Game is way too picky when it comes to a single packet loss

[Claviusnex]

Really need a sniffer to dig into why the packet loss. Also need to know the ISP static routes and their dynamic routing algorithm/tables. It sounds like the link to the NY node might be the ISP default route. It is possible that this is the ISP's largest pipe to the backbone provider so while the local Chicago node is closer to your location it could actually have less capacity than your current route. Not saying this is good or acceptable just reality. One possible cause of packet loss is the dynamic loss of routes. These eventually get discovered by the network and the new routing tables get propagated. Normally this occurs fast enough that the TCP communication stack will recover on a retry but sometimes it occurs slowly enough that connections start timing out. The longer the route the more opportunity for routes to change. Again this is just hypothesizing and a sniffer is what's needed to see what is really going on.

[Jijifli]

snip

In my case, it was my end, and I knew it was my end (Apparently I overheated my driver for it.) That doesn't just mean there's nothing that needs to get fixed when it does happen. If I get a packet loss, sure, the game obviously shouldn't take my inputs because it can't get my data, but that's the thing. It shouldn't, but it does, and incorrectly.

I HAD to stop leveling scholar for some time over it. If I used Aetherflow right when my loss happened, I'd be given 0 stacks, but still take the 60 second cooldown. Scholar needs that aetherflow for so many skills, and just because I missed 1 tick it extra screws me? If it just made it so I had to wait until it fixed itself, I'd be cool with it, but the system messes up cooldowns so badly...

[Claviusnex]

That doesn't just mean there's nothing that needs to get fixed when it does happen.

I didn't say that there isn't anything to fix but unless you know what the underlying events are you don't have any way to know where to look to resolve the problem. Traceroute and ping are fine for basic network diagnosis but they do not necessarily take the same routes that the application will. They also don't use the TCP stack but instead use ICMP or UDP neither of which are reliable transport so where they show packet loss a TCP based application might have recovered on a retransmission. Again the only thing that will show what networking protocols are being used and what is really going between the game server and client is a sniffer. From this you can then define the problem and whose responsibility it is to resolve.

I was also trying to explain that physical proximity does not guarantee a faster route. A small capacity connection to a local node might have a faster ping time but actually be slower with real data than a larger pipe to a remote node. This by itself is not good or bad. Most likely is is based on ISP contract costs with the backbone provider so even if it could be improved you would have to build a business case to justify the increased costs to the SIP. Of course it could be something stupid such as "We've always done it this way".

[Snow_Princess]

Really need a sniffer to dig into why the packet loss. Also need to know the ISP static routes and their dynamic routing algorithm/tables. It sounds like the link to the NY node might be the ISP default route. It is possible that this is the ISP's largest pipe to the backbone provider so while the local Chicago node is closer to your location it could actually haveless capacity than your current route. Not saying this is good or acceptable just reality. One possible cause of packet loss is the dynamic loss of routes. These eventually get discovered by the network and the new routing tables get propagated. Normally this occurs fast enough that the TCP communication stack will recover on a retry but sometimes it occurs slowly enough that connections start timing out. The longer the route the more opportunity for routes to change. Again this is just hypothesizing and a sniffer is what's needed to see what is really going on.

I am not asking why I have packet loss, my ISP is being cheap, the vpn fixes that because ya they prob do it what is more convenient for them, not what is better performance. This is simply showing I had a single packet loss, proven by the CMD, when I mounted and the cast failed. The only purpose of this thread is to show why they need to change the coding and stop over burning the servers unnecessary that is stopping us from having glam log because of the obsessive server confirmation checking that does not do anything positive and all negative.

I didn't say that there isn't anything to fix but unless you know what the underlying events are you don't have any way to know where to look to resolve the problem. Traceroute and ping are fine for basic network diagnosis but they do not necessarily take the same routes that the application will. They also don't use the TCP stack but instead use ICMP or UDP neither of which are reliable transport so where they show packet loss a TCP based application might have recovered on a retransmission. Again the only thing that will show what networking protocols are being used and what is really going between the game server and client is a sniffer. From this you can then define the problem and whose responsibility it is to resolve.

I was also trying to explain that physical proximity does not guarantee a faster route. A small capacity connection to a local node might have a faster ping time but actually be slower with real data than a larger pipe to a remote node. This by itself is not good or bad. Most likely is is based on ISP contract costs with the backbone provider so even if it could be improved you would have to build a business case to justify the increased costs to the SIP. Of course it could be something stupid such as "We've always done it this way".

Has nothing to do with the reasoning why a single packet loss happen, what is being discussed is how the game handles it. WoW or FFXI remain to function for a bit and when it gets streamlined again it recovers, here it just lets you lose your abilities or disconnects you from the game unnecessarily and have your character logged on too long after a D/C.