers. We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. Is it possible to get more Speed with tinc on this machines? I think we have compression and encryption already turned off so what is using the cpu? Our Tinc configuration looks like this: - Name = 0014224074A7 Mode = Switch Port = 655 #PingTimeout = 30 Hostnames=yes PMTUDiscovery=yes Cipher = none Compress = 0 Digest = none IndirectData = yes ConnectTo=0021912CF309 ConnectTo=00240117B755 ConnectTo=batgw ConnectTo=0022B0967CD7 ConnectTo=0014224074A7 -- If there is no way to get more speed, do you know another VPN-Solution which is better concerning speed? We dont need security because the network is completely open, but we need speed. Thank you Clemens ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
On Mon, Sep 20, 2010 at 04:37:03PM +0200, Clemens John wrote: we are using Tinc in our Freifunk Network in Oldenburg for internode connections over the internet. So Tinc is running on OpenWrt 10.03 on Dlink Dir-300 Routers. We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. Is it possible to get more Speed with tinc on this machines? I think we have compression and encryption already turned off so what is using the cpu? [...] If there is no way to get more speed, do you know another VPN-Solution which is better concerning speed? We dont need security because the network is completely open, but we need speed. Are you sure you have 1.6 MB/sec both upstream and downstream? You can try other VPN software, such as OpenVPN or perhaps VDE, and see if those are faster. Maybe you can also try profiling tinc (either with valgrind's callgrind tool, or compile tinc with -pg in both CFLAGS and LDFLAGS so you can use gprof) to see where the bottleneck is. However, the Dir-300 has a MIPS processor running at 182 MHz with only 16 kB instruction and 16 kB data cache. That means it is not very fast, especially not if a packet has to go all the way to userspace to be processed by tinc and back. -- Met vriendelijke groet / with kind regards, Guus Sliepen g...@tinc-vpn.org signature.asc Description: Digital signature ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
On Mon, Sep 20, 2010 at 11:27 AM, Guus Sliepen g...@tinc-vpn.org wrote: On Mon, Sep 20, 2010 at 04:37:03PM +0200, Clemens John wrote: we are using Tinc in our Freifunk Network in Oldenburg for internode connections over the internet. So Tinc is running on OpenWrt 10.03 on Dlink Dir-300 Routers. We all have enough internet bandwith (1,6 MB/sec and more) but we only get a maximum speed of ~350KB/sec between two tinc nodes because then tinc uses 99% of the cpu. Is it possible to get more Speed with tinc on this machines? I think we have compression and encryption already turned off so what is using the cpu? [...] If there is no way to get more speed, do you know another VPN-Solution which is better concerning speed? We dont need security because the network is completely open, but we need speed. Are you sure you have 1.6 MB/sec both upstream and downstream? You can try other VPN software, such as OpenVPN or perhaps VDE, and see if those are faster. Maybe you can also try profiling tinc (either with valgrind's callgrind tool, or compile tinc with -pg in both CFLAGS and LDFLAGS so you can use gprof) to see where the bottleneck is. However, the Dir-300 has a MIPS processor running at 182 MHz with only 16 kB instruction and 16 kB data cache. That means it is not very fast, especially not if a packet has to go all the way to userspace to be processed by tinc and back. -- Met vriendelijke groet / with kind regards, Guus Sliepen g...@tinc-vpn.org -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (GNU/Linux) iEYEARECAAYFAkyXi2AACgkQAxLow12M2nsYMgCeKWMubZfL1xpJrpaNEe0DKihi RLcAoIkxbuk0JU2SADcZ5uRNdUyvpuEI =3Owl -END PGP SIGNATURE- ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc Have you tried nice? On a full PC platform running Fedora 13 our gateway tinc node does much much better with nice. Even though there were plenty of unused CPU cycles, changing /etc/init.d/tincd by prepending 'nice -n 20' to tincd --net=YourNetName made performance much more consistent. Otherwise, ping times would range from 2milliseconds to 4000milliseconds. ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
Re: Tinc performance on a Dir-300
On Mon, Sep 20, 2010 at 02:29:44PM -0500, Rob Townley wrote: Have you tried nice? On a full PC platform running Fedora 13 our gateway tinc node does much much better with nice. Even though there were plenty of unused CPU cycles, changing /etc/init.d/tincd by prepending 'nice -n 20' to tincd --net=YourNetName made performance much more consistent. Otherwise, ping times would range from 2milliseconds to 4000milliseconds. Note that nice -n 20 actually lowers the priority. However, the kernel does behave differently when you nice a process, it could be that tinc does receive some kind of benefit from this. -- Met vriendelijke groet / with kind regards, Guus Sliepen g...@tinc-vpn.org signature.asc Description: Digital signature ___ tinc mailing list tinc@tinc-vpn.org http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc