Hello All, I subscribe to Linux Journal, which as gone digital only. When I go to get the digital copy, things time out with a connection closed message when I use wget. It can be anywhere from less than 1% to somewhere below 20%, and currently not doing partial downloads to continue, having to restart from scratch. I ran traceroute, and attach the results. Some of the times appear way out of whack. I am on the Telstra BigPond Ultimate Dial Up Plan, so each time I "connect", it is a local call fee, and a maximum of ten hours before being disconnected, along with the limitations of an _external_ 56K hardware modem. Just to put the lie to some of my experience, the current invocation is past 22%, but the traceroute was running concurrently. That should not have an effect, but then I think that some of the systems are badly misconfigured. It would be good to put a suitable "explosive" under the irresponsible sysadmins, and the incompetent management. Regards, Mark Trickett
On 03/08/12 21:34, Mark Trickett wrote:
Hello All,
I subscribe to Linux Journal, which as gone digital only. When I go to get the digital copy, things time out with a connection closed message when I use wget. It can be anywhere from less than 1% to somewhere below 20%, and currently not doing partial downloads to continue, having to restart from scratch.
I ran traceroute, and attach the results. Some of the times appear way out of whack. I am on the Telstra BigPond Ultimate Dial Up Plan, so each time I "connect", it is a local call fee, and a maximum of ten hours before being disconnected, along with the limitations of an _external_ 56K hardware modem.
Just to put the lie to some of my experience, the current invocation is past 22%, but the traceroute was running concurrently. That should not have an effect, but then I think that some of the systems are badly misconfigured. It would be good to put a suitable "explosive" under the irresponsible sysadmins, and the incompetent management.
Regards,
Mark Trickett
_______________________________________________ luv-main mailing list luv-main@luv.asn.au http://lists.luv.asn.au/listinfo/luv-main
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/ It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
Hello Duncan, I have trimmed the quoted material. On Fri, 2012-08-03 at 21:45 +1000, Allan Duncan wrote:
On 03/08/12 21:34, Mark Trickett wrote:
Hello All,
Just to put the lie to some of my experience, the current invocation is past 22%, but the traceroute was running concurrently. That should not have an effect, but then I think that some of the systems are badly misconfigured. It would be good to put a suitable "explosive" under the irresponsible sysadmins, and the incompetent management.
Regards,
Mark Trickett
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
Interesting, I would like to read, but that page is only available to Internode customers. They do offer broadband connections, but that is not an option, my copper pair is terminated on equipment that can provide ISDN, but no longer offered or supported, but not ADSL. I am considering the wireless options, but they have an unreasonable price premium for more than a minimal data quota.
It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
Regards, Mark Trickett
On Fri, Aug 03, 2012 at 09:45:57PM +1000, Allan Duncan wrote:
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
well spotted. Mark might also want to try mtr rather than traceroute. It can display MPLS labels (press "u" in the curses interface or use "-e" on the command line) i usually use the --curses mode but the --report mode is more like traditional traceroute (and IMO more readable) e.g. $ mtr -e --report --report-wide download.linuxjournal.com HOST: ganesh Loss% Snt Last Avg Best Wrst StDev 1.|-- nexthop.vic.iinet.net.au 0.0% 10 14.6 14.2 13.8 14.6 0.3 2.|-- te7-2.mel-pipe-bdr1.iinet.net.au 0.0% 10 14.5 14.2 13.9 14.8 0.3 3.|-- xe-0-1-0-0.syd-ult-core1.iinet.net.au 0.0% 10 24.7 24.4 23.9 25.0 0.3 4.|-- te3-1-0.bdr1.syd7.internode.on.net 0.0% 10 23.7 23.5 22.7 24.7 0.5 5.|-- po3-0-0.bdr1.sjc2.internode.on.net 0.0% 10 205.9 205.9 205.4 206.3 0.3 6.|-- te1-7.bbr01.eq01.sjc01.networklayer.com 0.0% 10 204.8 204.2 203.7 204.8 0.4 7.|-- ae7.bbr02.eq01.sjc02.networklayer.com 0.0% 10 205.5 205.7 205.3 206.2 0.3 | +-- [MPLS: Lbl 356132 Exp 0 S 1 TTL 1] 8.|-- ae0.bbr02.cs01.lax01.networklayer.com 0.0% 10 212.5 212.8 212.4 213.2 0.3 | +-- [MPLS: Lbl 308816 Exp 0 S 1 TTL 1] 9.|-- ae7.bbr01.cs01.lax01.networklayer.com 0.0% 10 213.7 213.9 212.4 220.7 2.5 | +-- [MPLS: Lbl 567317 Exp 0 S 1 TTL 1] 10.|-- ae19.bbr01.eq01.dal03.networklayer.com 20.0% 10 241.7 241.6 241.3 242.0 0.3 11.|-- po31.dsr01.dllstx3.networklayer.com 0.0% 10 243.3 243.8 243.2 246.8 1.1 12.|-- te4-1.dsr01.dllstx2.networklayer.com 10.0% 10 241.7 243.3 241.2 254.0 4.1 | +-- [MPLS: Lbl 13666 Exp 0 S 1 TTL 1] 13.|-- te3-1.car16.dllstx6.networklayer.com 10.0% 10 244.6 244.6 244.3 245.2 0.3 14.|-- saratoga.site5.com 10.0% 10 239.8 239.9 239.6 240.5 0.3 I think Mark has mentioned problems downloading the Linux Journal before. Makes me wonder what the MTU and MRU on his PPP interface is and whether he (or Telstra) is blocking ICMP packets (in particular ICMP Fragmentation Required msgs). Mark, try setting the MTU and MRU to some fairly low value (the default of 1500 is too big for a 56K modem connection anyway). something like 600 or 900 each perhaps. and then see if the download works. IF this works, there'll be a bit more throughput loss to packet overhead (more and smaller packets) but a slightly slower download that works is better than one that doesn't :) it's been far too long since i had a modem connection, i can't remember what I used to set mine at. 600 is a good test value - if that doesn't help then MRU and MTU issues are not the source of the problem and there's no point wasting time trying to find the highest value that works reliably (just set them back to default values). If it does help then gradually increase them (in increments of 50 or 100) until the download stops working again - you have now discovered a good value for MTU and MRU for your connection. craig -- craig sanders <cas@taz.net.au>
On 04/08/2012, at 8:59, Craig Sanders <cas@taz.net.au> wrote:
On Fri, Aug 03, 2012 at 09:45:57PM +1000, Allan Duncan wrote:
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
well spotted.
<snip>
craig
I take it you're both on Internode as that link gives me a 403 page. "Access to this content or feature is Restricted Sorry, but access to this content or feature is restricted to Internode Customers connected to Internode's network. You must be connected via Internode's network to view or use this content or feature. If you would like to obtain an Internode broadband service please click here." Edward
Edward Savage wrote:
On 04/08/2012, at 8:59, Craig Sanders <cas@taz.net.au> wrote:
http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/ I take it you're both on Internode as that link gives me a 403 page.
I can't in good conscience circumvent that for you, but I will give you the link off that page to a "more technical version" by Cisco: http://www.cisco.com/en/US/tech/tk436/tk428/technologies_tech_note09186a0080... (And yes, iinet bought internode late ast year: https://en.wikipedia.org/wiki/Internode_(ISP)#Acquisition_by_iiNet )
On 04/08/12 10:02, Edward Savage wrote:
On 04/08/2012, at 8:59, Craig Sanders <cas@taz.net.au <mailto:cas@taz.net.au>> wrote:
On Fri, Aug 03, 2012 at 09:45:57PM +1000, Allan Duncan wrote:
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
well spotted.
<snip>
craig
I take it you're both on Internode as that link gives me a 403 page.
"Access to this content or feature is Restricted
Ah, sorry about that. I'd summarise what it said, but there is another reference in this thread for a Cisco page.
