TRACEROUTE thread

[[GOA]Erivoss]

Please remember that we need to know where you're pinging from - so blanking out the first couple of hops on your ping plot really doesn't help. Also, we need to know the time and date of the ping plots, so blanking out that bit >really< doesn't help us either.

If you don't want people here knowing what your IP is then send the ping plots via rightnow or to me directly.

 

[Bahumat]

Please remember that we need to know where you're pinging from - so blanking out the first couple of hops on your ping plot really doesn't help. Also, we need to know the time and date of the ping plots, so blanking out that bit >really< doesn't help us either.

If you don't want people here knowing what your IP is then send the ping plots via rightnow or to me directly.

as monkeys go, your one clever monkey!

*continues to examine Erivoss's avatar*

dont move much tho

 

[Gamah]

My ping plot trial time has run out, know any way to "get around it"?

 

[IainC]

My ping plot trial time has run out, know any way to "get around it"?

It will prompt you for the registration code when you open it and again when you start a trace however the 'register later' button will become active in both cases after about 5 seconds so just click that to keep using the trial.

 

[chikara]

That's a perfectly healthy pingplot, no problems at all in that one and a very low average ping.

The one that Wodzie posted does show an issue that we can pass on but the trace length is too short to draw definite conclusions from, these plots need to be at least ten minutes long otherwise they don't really help.

What´s wrong with that pingplot? average ping of 46 is bad now?

I seriously beg to differ with the both of you. I would ask you to take a look at that image again and look at the 2nd column headed PL% (percentage packet loss).

All the ping plots that I have posted or sent to right now have shown packet loss at the entry point to the GOA network. If you're going to judge response time over loss as your benchmark for an acceptable gaming experience then I am amazed.

With regards to the GOA staff, can't you ask your tech people to set up beacon servers to monitor this sort of thing? As GOA is actually owned by France Telecom, I can't image them refusing this sort of request to monitor their network for faults.

 

[IainC]

I seriously beg to differ with the both of you. I would ask you to take a look at that image again and look at the 2nd column headed PL% (percentage packet loss).

All the ping plots that I have posted or sent to right now have shown packet loss at the entry point to the GOA network. If you're going to judge response time over loss as your benchmark for an acceptable gaming experience then I am amazed.

With regards to the GOA staff, can't you ask your tech people to set up beacon servers to monitor this sort of thing? As GOA is actually owned by France Telecom, I can't image them refusing this sort of request to monitor their network for faults.

That packet loss isn't an error though. All it shows are routers which are configured not to respond to ping requests which is common for busy routers, the fact that you can reach servers behind those routers shows clearly that you aren't actually getting 100% packet loss, you just aren't getting any answer from those hops in your path.

The problem isn't one of hubs misbehaving usually, we can monitor every hub in Europe and we wouldn't see the problems that people are having. Any infrastructure provider will also monitor their hubs very closely and hubs that break down are quickly fixed or have traffic routed around them. The issue is usually with routing from one hub to another and that's not something that we can see from where we are. If we ran a trace back to your ISP backbone from our platform, we'd go through a different path to the one that you take in the opposite direction. This would of course tell us nothing about the problems that you may be experiencing.

 

[chikara]

That packet loss isn't an error though. All it shows are routers which are configured not to respond to ping requests which is common for busy routers, the fact that you can reach servers behind those routers shows clearly that you aren't actually getting 100% packet loss, you just aren't getting any answer from those hops in your path.

The problem isn't one of hubs misbehaving usually, we can monitor every hub in Europe and we wouldn't see the problems that people are having. Any infrastructure provider will also monitor their hubs very closely and hubs that break down are quickly fixed or have traffic routed around them. The issue is usually with routing from one hub to another and that's not something that we can see from where we are. If we ran a trace back to your ISP backbone from our platform, we'd go through a different path to the one that you take in the opposite direction. This would of course tell us nothing about the problems that you may be experiencing.

Yes I realise that the hops with 100% packet loss are the Opentransit routers that do not respond to ping requests. I was referring more to the shocking 27% packet loss when I was hitting glastonbury (hop 19 on the image), and 18% on the hop before.

I have used my MS paint skills to highlight this:

I'm not suggesting that you monitor the whole of europe, I'm suggesting you monitor your internet network and your link onto the Opentransit backbone.

Every time I start getting that magic LD feeling in game, and have started ping plotter up, all fingers unfortunately start to point to your platform. Sure people are going to go LD because of problems on their own ISPs networks and what not, and granted you are going to defend your own set up, and I don't blame you for doing so. After all, the majority of the time things are working ok. But when people start posting evidence of high pings and packet loss (especially the packet loss) which are only beginning on the hop after your link to Opentransit, you have to start asking what you folks can do to sort out the problem. When people post plots after a mass LD, it's all well and good to blame some central european server, coz Joe Bloggs happens to have packet loss on his ISPs servers too. But for every person that has that, there is also another person that is having no lag before they hit the GOA network.

I'm willing to bet you a month's subscription that you have a problem with 1 of the 2 routers that links your network to the opentransit backbone. A beacon server on the same network as your gaming servers, talking to one on the Opentransit network would enable you to tell me right here and now whether that's the case or not. TBH it's quite shameful that instead of being able to do this, GOA has to ask for people to send ping plot after ping plot. This is something that should be happening automatically. 11 pages of this thread alone, plus all those other thread that have appeared about the subject, should be enough to say that you folks need something like this set up.

I'll leave you with a repost of the rest of the plots that I've posted on FH. You can check the dates and times yourself. Although many of them are days / weeks old by now, they all happened during periods of mass LDs if I recall, and should therefore be enough for you to realise you do actually have a problem. Packet loss is your biggest enemy. Deny it if you want, but then it just turns into a PR exercise.

High latency makes things sluggish but can still be playable. Dropping the packets is what's leading to zero response for large periods / LD.

/me shrugs

 

[glabbin]

[URL=http://suprfile.com/get.php?id=7bmk90] [/URL]

hm, cant connect

 

[Gahn]

4s delay now and then: