| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| *** | himuraken has quit IRC (Ping timeout: 272 seconds) | [00:04] |
| himuraken has joined #arpnetworks | [00:16] | |
| .... (idle for 18mn) | ||
| himuraken has quit IRC (Ping timeout: 256 seconds) | [00:34] | |
| himuraken has joined #arpnetworks | [00:46] | |
| ........................ (idle for 1h56mn) | ||
| himuraken has quit IRC (Ping timeout: 245 seconds) | [02:42] | |
| himuraken has joined #arpnetworks | [02:54] | |
| ................................... (idle for 2h54mn) | ||
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [05:48] | |
| ..... (idle for 20mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [06:08] | |
| ..... (idle for 21mn) | ||
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [06:29] | |
| ...... (idle for 27mn) | ||
| amdprophet has joined #arpnetworks | [06:56] | |
| dzup has quit IRC (Ping timeout: 246 seconds) | [07:10] | |
| .... (idle for 19mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [07:29] | |
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [07:39] | |
| ......... (idle for 41mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [08:20] | |
| .......................... (idle for 2h5mn) | ||
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [10:25] | |
| ...... (idle for 28mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [10:53] | |
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [11:04] | |
| heavysixer has quit IRC (Quit: heavysixer) | [11:17] | |
| ....... (idle for 34mn) | ||
| mnathani has quit IRC (Read error: Connection reset by peer)
mnathani has joined #arpnetworks | [11:51] | |
| ............ (idle for 59mn) | ||
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [12:51] | |
| ............... (idle for 1h12mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [14:03] | |
| scottschecter has quit IRC (Quit: WeeChat 0.4.0) | [14:12] | |
| ..... (idle for 23mn) | ||
| scottschecter has joined #arpnetworks
amdprophet has quit IRC (Remote host closed the connection) | [14:35] | |
| ........ (idle for 38mn) | ||
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [15:17] | |
| ..... (idle for 20mn) | ||
| heavysixer has quit IRC (Quit: heavysixer) | [15:37] | |
| ............................... (idle for 2h30mn) | ||
| bGeorge has quit IRC (Quit: Bye.) | [18:07] | |
| heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer henderb has quit IRC (Ping timeout: 255 seconds) henderb has joined #arpnetworks bGeorge has joined #arpnetworks | [18:19] | |
| heavysixer has quit IRC (Quit: heavysixer)
jm|laptop has joined #arpnetworks | [18:34] | |
| jm|laptop | hi
I have the super-basic "small" VPS plan is my speed limited? I'm trying to download a dataset from my VPS to here - and it seems really throttled to like 1M | [18:35] |
| *** | amdprophet has joined #arpnetworks | [18:39] |
| mnathani | jm | [18:40] |
| jm|laptop | iperf seems to bear it out :/ | [18:40] |
| mnathani | jm|laptop: where is "here"
VPS should be able to hit around 100mbit | [18:40] |
| jm|laptop | here is my home fibre dual-xDSL | [18:41] |
| mnathani | east coast? west coast? europe? asia? | [18:41] |
| jm|laptop | oh. UK | [18:42] |
| mnathani | k | [18:42] |
| jm|laptop | icmp_req=1 ttl=44 time=166 ms | [18:42] |
| mnathani | v4 or v6? | [18:43] |
| jm|laptop | both / either
scp ipv6 and http ipv4 to prove it wasn't ssh overhead and iperf ipv4 it just stalled again >.< | [18:44] |
| mnathani | try the arpnetworks speed test which is on gigabit: http://arpnetworks.com/100mb.bin | [18:46] |
| jm|laptop | that's coming at like 251K/s
although I still have my other scp going on jm|laptop mtrs no obvious loss, slows down cross-Atlantic so I have a colo in Manchester pulling 1.7M from that .bin, ramhost London pulls ~500K ramhost down to 120K now I wonder if this is a peering thing | [18:47] |
| CaZe | moar routes | [18:53] |
| mercutio | could be
or transit thing | [18:54] |
| jm|laptop | seems to go myISP > L3 LON > L3 NYC > nlayer > us.pktxchng > mzima
who is nlayer? | [18:55] |
