Forums » General
NW lag observation, an invitation to the devs.
The Nation War regularly sees the biggest number of players in a single sector than any other event or general activity in VO. 40 players is usual for the sunday NW. It is perhaps unique in that all the sector activity is player generated ; there are no npcs involved.
Various players , including myself, have noticed that the problems of lag, missing hits and jumpiness have become worse in recent months to the extent that the first five minutes of the war can be extremely frustrating.
If vo becomes more populated, as we all hope it will, this lagginess in sectors where there are a large number of players will become more of general rather than a specific NW problem, and may detract from the game for many .
I appreciate that the devs are busy , but the NW provides a unique opportunity to see these problems first hand. I would be more than happy to see you come to a Nation war , observe or join in, and use clever " dev tools " to try and determine what is happening and how it can be alleviated.
The idea of using the NW as a platform for testing this is simple; it is something which could not be observed anywhere else in VO at the moment. I know that a number of other players are as concerned about this issue as I am, so an open invitation, please feel that you are welcome to use the NW in whatever way you consider is of use .
Perhaps this should be in bugs or suggestions, but I do feel I had to get the debate started.
Ecka
Various players , including myself, have noticed that the problems of lag, missing hits and jumpiness have become worse in recent months to the extent that the first five minutes of the war can be extremely frustrating.
If vo becomes more populated, as we all hope it will, this lagginess in sectors where there are a large number of players will become more of general rather than a specific NW problem, and may detract from the game for many .
I appreciate that the devs are busy , but the NW provides a unique opportunity to see these problems first hand. I would be more than happy to see you come to a Nation war , observe or join in, and use clever " dev tools " to try and determine what is happening and how it can be alleviated.
The idea of using the NW as a platform for testing this is simple; it is something which could not be observed anywhere else in VO at the moment. I know that a number of other players are as concerned about this issue as I am, so an open invitation, please feel that you are welcome to use the NW in whatever way you consider is of use .
Perhaps this should be in bugs or suggestions, but I do feel I had to get the debate started.
Ecka
The largest Nation War I ran had near 60 people in it. Bad in and of itself. But most people had rockets, which made it very very bad.
I would also mention that this doesn't even factor in the problem that Swarm Missiles are already out because of the lag issues. We might start to see a more limited arsenal if VO & the NW grow.
Lagging often seems related to ping. Perhaps a new ping tool could help. It would return a list of every player in the sector and their ping. Players with excessively high pings could be excluded. This might also help regulate any players purpously manipulating their ping to gain advantage.
Too easy, right Lecter? ;)
The trouble is defining what kind of "lag" you guys are talking about. Rockets, for instance, should have little impact on your network performance or the server, but might slow people down graphically from all the trails.
I'm fine with stopping by at some point, although we do keep continuous graphs of the game's performance (individual cluster server CPU usage, network traffic, core server CPU usage, so on). I don't see anything that especially stands out for Sunday. Our bandwidth usage was well within limits, the cluster server load was characteristically spiky without the hallmarks of any of the problems that occasionally crop up. The core server load was negligible. But, the graphs don't show everything.
What specific time was the NW? I can guess from the peak user graphs, but it would be better to know (I'm sure I can check a newspost or some such, just making sure).
What were people's individual pings at the time when things became bad? The reports of missing hits and jumpiness are the hallmarks of a bandwidth or latency problem (pingtime). The server itself was doing fine, in terms of bandwidth. Obviously, if there was an internet weather issue that happened to coincide, there's little we could have done about that. We can start graphing ping response times to certain remote hosts, but that can be sort of a difficult thing to measure (routers, for instance, will de-prioritize direct ICMP responses when under load, giving inaccurate latency spikes while the ping-through times remain normal.
If the game is becoming slow on the graphical client side, that's a whole different issue. In that case, I can look at building a reduced-detail version of missile trails or some such, which are configurable under detail settings.
There are a few things we can look at if we're there at the time, but honestly, the best analysis would be provided by you guys, monitoring your framerates and ping times using the features built into the client. We can only see what the server is seeing, we have no way of knowing if the game is "slow" on your computer, for instance.
I'm fine with stopping by at some point, although we do keep continuous graphs of the game's performance (individual cluster server CPU usage, network traffic, core server CPU usage, so on). I don't see anything that especially stands out for Sunday. Our bandwidth usage was well within limits, the cluster server load was characteristically spiky without the hallmarks of any of the problems that occasionally crop up. The core server load was negligible. But, the graphs don't show everything.
What specific time was the NW? I can guess from the peak user graphs, but it would be better to know (I'm sure I can check a newspost or some such, just making sure).
What were people's individual pings at the time when things became bad? The reports of missing hits and jumpiness are the hallmarks of a bandwidth or latency problem (pingtime). The server itself was doing fine, in terms of bandwidth. Obviously, if there was an internet weather issue that happened to coincide, there's little we could have done about that. We can start graphing ping response times to certain remote hosts, but that can be sort of a difficult thing to measure (routers, for instance, will de-prioritize direct ICMP responses when under load, giving inaccurate latency spikes while the ping-through times remain normal.
If the game is becoming slow on the graphical client side, that's a whole different issue. In that case, I can look at building a reduced-detail version of missile trails or some such, which are configurable under detail settings.
There are a few things we can look at if we're there at the time, but honestly, the best analysis would be provided by you guys, monitoring your framerates and ping times using the features built into the client. We can only see what the server is seeing, we have no way of knowing if the game is "slow" on your computer, for instance.
Could there be a way of recording that? At least for the moment, I think it would helpful just to confirm wether this was a graphical or networking problem.
Nation War was at 20:00 GMT on Sunday. 3PM CDT, if I did my timezone math correctly.
Zamzx (and thread), from what I've seen there is the inherent problem of latency just from so many people in the sector, causing the missed shots etc, but when weapons such as swarms go off (I've personally never seen a slowdown from just rockets) there is a graphical slowdown as well.
In short...both? (probably)
But I think the network issues are the more pressing concern. And as Ecka said this has been a repeated problem so I would be skeptical if it came from weather/outside circumstances.
In short...both? (probably)
But I think the network issues are the more pressing concern. And as Ecka said this has been a repeated problem so I would be skeptical if it came from weather/outside circumstances.
Well, there's no reason for the devs not to cover all of their bases. We already know that they can monitor all of (their) networking activity at their end. I'm suggesting that they look into a way of recording a players /ping and /fps or any other relevant information. We have a small enough community now that if the devs were to ship out an optional mod that recorded that for the players, we should be able to get pretty much everyone in the NW to install it and turn over the data.
People are welcome to leave a continuous "ping" going in the background, in another window, while they run VO. For instance, in Windows you would do something like:
ping -t www.vendetta-online.com
From a command prompt. MacOS you would remove the "-t", Linux would be the same. Using our webserver is fine, as they're all on the same link in the same server room. When you abort the ping (ctrl-c), it will display the min/max/average over the period when you were measuring. If you want additional data, you could redirect the output to a file, a-la:
ping -t www.vendetta-online.com > ping.log
Aside from this, you can also use "/toggleframerate" in the game, as well as "/toggleframerategraph" (I think that still works?) to get an idea of how fast the game is running on your machine.
There is no known "inherent problem of latency just from so many people in the sector": bandwidth over-saturation would cause this and we have yet to see if this is the case (at least at our end, on the client end it's very possible, especially for modem users or those in distant locations). There is, however, a *lot* of misunderstanding about what causes these kinds of issues. The best advice I can give, for the moment, is to gather as much information as possible from as many affected areas as possible. For instance, everyone having a ping going in the background, everyone watching their framerate, taking note of what has changed if the situation improves. Ie, the original complaint said it was annoying for the "first five minutes", what happens at that point.. are there just less ships around? Or have people respawned in replacement ships, and are present but just not shooting and flying around? These indicate different kinds of issues.
One of us will try to stop by on Sunday, to see what we can. We also have pre-existing tickets for creation of more game and network monitoring, but I can't promise anything before the next NW. I also have an email in to our ISP, just to see if they're aware of any issues. But for the moment, that's the best I can do.
If this is a real problem at our end, we'll definitely try to sort it out. To figure that out, though, we'll need a lot of debugging help from all of you, during the situation in question.
ping -t www.vendetta-online.com
From a command prompt. MacOS you would remove the "-t", Linux would be the same. Using our webserver is fine, as they're all on the same link in the same server room. When you abort the ping (ctrl-c), it will display the min/max/average over the period when you were measuring. If you want additional data, you could redirect the output to a file, a-la:
ping -t www.vendetta-online.com > ping.log
Aside from this, you can also use "/toggleframerate" in the game, as well as "/toggleframerategraph" (I think that still works?) to get an idea of how fast the game is running on your machine.
There is no known "inherent problem of latency just from so many people in the sector": bandwidth over-saturation would cause this and we have yet to see if this is the case (at least at our end, on the client end it's very possible, especially for modem users or those in distant locations). There is, however, a *lot* of misunderstanding about what causes these kinds of issues. The best advice I can give, for the moment, is to gather as much information as possible from as many affected areas as possible. For instance, everyone having a ping going in the background, everyone watching their framerate, taking note of what has changed if the situation improves. Ie, the original complaint said it was annoying for the "first five minutes", what happens at that point.. are there just less ships around? Or have people respawned in replacement ships, and are present but just not shooting and flying around? These indicate different kinds of issues.
One of us will try to stop by on Sunday, to see what we can. We also have pre-existing tickets for creation of more game and network monitoring, but I can't promise anything before the next NW. I also have an email in to our ISP, just to see if they're aware of any issues. But for the moment, that's the best I can do.
If this is a real problem at our end, we'll definitely try to sort it out. To figure that out, though, we'll need a lot of debugging help from all of you, during the situation in question.
The only times I ever experience shots not registering is during BS or NW or other occasions with a lot of ships in the sector (which is admittedly rare =P) The first thing I do when I encounter this is check my ping which usually resides around 70-80 and remains there even when my shots stop registering. With this in mind, I've never seen a correlation between my ping and shots not registering. However, when the problem does occur I've always been in some sector with a large amount of ships, NPC or PC. Hope this helps. Like Ecka said, the problem is not very apparent right now, but if the game ever expands, it might become a huge problem.
The thing is.. it's relatively explicable with BS, due to the complexity of the interacting capital ships and the relative "weight" of all those NPCs firing on each other. This slows down the sector and makes it stop registering hits. We've put in a lot of work to improve that case.
Pure PvP, on the other hand, doesn't have most of this going on. The sector does "sanity check" the impacts between players, to prevent cheating, but this should be a hell of a lot lighter weight than anything BS-like. So, the correlation isn't very clear to me.
It is, however, possible that you're seeing two completely different issues that present the same symptoms. In BS, you see shots not registering simply because the sector is overloaded and is unable to see that your shot as landed. The NW, you see shots not registering because the remote player's bandwidth is overloaded. Ie, the same problem we have with people "cheating" by saturating their own bandwidth. We do have a fix planned for this, but I'm not sure exactly when it'll appear.
At any rate, the trouble for us is the.. wide disparities between the situations, rather than what they have in common. Anyway, continued feedback is welcome.
Pure PvP, on the other hand, doesn't have most of this going on. The sector does "sanity check" the impacts between players, to prevent cheating, but this should be a hell of a lot lighter weight than anything BS-like. So, the correlation isn't very clear to me.
It is, however, possible that you're seeing two completely different issues that present the same symptoms. In BS, you see shots not registering simply because the sector is overloaded and is unable to see that your shot as landed. The NW, you see shots not registering because the remote player's bandwidth is overloaded. Ie, the same problem we have with people "cheating" by saturating their own bandwidth. We do have a fix planned for this, but I'm not sure exactly when it'll appear.
At any rate, the trouble for us is the.. wide disparities between the situations, rather than what they have in common. Anyway, continued feedback is welcome.
Ok, Thanks for the feedback all round . Let's observe and record , see what we can turn up between us .
One more thing to put out there: WiFi is notorious for sudden latency spikes, packet loss and other difficult-to-track-down problems (say, your neighbor turns on their microwave, or starts using their cordless phone or something). So please measure from a wired connection. Wireless adds way to much weirdness into the equation.
Preferably a real ethernet wired connection. USB connections can also start to suck as CPU usage increases.
Preferably a real ethernet wired connection. USB connections can also start to suck as CPU usage increases.
I could record the nation war like I did for that RP thread of mine. I'll need to bind a key for checking my ping every so often, though, because I'm too competitive to take timeouts to chat. ;)
Incarnate: what do you think about creating a ping tool that records the ping of everyone in the sector?
I think that this is the best way to discover if latancy is an issue. If individual users are purposly saturating their own bandwidth, they will not report their own higher ping times. We need a way to see another players ping. This could be useful in every PVP encounter. A player could decide to engage based not just on player stats, but ping also.
I think that this is the best way to discover if latancy is an issue. If individual users are purposly saturating their own bandwidth, they will not report their own higher ping times. We need a way to see another players ping. This could be useful in every PVP encounter. A player could decide to engage based not just on player stats, but ping also.
Here's how the devs and the VO community could work together to analyze the ping question.
1. Devs add an argument to the /ping command:
/ping playername
2. Drazed adds the new ping info as an option to targetls.
3. Another lua guru codes a ping recording tool for use in NW.
1. Devs add an argument to the /ping command:
/ping playername
2. Drazed adds the new ping info as an option to targetls.
3. Another lua guru codes a ping recording tool for use in NW.
I've always wanted to be able to see another players ping in-relation to the server. Some folk might argue that this makes VO even more FPS'ish, but I really do not think it would. The ping readout should be placed on the +toplist, and would only need to be updated like once every minute.
Well, I've only hit lag once during a NW, but it was quite major, roughly 15 seconds where there was no information coming in at all- I believe it was most likely a node on the path between my comp and the VO servers shutting down and forcing a reroute.
As for general performance, I have not seen anything unexpected happening. NW combat proceeds smoothly for me. True, some distant players (2000m+) will get a bit jumpy, but this always happens at long range. If I recall correctly, this is actually how it's supposed to work- prioritize location and status updates for closer players, and let those that are too far off to have a realistic chance of engaging be updated less frequently.
I'll do the recommended ping/FPS monitoring during the next NW. I run VO on the 'low' end of the graphical options, and have the background turned off entirely for NW, because I am very familiar with system lag, and how to counter such preemptively.
And since nobody actually answered Incarnate's question of what goes on in the first five minutes of a NW- All sides lose roughly one half to two thirds of their forces in a hail of rockets and energy shots, resulting in a dramatic drop in the sector population. Returning players stay in buses, at a distance, usually holding still with occasional rotation to keep the battle in view. Surviving players usually run out of, or at least conserve, their last shots of ammo, so fighting moves almost exclusively into energy weapons. This progression would remove causes for both network lag and system lag, so it's not very clear to me which it would be.
As for general performance, I have not seen anything unexpected happening. NW combat proceeds smoothly for me. True, some distant players (2000m+) will get a bit jumpy, but this always happens at long range. If I recall correctly, this is actually how it's supposed to work- prioritize location and status updates for closer players, and let those that are too far off to have a realistic chance of engaging be updated less frequently.
I'll do the recommended ping/FPS monitoring during the next NW. I run VO on the 'low' end of the graphical options, and have the background turned off entirely for NW, because I am very familiar with system lag, and how to counter such preemptively.
And since nobody actually answered Incarnate's question of what goes on in the first five minutes of a NW- All sides lose roughly one half to two thirds of their forces in a hail of rockets and energy shots, resulting in a dramatic drop in the sector population. Returning players stay in buses, at a distance, usually holding still with occasional rotation to keep the battle in view. Surviving players usually run out of, or at least conserve, their last shots of ammo, so fighting moves almost exclusively into energy weapons. This progression would remove causes for both network lag and system lag, so it's not very clear to me which it would be.