Netcabo e os Packet Loss

Twikhor

Twikhor

Power Member
Boas!

Estou abrir este tópico devido aos packet loss que são cada vez mais frequentes na netcabo.
Esta situação é provavelmente exacerbada pelo aumento dos limites de download, mas a verdade é que antes das alterações dos tarifários já havia bastantes pessoas com este problema.

Eu tenho uma taxa de packet loss que varia entre os 10% e os 40% a qualquer hora do dia e em qualquer ligação que passe pelos servidores da marconi (cprm.net); situação esta que já se arrasta há mais de 2 meses sem qualquer explicação por parte da netcabo :wow:
Como devem imaginar isto condiciona, e muito, a utilização da net especialmente quando jogo em servidores internacionais (pings elevados).

Pelo que pude averiguar este problema é comum em Coimbra e segundo um user de outro fórum isto acontece devido à netcabo afunilar certas ranges de IPs no mesmo nodo o que faz com que certos pedidos nunca chegam a obter resposta daí os packet loss.

Queria portanto saber se há mais pessoas na mesma situação e se alguém tem outra explicação para estes packet loss já que a netcabo não se digna a informar os clientes...

P.S.- Deixo aqui os links dos trace routes que fiz:

http://pwp.netcabo.pt/twister1/tracert1.jpg
http://pwp.netcabo.pt/twister1/tracert2.jpg
http://pwp.netcabo.pt/twister1/tracert3.jpg
 
Última edição:
>ping -t 195.8.10.37 um dos servidores da marconi que tiveste mais perdas(tinha inicialmente o ping ao hotmail e queria por era a um ip da marconi, mas qd fiz o ping tava ja a pensar no traceroute ao hotmail e trokei me td :P por isso editei novamente o post)

A enviar para 195.8.10.37 com 32 bytes de dados:

Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=16ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=6ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=6ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=8ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=8ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=8ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=8ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=6ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=8ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=13ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=10ms TTL=249
Resposta de 195.8.10.37: bytes=32 tempo=7ms TTL=249

Estatísticas de ping para 195.8.10.37:
Pacotes: Enviados = 21, Recebidos = 21, Perdidos = 0 (perda: 0%),
Tempo aproximado de ida e volta em milissegundos:
Mínimo = 6ms, Máximo = 16ms, Média = 7ms
nem uma perda.. arredores de lisboa...
envia mail com essas imagens para a netcabo e para a marconi..
perdas perdas so comecas a ter dentro da rede da marconi.. mesmo assim envia mail para os dois lados..


>tracert www.hotmail.com

A rastrear a rota para www.hotmail.aate.nsatc.net [208.173.208.152]
até um máximo de 30 saltos:

2 7 ms 9 ms 8 ms a212-113-162-94.netcabo.pt [212.113.162.94]
3 8 ms 6 ms 7 ms ge-5-1.cr01pc01.netcabo.net [212.113.160.210]
4 13 ms 8 ms 7 ms a212-113-176-73.netcabo.pt [212.113.176.73]
5 10 ms 7 ms 11 ms lis1-br1-gi-6-2.cprm.net [195.8.10.37]
6 6 ms * 7 ms lis1-cr1-gi-14-2.cprm.net [195.8.30.5]
7 82 ms 83 ms 172 ms so-2-2-0-ecr1.mia.cw.net [195.2.6.37]
8 114 ms 122 ms 113 ms so-4-0-0-1-dcr1.was.cw.net [195.2.3.57]
9 * 143 ms 126 ms so-7-0-0-dcr1.nyk.cw.net [195.2.3.2]
10 197 ms 197 ms 193 ms so-2-0-0-dcr2.tsd.cw.net [195.2.10.110]
11 198 ms 193 ms 193 ms as0-dcr1.tsd.cw.net [195.2.10.165]
12 205 ms 200 ms 204 ms so-0-0-0-dcr2.amd.cw.net [195.2.10.145]
13 200 ms 197 ms 200 ms so-5-0-0-bcr3.amd.cw.net [195.2.10.85]
14 201 ms 203 ms 200 ms 208.173.208.152

Rastreio concluído.

nestes dois testes e a estas horas eu pelo - n tou a ter problemas
 
Última edição:
Código: 
Tracing route to 80.239.178.114 over a maximum of 30 hops

  1    <1 ms    <1 ms    <1 ms  192.168.1.3 
  2    58 ms    30 ms    30 ms  10.41.223.254 
  3   155 ms    20 ms     7 ms  a212-113-162-94.netcabo.pt [212.113.162.94] 
  4     9 ms     6 ms     6 ms  ge-5-2.cr01pc02.netcabo.net [212.113.160.214] 
  5    10 ms     9 ms     5 ms  a212-113-176-69.netcabo.pt [212.113.176.69] 
  6     9 ms     7 ms     6 ms  lis2-br1-gi-6-1.cprm.net [195.8.10.33] 
  7    38 ms     9 ms    93 ms  lis1-cr1-gi-9-0.cprm.net [195.8.0.89] 
  8    41 ms    40 ms    41 ms  lon2-cr1-po-1-0.cprm.net [195.8.0.94] 
  9    35 ms   106 ms   141 ms  ldn-th-i1-geth3-0.telia.net [195.66.226.48] 
 10    35 ms    34 ms    34 ms  ldn-bb2-pos0-2-0.telia.net [213.248.64.93] 
 11    50 ms    46 ms    48 ms  prs-bb2-pos6-0-0.telia.net [213.248.65.114] 
 12   117 ms    49 ms    70 ms  prs-b2-pos11-0.telia.net [213.248.65.226] 
 13    48 ms    51 ms    47 ms  prs-tc-ks51-geth0-1.telia.net [213.248.70.22]

Sem problemas. :)

Liga para o SAC da netcabo, diz que queres falar com alguém do Departamento de Fidelização e ameaça-los que se o problema não estiver resolvido em 5 dias úteis -> bye bye! :x2:

[[]]

PS: Já agora, deixa esse proggy e faz testes com a linha de comandos. ;)
PS2: Traceroute com P2P a 20kB/s (upstream) :D
 
Estive a fazer mais alguns traceroutes e cheguei à conclusão que o problema não se encontra na cprm.net mas sim na transição da rede netcabo para qualquer outra rede:

tracert4.jpg


xoty disse:
envia mail com essas imagens para a netcabo e para a marconi..
Clicar para expandir...

Já enviei e da marconi recebi a resposta que o problema não estava na rede cprm.net (pelo e-mail pareceu que sabiam o que se passava mas não podiam dar mais informações)
Da netcabo recebi a resposta do costume: "Na sequência do seu contacto, que desde já agradecemos, informamos que iremos contactá-lo telefonicamente no sentido de dar seguimento ao tratamento desta questão."





redalert disse:
PS: Já agora, deixa esse proggy e faz testes com a linha de comandos.
Clicar para expandir...

Com linha de comandos o resultado é o mesmo...


Cumprimentos
 
O texto está em Inglês porque foi postado no forum da Blizzard.

Ok, I spent another afternoon on the phone with a few people, and will try to compile all the info on this post.

First, my ISP is also Netcabo, and I'm having latency problems as most/everyone from that ISP. Most notably the huge (sometimes over 60sec) lag spikes in-game which make anything impossible - not even chatting, let alone playing.

As I'm sure most did, I contacted Netcabo several times since September (when the problem seemed to have started). I only use this link on weekends (Fri-Sun), during the week I have access to a Cabovisão connection (which has had no problems with latency or packet loss whatsoever). Because of this I didn't keep that constant track of the Netcabo link status (everyday basis), my detection of problems was like having fine latencies on a weekend, horrible on the next, and so on...

Nevertheless few weekends were playable, only in late November/early December the latencies came back to normal, I even thought they had finally fixed it. But coming home for vacations on the 21th December I saw that it wasn't fixed, in fact it's worse now than ever was since September (for me at least).

I contacted Netcabo again and pushed them to let me speak with someone in "Tech" (at least someone who knew more about networking than those that answer our calls), a few calls and days later I managed to do it and was asked to do a traceroute. I showed them the results and, as the problem is recorded as starting in CPRM/MID, they told me: "As you see there's no packet loss within the Netcabo network, it starts in CPRM thus we can't do anything else than report it to them". They added that they had a lot of similar complaints and that all were forwarded to CPRM.

Fortunately I have a cousin working in PT (almost retired too, so fairly good connections inside) who gave me a direct contact to an Engineer in CPRM/MID. After my first contact he asked me to e-mail all the traceroutes I've been doing so he could look into it; he's also a Netcabo client and he already noticed the poor link he's been having.

The funny but not surprising thing he told me when I said Netcabo had forwarded all the complaints to them was that he had a total of two/three contacts from Netcabo blaming them for the packet loss. He's only sad that they didn't complain in written (nor did he managed to get them to do it) as they'd be burned with proof that the problem was theirs.

After looking at the traceroutes, and after me reading these forums and reading about different theories, he told me this:

1) That latency spot we all see in Telia.level3 (two consecutive routers even), according to his personal opinion, will have little if any impact on gaming. He says it's normal for traceroute to report unusual higher latencies along the way, which decrease on the following hops - which is our case. If you look at the latencies within Telia Paris network (or do a TCP ping to the WoW login servers) they're fairly below 100ms. NOTE - sometimes we have over 250ms there and Telia/Blizzard are to blame for that, but it's not a 24/7 issue and certainly not our current problem.

2) The looking glass reports that show the reverse path going through NY are not valid, as each hop decides where to route through. This means that even though that particular hop may route through NY, some other closer to WoW is doing the correct routing (which is proved by the small latencies).

3) He did traceroutes and pings to WoW login servers and showed the same 50-80ms I was recording from my computer *BUT WITH NO PACKET LOSS*. This more than proves on its own that the problem is not the link from CPRM/MID (Portugal) -> Telia (France) While there is room for debate about some oscillating latencies, there's no point in bashing the London routers for what's happening to us.

4) As a last test, I allowed my PC to be pinged and he did the same tests towards this direction (CPRM->my PC). Guess what? Close to the very same 20% packet loss I record on my traceroutes.



*What this ALL means*

The uplink signal we have is decent, if you try to do an upload test you will most likely find yourself having close to the full bandwidth (upload ONLY).

As such, there is no packet loss on the way out, all data reaches WoW normally. If you talk to a friend in-game for a while and suddenly get a lag spike, you will notice he gets your message as soon as you press enter, but you only see it on your screen when the lag spike ends.

The problem is that on its way back, when the signal goes from CPRM->Netcabo, IT IS NOT routed through the same path/hops. One or more of those hops used on the return path have way too much traffic and lose around 20% of all the packets. THIS is the reason for our brilliant lag spikes.

Once again I contacted Netcabo (but now with this info), and they merely throwed me what some have already said: they're aware of this issue and it's scheduled to be resolved by late/end of December. I guess it means they have 4 days to fix it... (yeah right)

I have gone through many tests and theories, and only this CPRM-centered test could prove it; the link is fine from Portugal->France (explaining why Cabovisão users that go through CPRM have no problems), but horrible from CPRM->Netcabo.

Time to cross your fingers... as for me, I'll be hoping to get Clix at home ASAP, I'm fed up with these greedy ignorants.
Clicar para expandir...

O problema não está na saída, está no retorno da informação, CPRM --> Netcabo. Ou seja problema da Netcabo.
 
Inicie sessão ou registe-se para poder participar.
Back
Topo