| mercutio | nlayer is fine
check other direction some of the any2ix peers can be kind of slow with transit back atrato i've noticed be slow before for example | [18:55] |
| jm|laptop | mzima > big pause > nlayer > L3 LA > L3 SJ > LA NYC x3 > L3 LON > myISP
it really bounces around L3 :S | [18:56] |
| mercutio | that really should be fine unless your isp has congestion at level 3 | [18:57] |
| jm|laptop | 18 level3.restless.thn.aaisp.net.uk (195.50.95.67) 145.879 ms 146.302 ms 146.025 ms
my isp peers L3 | [18:57] |
| mercutio | 9. ae-82-82.ebr2.NewYork1.Level3.net 0.0% 6 138.8 139.0 138.8 139.5 0.3
ae-72-72.ebr2.NewYork1.Level3.net like that etc? | [18:57] |
| jm|laptop | yeah. With an ominous * * * in between | [18:57] |
| mercutio | mtr is good at skipping over that :) | [18:57] |
| jm|laptop | 14 ae-91-91.ebr1.NewYork1.Level3.net (4.69.134.77) 146.221 ms ae-61-61.ebr1.NewYork1.Level3.net (4.69.134.65) 147.240 ms ae-91-91.ebr1.NewYork1.Level3.net (4.69.134.77) 146.204 ms
15 * * * 16 ae-59-224.csw2.London1.Level3.net (4.69.153.142) 147.207 ms 147.286 ms ae-56-221.csw2.London1.Level3.net (4.69.153.130) | [18:58] |
| mercutio | it does that here on another route not to arp from a uk vps | [18:58] |
| jm|laptop | I guess its sub-Atlantic node :) | [18:58] |
| mercutio | with like two dead hops between london and newyork
you have the same mpls exit path thingy where new york shows high ping like london | [18:58] |
| jm|laptop | 150ms ? | [18:59] |
| mercutio | yeh it's the same on london and new york though
rather than showing the latency from london to new york you're not using openbsd or something are you? | [18:59] |
| jm|laptop | ..... where? | [19:01] |
| mercutio | openbsd's tcp/ip stack can be kind of bad for perfomrance transatlanttic | [19:01] |
| jm|laptop | my router runs OpenBSD | [19:01] |
| mercutio | takes longer to ramp up only seems to increase window size with scaling
router doens't matter it's only end point | [19:01] |
| jm|laptop | then no, Debian | [19:02] |
| mercutio | yeah debian defaults should be fine even | [19:02] |
| jm|laptop | so ARP /doesn't/ throttle? | [19:02] |
| mercutio | nah it doesn't throttle
and level3 usually has good speeds unless someone doesn't buy enough level 3 transit or such | [19:02] |
| jm|laptop | just coincidence that my monitor is showing almost /exactly/ 1M :) | [19:03] |
| mercutio | you're at like 150 msec altency? | [19:03] |
| jm|laptop | I'm not at liberty to disclose my altency | [19:03] |
| mercutio | well guessing that | [19:04] |
| jm|laptop | 3 packets transmitted, 3 received, 0% packet loss, time 2002ms
rtt min/avg/max/mdev = 166.583/167.351/167.918/0.735 ms | [19:04] |
| mercutio | your window size would be about 160k | [19:04] |
| *** | heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer | [19:04] |
| mercutio | which isn't generally a size anything will cap at | [19:04] |
| jm|laptop | when my scp finishes I might try another iperf | [19:04] |
| mercutio | ohh
it's not old scp is it? | [19:05] |
| jm|laptop | no | [19:05] |
| mercutio | ok | [19:05] |
| jm|laptop | and I tried http also! (see above) | [19:05] |
| mercutio | ssh used to have window size issues
ahh ok well assuming you have more than 128mb ram | [19:05] |
| jm|laptop | [02:44] <jm|laptop> scp ipv6 and http ipv4 to prove it wasn't ssh overhead | [19:05] |
| mercutio | linux should pick big enough tcp window size
ipv6 was the same? | [19:05] |
| jm|laptop | yup
scp is ipv6 | [19:05] |
| mercutio | ok that's really bizzare
i've often found ipv6 and ipv4 give completely diff performance sometimes one is like twice as fast as the other | [19:06] |
| jm|laptop | doesn't make a lot of sense | [19:06] |
| mercutio | yeah | [19:06] |
| jm|laptop | unless the whole ipv6 topology is different
which doesn't make a lot of sense | [19:06] |
| mercutio | oh it is often i think
i dunno i even reproduced that with the same provider being used for ipv4 and ipv6 it may just be load balancing tkaing diff path | [19:06] |
| jm|laptop | were they cheap? | [19:07] |
| mercutio | suppose?
i dunno | [19:07] |
| jm|laptop | maybe the employed a different ipv6 gateway | [19:07] |
| mercutio | well arp dose
does | [19:07] |
| jm|laptop | s/different/less congested/ | [19:07] |
| mercutio | i can't remember which was faster
sometimes people don't shape ipv6 properly etc too and shape ipv4 | [19:07] |
| jm|laptop | hm
722141184 98% 407.45kB/s 0:00:29 | [19:08] |
| mercutio | ok i need to try again :) | [19:08] |
| jm|laptop | hmm that's weirder still | [19:09] |
| mercutio | weird both are going slow atm | [19:09] |
| jm|laptop | I had two scp making that 1M | [19:09] |
| mercutio | dude i'm getting terrible speeds compared to normal from this host too
it's a friends connection but it usually does faster | [19:09] |
| jm|laptop | and one has finished so I expected other to take 'spare' bandwidth
but it's 500k wtf 587956224 80% 385.63kB/s 0:06:10 like something is making 500K per connection :S | [19:09] |
| mercutio | i got 647k/sec with ipv6 and 509k/sec with ipv4
for a 10 mb file with my normal provider getting 2267k/sec average 130 msec ping versus 138 msec on the slow | [19:10] |
| jm|laptop | [ 3] 0.0-10.2 sec 7.12 MBytes 5.83 Mbits/sec
^^ my other colo | [19:11] |
| mercutio | and 139 msec with ipv6 on the slow
hmm % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 10.0M 100 10.0M 0 0 509k 0 0:00:20 0:00:20 --:--:-- 504k and that's with any2ix peer | [19:11] |
| jm|laptop | (and my values include wireless overhead to the laptop) | [19:12] |
| mercutio | it's prob this stupid upstream :/ | [19:12] |
| jm|laptop | oh well | [19:12] |
| mercutio | wireless doesn't really slow things down unless you're far away | [19:12] |
| jm|laptop | part1 of my dataset is here so I can grok it | [19:12] |
| mercutio | oh there's packet loss to last hop that's really weird | [19:12] |
| jm|laptop | thanks for clarifications :) | [19:13] |
| mercutio | well it can limit you, but doesn't slow down unless you're past bandwidth and it's giving high pings locally
basically | [19:13] |
| jm|laptop | wait what?
how can I check my bandwidth limit? | [19:13] |
| *** | _mnathani_ has joined #arpnetworks | [19:14] |
| mercutio | i meant on wireless
like if your wirelss can do 20 megabit | [19:14] |
| jm|laptop | oh | [19:14] |
| mercutio | and you're pushing 25 megabit it'll slow it down to 20 megabit | [19:14] |
| jm|laptop | I have 5GHz N | [19:14] |
| mercutio | but if you're pushing 5 megabit then it'll be same speed on wireless and ethernet
it doesn't tend to give loss it can, but usually bandwidth is heavily reduced before that pings spike etc | [19:14] |
| jm|laptop | [ 3] 0.0-10.0 sec 126 MBytes 106 Mbits/sec
not bad for wireless :) | [19:15] |
| mercutio | yeah
so shouldn't be an issue cos you have dual dsl oh how does your dual dsl work? | [19:15] |
| jm|laptop | I pull about 50M from my dual xdsl | [19:15] |
| mercutio | do you have to do multiple connections to max it out?
or is it bonding? | [19:15] |
| jm|laptop | bondage | [19:15] |
| mercutio | cool
mlpppp :) | [19:15] |
| jm|laptop | nth packet downstream | [19:16] |
| mercutio | oh | [19:16] |
| jm|laptop | no | [19:16] |
| mercutio | that's better
so it doesn't split each packet half between each does that give out of order issues? | [19:16] |
| jm|laptop | aiui it balances based on latency from LCP data
yes; but TCP/IP is designed to deal with that :) | [19:16] |
| mercutio | err
tcp/ip sucks with out of orer :) perfomrance wise it's more designed to cope than deals well with :) | [19:17] |
| jm|laptop | worksforme dot com | [19:17] |
| mercutio | heh | [19:17] |
| jm|laptop | if I had a more expensive router at this end I could handle it better
but obsd router doesn't do too bad a job | [19:17] |
| mercutio | i'm surprised it looks at lcp data
i wanted to play with bonding some day i was trying to think up novel ways to do it better | [19:18] |
| jm|laptop | my ISP is pretty switched on with this stuff
I get per second graphing and stuff | [19:18] |
| mercutio | any idea how they're doing it?
i was playing with ppp compression a while back it works fine until you get packet drops and it seems to make more sense on upstream than downstream for adsl | [19:18] |
| jm|laptop | they made their own router | [19:19] |
| mercutio | oh wow
ok that's good to know i didn't think anything was doing those kinds of smarts normally | [19:19] |
| jm|laptop | http://wiki.aa.org.uk/index.php/FireBrick
http://www.firebrick.co.uk/products_2700.php this http://www.firebrick.co.uk/ or indeed this :'/ | [19:20] |
| mercutio | oh
i read something on this site before about some special ppp server i think | [19:21] |
| jm|laptop | which part of the PPP sever?
it's not that clever :/ | [19:21] |
| mercutio | To bond downlink on DSL lines you need the ISP to send traffic down multiple lines. The FB6202 LNS can do this on multiple L2TP links with per link speed controls. Bonding downlink does not need anything at the receiving end as the packets simply arrive on one of multiple simple broadband routers and go on to the LAN. It is obviously a good idea to use an FB2700 at the customer end to also bond uplink
though. it's not acutally specific to the termination device | [19:21] |
| *** | mnathani has quit IRC (*.net *.split)
Ehtyar has quit IRC (*.net *.split) hazardous has quit IRC (*.net *.split) Lucifer7 has quit IRC (*.net *.split) toddf has quit IRC (*.net *.split) | [19:21] |
| mercutio | but the sending ppp
Because bonding works on a per-packet basis, even a single session can make use of multiple lines. However, it is worth bearing in mind that whilst IP does not guarantee packet order, most TCP stacks will have trouble if you bond more than about 4 lines or if the lines are very different latency. | [19:21] |
| jm|laptop | that's multihoming | [19:22] |
| mercutio | nah that's bonding dsl | [19:22] |
| jm|laptop | the former
re: the latter, I only have two :) | [19:22] |
| mercutio | it's both under bonding
yeh but it does out of order a bit ok | [19:23] |
| jm|laptop | and the latency is pretty much similar because it comes to my house on the same 6-string drop :) | [19:23] |
| mercutio | it's actually really easy for it to be different
if lacp etc is being used | [19:23] |
| jm|laptop | yup
again: wfm.com | [19:24] |
| mercutio | it'll matter less for big packets though | [19:24] |
| jm|laptop | I have poor-mans bonding at this end
an Alix mpath routing | [19:24] |
| mercutio | anyway, i'd try a udp test | [19:24] |
| jm|laptop | whence? | [19:24] |
| mercutio | iperf -u | [19:24] |
| jm|laptop | whence? | [19:24] |
| mercutio | which'll tell you if there' any packet loss
and it'll tell you how much is out of order if there's loss it's prob reason it's going slow | [19:24] |
| jm|laptop | [ ID] Interval Transfer Bandwidth
[ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec [ 3] Sent 893 datagrams [ 3] WARNING: did not receive ack of last datagram after 10 tries. | [19:25] |
| mercutio | if there's no loss and no out of order your isp may be doing something weird with tcp/ip traffic
you need iperf -s -u on server iperf -c -b 2m on client where client uploads to server well 1mbit is fine too | [19:25] |
| jm|laptop | [ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec 0.406 ms 2/ 893 (0.22%) | [19:26] |
| mercutio | you can use -i 1 as well on server to list per second
hmm | [19:26] |
| jm|laptop | hmm indeed :) | [19:26] |
| mercutio | maybe try -l 512
on both to send smaller packets and i think it's -t 30 to do 30 second test | [19:26] |
| jm|laptop | [ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec 0.225 ms 0/ 2561 (0%)
[ 3] 0.0-10.0 sec 1 datagrams received out-of-order | [19:27] |
| mercutio | that seems fine
hmmm | [19:27] |
| jm|laptop | I know: except for the speed :) | [19:28] |
| mercutio | that's cos you have -1m
-b 1m try -b2m | [19:28] |
| jm|laptop | hm | [19:28] |
| mercutio | udp sends at fixed speed
but generally speaking if udp is fine at low speeds without loss then the link is going ok | [19:28] |
| jm|laptop | [ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec 0.216 ms 0/ 2561 (0%)
[ 3] 0.0-10.0 sec 1 datagrams received out-of-order something really only wants 1Mb/s :) | [19:28] |
| mercutio | -b 2m
on the client | [19:29] |
| jm|laptop | I did that | [19:29] |
| mercutio | oh
are you sending from your dsl to arp? you have to do arp to dsl | [19:29] |
| jm|laptop | $ iperf -u -l 512 -b2m -c arp | [19:29] |
| mercutio | yeh do it other way around
cos client sends to server | [19:29] |
| jm|laptop | how? I have NAT?