On 04/08/2012, at 8:59 AM, Craig Sanders <cas@taz.net.au> wrote:
Mark might also want to try mtr rather than traceroute. It can display MPLS labels (press "u" in the curses interface or use "-e" on the command line)
i usually use the --curses mode but the --report mode is more like traditional traceroute (and IMO more readable)
I thought I'd play along at home - I've had some network performance issues at home which I thought were WiFi related but as I'd never used mtr I installed it and gave it a go: [root@kerberos ~]# mtr --report -c 10 -e www.latrobe.edu.au HOST: kerberos.gb.davies.net.au Loss% Snt Last Avg Best Wrst StDev 1.|-- upstairs 0.0% 10 0.8 0.8 0.8 1.1 0.1 2.|-- 10.79.0.1 0.0% 10 812.6 786.3 515.2 1206. 279.5 3.|-- riv3-pos3-4.gw.optusnet.c 0.0% 10 797.9 778.4 462.5 1205. 270.8 4.|-- xe-6-1-0.md12.optus.net.a 0.0% 10 784.4 779.7 539.5 1205. 253.0 5.|-- ??? 100.0 10 0.0 0.0 0.0 0.0 0.0 6.|-- TenGigE0-3-0-5.lon70.Melb 20.0% 10 739.8 762.1 541.1 1227. 277.0 7.|-- Bundle-Ether2.exi-core1.M 10.0% 10 728.9 762.4 551.1 1199. 259.5 8.|-- TenGigabitEthernet9-1.lon 20.0% 10 710.4 689.7 546.2 1206. 218.7 9.|-- aarnet4.lnk.telstra.net 11.1% 9 697.9 682.9 549.4 1180. 206.4 10.|-- ge-5-0-0.bb1.b.mel.aarnet 11.1% 9 683.2 674.9 537.1 1173. 208.9 11.|-- gigabitethernet0.er1.latr 11.1% 9 740.9 686.7 548.7 1168. 207.3 12.|-- gw1.er1.latrobe.cpe.aarne 11.1% 9 774.7 728.6 534.5 1185. 217.4 13.|-- 131.172.1.30 11.1% 9 807.4 703.9 520.9 1208. 225.6 14.|-- ??? 100.0 8 0.0 0.0 0.0 0.0 0.0 15.|-- ??? 100.0 8 0.0 0.0 0.0 0.0 0.0 16.|-- latrobe.edu.au 12.5% 8 767.5 622.7 520.8 767.5 88.6 Ouch - 700+ms pings to my Optus cable box. Once I work out how to power cycle it (it's a newish cable modem with built in battery backup) I'll see if this helps - any other suggestions welcome… Huw Davies | e-mail: Huw.Davies@kerberos.davies.net.au Melbourne | "If soccer was meant to be played in the Australia | air, the sky would be painted green"
Following this thread with interest I thought I'd have a look too. I thought I'd play along at home - I've had some network performance issues at home which I thought were WiFi related but as I'd never used mtr I installed it and gave it a go: [root@kerberos ~]# mtr --report -c 10 -e www.latrobe.edu.au HOST: kerberos.gb.davies.net.au Loss% Snt Last Avg Best Wrst StDev 1.|-- upstairs 0.0% 10 0.8 0.8 0.8 1.1 0.1 2.|-- 10.79.0.1 0.0% 10 812.6 786.3 515.2 1206. 279.5 3.|-- riv3-pos3-4.gw.optusnet.c 0.0% 10 797.9 778.4 462.5 1205. 270.8 4.|-- xe-6-1-0.md12.optus.net.a 0.0% 10 784.4 779.7 539.5 1205. 253.0 5.|-- ??? 100.0 10 0.0 0.0 0.0 0.0 0.0 6.|-- TenGigE0-3-0-5.lon70.Melb 20.0% 10 739.8 762.1 541.1 1227. 277.0 7.|-- Bundle-Ether2.exi-core1.M 10.0% 10 728.9 762.4 551.1 1199. 259.5 8.|-- TenGigabitEthernet9-1.lon 20.0% 10 710.4 689.7 546.2 1206. 218.7 9.|-- aarnet4.lnk.telstra.net 11.1% 9 697.9 682.9 549.4 1180. 206.4 10.|-- ge-5-0-0.bb1.b.mel.aarnet 11.1% 9 683.2 674.9 537.1 1173. 208.9 11.|-- gigabitethernet0.er1.latr 11.1% 9 740.9 686.7 548.7 1168. 207.3 12.|-- gw1.er1.latrobe.cpe.aarne 11.1% 9 774.7 728.6 534.5 1185. 217.4 13.|-- 131.172.1.30 11.1% 9 807.4 703.9 520.9 1208. 225.6 14.|-- ??? 100.0 8 0.0 0.0 0.0 0.0 0.0 15.|-- ??? 100.0 8 0.0 0.0 0.0 0.0 0.0 16.|-- latrobe.edu.au 12.5% 8 767.5 622.7 520.8 767.5 88.6 Ouch - 700+ms pings to my Optus cable box. My numbers below may give you some sort of reference point - using a not really brilliant ADSL2 connection bob@Comp-NC:~$ mtr --report -c 10 -e www.latrobe.edu.au mtr: invalid option -- 'e' HOST: Comp-NC Loss% Snt Last Avg Best Wrst StDev 1.|-- 192.168.72.200 0.0% 10 0.4 0.4 0.4 0.6 0.0 2.|-- ge1-0-306.adsl0.vic.aanet 0.0% 10 43.4 41.8 40.9 43.4 0.7 3.|-- te0-0-0-306.core0.mel.dft 0.0% 10 43.0 42.0 40.6 43.3 1.0 4.|-- telstra.mel.dft.com.au 0.0% 10 43.6 43.1 41.9 44.1 0.7 5.|-- Bundle-Ether11.win24.Melb 0.0% 10 49.9 45.8 41.1 51.3 3.6 6.|-- Bundle-Ether6.win-core1.m 0.0% 10 44.0 48.9 41.9 54.8 4.0 7.|-- TenGigabitEthernet9-2.lon 0.0% 10 43.4 63.9 40.9 160.1 44.4 8.|-- aarnet4.lnk.telstra.net 0.0% 10 43.0 42.7 41.7 43.3 0.6 9.|-- ge-5-0-0.bb1.b.mel.aarnet 0.0% 10 42.3 43.1 42.0 44.2 0.7 10.|-- gigabitethernet0.er1.latr 0.0% 10 43.0 46.4 41.7 78.9 11.4 11.|-- gw1.er1.latrobe.cpe.aarne 0.0% 10 154.8 68.6 41.4 172.8 50.5 12.|-- 131.172.1.30 0.0% 10 44.8 43.4 42.3 44.8 0.8 13.|-- ??? 100.0 10 0.0 0.0 0.0 0.0 0.0 14.|-- ??? 100.0 10 0.0 0.0 0.0 0.0 0.0 15.|-- www.latrobe.edu.au 0.0% 10 43.0 43.4 41.7 44.8 0.9 Bob
Bob <enquiries@lorneholidays.com.au> wrote:
Ouch - 700+ms pings to my Optus cable box.
It might be worth setting up a ping from your local host to that box and letting it run for a while just to confirm. You could try it with different traffic loads too. I think the said box might benefit from a firmware upgrade, if there is one available.
On Sun, 2012-08-05 at 15:49 +1000, Jason White wrote:
Bob <enquiries@lorneholidays.com.au> wrote:
Ouch - 700+ms pings to my Optus cable box.
It might be worth setting up a ping from your local host to that box and letting it run for a while just to confirm. You could try it with different traffic loads too.
I think the said box might benefit from a firmware upgrade, if there is one available.
_______________________________________________ luv-main mailing list luv-main@luv.asn.au http://lists.luv.asn.au/listinfo/luv-main
Did I miss the start of this thread? How fast is the network card in said box? $35 might fix it? Andrew Greig
Craig Sanders <cas@taz.net.au> wrote:
Mark might also want to try mtr rather than traceroute. It can display MPLS labels (press "u" in the curses interface or use "-e" on the command line)
i usually use the --curses mode but the --report mode is more like traditional traceroute (and IMO more readable)
Thanks, that's interesting and useful to know. I hadn't looked at it earlier. The manual page warns that it generates a lot of network traffic; I hope it isn't enough traffic to cause network providers any concerns. In my test, it sent 10 packets by default to each host along the path, i.e., with a suitably adjusted TTL. I wouldn't expect anyone to be worried about the additional load on their infrastructure that results from that level of traffic.
Hello Craig, On Sat, 2012-08-04 at 08:59 +1000, Craig Sanders wrote:
On Fri, Aug 03, 2012 at 09:45:57PM +1000, Allan Duncan wrote:
I suggest you read this first: http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
It won't help you with your download problem, but it helps explain strange numbers when MPLS gets invovled.
well spotted.
Mark might also want to try mtr rather than traceroute. It can display MPLS labels (press "u" in the curses interface or use "-e" on the command line)
Will need to sort out a small matter of removing an old kernel, and the modules. A script in the kernel package removed the modules package, before Apt/Synaptic could do. I need to clear that before I can install a package that provides mtr.
i usually use the --curses mode but the --report mode is more like traditional traceroute (and IMO more readable)
e.g.
$ mtr -e --report --report-wide download.linuxjournal.com HOST: ganesh Loss% Snt Last Avg Best Wrst StDev 1.|-- nexthop.vic.iinet.net.au 0.0% 10 14.6 14.2 13.8 14.6 0.3 2.|-- te7-2.mel-pipe-bdr1.iinet.net.au 0.0% 10 14.5 14.2 13.9 14.8 0.3 3.|-- xe-0-1-0-0.syd-ult-core1.iinet.net.au 0.0% 10 24.7 24.4 23.9 25.0 0.3 4.|-- te3-1-0.bdr1.syd7.internode.on.net 0.0% 10 23.7 23.5 22.7 24.7 0.5 5.|-- po3-0-0.bdr1.sjc2.internode.on.net 0.0% 10 205.9 205.9 205.4 206.3 0.3 6.|-- te1-7.bbr01.eq01.sjc01.networklayer.com 0.0% 10 204.8 204.2 203.7 204.8 0.4 7.|-- ae7.bbr02.eq01.sjc02.networklayer.com 0.0% 10 205.5 205.7 205.3 206.2 0.3 | +-- [MPLS: Lbl 356132 Exp 0 S 1 TTL 1] 8.|-- ae0.bbr02.cs01.lax01.networklayer.com 0.0% 10 212.5 212.8 212.4 213.2 0.3 | +-- [MPLS: Lbl 308816 Exp 0 S 1 TTL 1] 9.|-- ae7.bbr01.cs01.lax01.networklayer.com 0.0% 10 213.7 213.9 212.4 220.7 2.5 | +-- [MPLS: Lbl 567317 Exp 0 S 1 TTL 1] 10.|-- ae19.bbr01.eq01.dal03.networklayer.com 20.0% 10 241.7 241.6 241.3 242.0 0.3 11.|-- po31.dsr01.dllstx3.networklayer.com 0.0% 10 243.3 243.8 243.2 246.8 1.1 12.|-- te4-1.dsr01.dllstx2.networklayer.com 10.0% 10 241.7 243.3 241.2 254.0 4.1 | +-- [MPLS: Lbl 13666 Exp 0 S 1 TTL 1] 13.|-- te3-1.car16.dllstx6.networklayer.com 10.0% 10 244.6 244.6 244.3 245.2 0.3 14.|-- saratoga.site5.com 10.0% 10 239.8 239.9 239.6 240.5 0.3
I think Mark has mentioned problems downloading the Linux Journal before. Makes me wonder what the MTU and MRU on his PPP interface is and whether he (or Telstra) is blocking ICMP packets (in particular ICMP Fragmentation Required msgs).
The MTU and MRU were as default on install. I have now altered both to 600 in /etc/ppp/options, but will only see what effect when I next connect, probably Monday evening. Since it is a local call fee each connect, I tend to do only once a day.
Mark, try setting the MTU and MRU to some fairly low value (the default of 1500 is too big for a 56K modem connection anyway). something like 600 or 900 each perhaps. and then see if the download works.
I will try, and report.
IF this works, there'll be a bit more throughput loss to packet overhead (more and smaller packets) but a slightly slower download that works is better than one that doesn't :)
I am still looking for where there might be any proxy set up, or whether it might be something auto set up by the negotiations when I connect.
it's been far too long since i had a modem connection, i can't remember what I used to set mine at. 600 is a good test value - if that doesn't help then MRU and MTU issues are not the source of the problem and there's no point wasting time trying to find the highest value that works reliably (just set them back to default values). If it does help then gradually increase them (in increments of 50 or 100) until the download stops working again - you have now discovered a good value for MTU and MRU for your connection.
I can look up man pages, and read them, but they are not entirely meaningful at this time. I have other things in life, and a bit of comment from more experienced users does help a lot. I would appreciate whether the files listed in the pon/poff man page are sufficient, or whether I should dig further to set up dialup on Debian 6. I would also appreciate pointers as to the primary package to remove to uninstall Gnome Net Manager. I have looked with Synaptic, but not quite certain of the right candidate. The other matter is how to know whether I need chap or pap, or should set up both.
craig
Regards, Mark Trickett
Allan Duncan wrote:
http://www.internode.on.net/support/faq/tech_space/mpls_traceroute/
Thanks, that is useful to me, too.
Mark Trickett wrote:
I subscribe to Linux Journal, which as gone digital only. When I go to get the digital copy, things time out with a connection closed message when I use wget. It can be anywhere from less than 1% to somewhere below 20%, and currently not doing partial downloads to continue, having to restart from scratch.
Did you try wget -c?
participants (9)
-
Allan Duncan -
Andrew Greig -
Bob -
Craig Sanders -
Edward Savage -
Huw Davies -
Jason White -
Mark Trickett -
Trent W. Buck