JUST KIDDING | [19:29] |
| mercutio | port forward? | [19:29] |
| jm|laptop | my ISP is better than that :) | [19:29] |
| mercutio | you have /64 ? | [19:29] |
| jm|laptop | I have /26 and /48 actually | [19:30] |
| mercutio | you can do ipv6 with iperf but you need to set -V
and if you have differnt default udp return addresss than incoming address you need to bind on the server to the incoming address -B <ip> i think it is it seems much more complicated when desciribing it | [19:30] |
| jm|laptop | [ 3] 0.0-10.0 sec 1.25 MBytes 1.05 Mbits/sec 0.224 ms 1/ 2562 (0.039%)
[ 3] 0.0-10.0 sec 1 datagrams received out-of-order other direction something somewhere really seems to be restricting me to 1M :) anyway: off to research my dataset thanks for your time mercutio :) | [19:31] |
| *** | Ehtyar has joined #arpnetworks | [19:33] |
| mercutio | i'll just it | [19:34] |
| *** | up_the_irons has quit IRC (Ping timeout: 264 seconds)
mikeputnam has quit IRC (Ping timeout: 264 seconds) DiaboliK has quit IRC (Ping timeout: 264 seconds) [FBI] has quit IRC (Ping timeout: 264 seconds) [FBI] starts logging #arpnetworks at Sun Mar 31 19:49:47 2013 [FBI] has joined #arpnetworks | [19:41] |
| up_the_irons has joined #arpnetworks
ChanServ sets mode: +o up_the_irons | [19:54] | |
| himuraken has quit IRC (Ping timeout: 255 seconds) | [20:04] | |
| up_the_irons | mikeputn1m: lol nice iron maiden references earlier... | [20:04] |
| *** | dzup has joined #arpnetworks | [20:08] |
| mercutio | up_the_irons: do you normally allow /23s into your network now? | [20:10] |
| up_the_irons | mercutio: if it's from a peer
soon, i'll have everything, with the new router brb | [20:13] |
| staticsafe | new router o/ | [20:15] |
| *** | himuraken has joined #arpnetworks | [20:16] |
| mercutio | up_the_irons: trit's not a peer is it?
everything would be good, it's accepting a way worse secondary route | [20:18] |
| ............ (idle for 59mn) | ||
| *** | _mnathani_ has quit IRC ()
himuraken has quit IRC (Ping timeout: 255 seconds) | [21:17] |
| himuraken has joined #arpnetworks | [21:29] | |
| .... (idle for 19mn) | ||
| dzup has quit IRC (Ping timeout: 245 seconds) | [21:48] | |
| mnathani has joined #arpnetworks
dzup has joined #arpnetworks | [21:53] | |
| himuraken has quit IRC (Ping timeout: 240 seconds) | [22:01] | |
| himuraken has joined #arpnetworks | [22:13] | |
| .... (idle for 18mn) | ||
| himuraken has quit IRC (Ping timeout: 256 seconds) | [22:31] | |
| himuraken has joined #arpnetworks | [22:43] | |
| ............. (idle for 1h4mn) | ||
| Webhostbudd has joined #arpnetworks | [23:47] | |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |