Page 1 of 2 12 LastLast
Results 1 to 20 of 29
  1. #1
    Community Member
    Join Date
    Sep 2013
    Posts
    350

    Default Daily server reboots

    Since they are unable to fix the lag, a band aid solution is daily reboot.

    Clears connections.
    Clears ghost LFMs.

  2. #2
    Community Member Flavilandile's Avatar
    Join Date
    Aug 2010
    Location
    France
    Posts
    3,908

    Default

    A server reboot is a one hour and a half downtime.

    I'm sure Turbine can do it between 8PM and 9:30 PM every day, Boston Time... Every day.

    And it will not solve lag...
    Since most of the lag is caused by either your own computer ( though I expect you know that already )
    or the network between the computer and Turbine's server since they go through Cogent as a Carrier :

    http://forums.na.leagueoflegends.com....php?t=3688185
    http://sto-forum.perfectworld.com/sh....php?t=1280131
    http://forum.square-enix.com/ffxiv/t...lease-research
    http://www.dslreports.com/forum/r288...vice-requested
    http://blues.wowraider.net/blue/us2/...th-pictures%29

    The only case where Turbine might be the source of lag is when a whole server lags out at the same time.
    ( and even in that case, PNAP might be the guilty party, since they have a few Single Points of Failure in their Boston Datacenter for all Turbine's stuff. )
    On G-Land : Flavilandile, Blacklock, Yaelle, Millishande, Larilandile, Gildalinde, Tenalafel, and many other...

  3. #3
    Community Member
    Join Date
    Apr 2013
    Location
    Athens
    Posts
    2,105

    Default

    Quote Originally Posted by Flavilandile View Post
    A server reboot is a one hour and a half downtime.

    I'm sure Turbine can do it between 8PM and 9:30 PM every day, Boston Time... Every day.

    And it will not solve lag...
    Since most of the lag is caused by either your own computer ( though I expect you know that already )
    or the network between the computer and Turbine's server since they go through Cogent as a Carrier :

    http://forums.na.leagueoflegends.com....php?t=3688185
    http://sto-forum.perfectworld.com/sh....php?t=1280131
    http://forum.square-enix.com/ffxiv/t...lease-research
    http://www.dslreports.com/forum/r288...vice-requested
    http://blues.wowraider.net/blue/us2/...th-pictures%29

    The only case where Turbine might be the source of lag is when a whole server lags out at the same time.
    ( and even in that case, PNAP might be the guilty party, since they have a few Single Points of Failure in their Boston Datacenter for all Turbine's stuff. )
    Hi, while what you say is true and lag can indeed be caused by my computer (slow processing of graphics/data) or my connection (packet loss/high latency), those are not the only sources of lag.

    For example, a DDoS attack generates lag. But there are many types of DDoS attacks:
    - An attack that uses the attackers' superior bandwidth to flood the single point failure of the server's connection. That means that the server's incoming network is down and also that the hardware can be fried if it's not properly configured. Solution against this is to simply have greater bandwidth than the attackers (and that's where DDoS protection companies come in).
    - An attack that uses the server's large responses to flood the outgoing network with low incoming traffic. Again the network is down. But this time you can use clever algorithms to determine that this is an attack and send no responses to the attackers' messages.
    - An attack that uses the server's memory usage per connection to cause memory crash. Again, with clever memory allocation and similar algorithms to detect attacks, you can defend yourself against this kind of attack
    - An attack that uses the server's computational complexity per connection to cause the server to have to process more than it can handle with its processors. Again, with smart connection algorithms that require minimal computations per connection, you can avoid this kind of attack.

    However, nothing can save you from a bug in your code. If for some reason you are making unneeded calculations per object in your world, then you'll find yourself in condition 4 where your processors can't handle that much computations. Now, if you have some sort of time-stamping in your computations and declare that "this will occur between X and Y seconds. If you receive this computation in the time Y+2, don't bother calculating it, it's too old" then you'll find yourself with lag in the terms of "calculations are being dropped" and bugs such as:

    Quote Originally Posted by MoonlitSilver View Post
    It's not all bad, though! You know that bug where lagging out in an aoe dot like firewall can make it stick to you until you log out? Today that happened to the dragon, causing him to pick up a permanent firewall debuff from one of the Zhentarim wizards.

    Anyone care to explain how a bad connection on my part (which somehow only affects DDO and not anywhere else I visit on the internet) causes a spell cast by a mob to cause a lag-related bug on another mob?
    Cheers.

    EDIT: Also, I would like to point out that my connection and packet-loss as the links you provide state is in the green with 160ms latency as expected since a traceroute to Turbine servers has the same latency. All this while I can't move, nor cast any spells or do anything else in the game besides chat and voice chat (which are working perfectly fine)

    EDIT2: I noticed you are in Ghallanda. Try playing in Khyber (or Argonessen?) for a bit. You'll see what we mean and realize it's not just the usual lag here and there. If I was playing in Sarlona for example, I would be saying as well "pffft, the usual complaints about lag" because Sarlona seems to have no issues.
    Last edited by Faltout; 12-29-2014 at 07:09 AM.
    My main server is Khyber. Have toons in almost every server for favor purposes. The Faltouts

  4. #4
    Community Member moomooprincess's Avatar
    Join Date
    Mar 2008
    Posts
    3,981

    Default

    Do they always go through Cogent? I thought Turbine only switched to Cogent once the MCI Worldcom(dba Verizon) connection went wonky.

    In my grey matter, it seemed the Cogent was the good one but the Datacenter always switches back to the Verizon one after the problems have been solved. Obviously, Cogent must charge more money to be run through them than Verizon does.
    Recipemaker Guild: Top Chef School of Recipes

    Event Statistics: Risia, Festivult, Midwinter, Daily Dice, Mimic Hunt

  5. #5
    2015 DDO Players Council Seikojin's Avatar
    Join Date
    Feb 2006
    Location
    gamertown usa
    Posts
    7,435

    Default

    Just gonna say it until everyone learns it. 90% of lag in all mmos is due to timeout handles dropping data and the server having to re-perform the request.

  6. #6
    Community Member Flavilandile's Avatar
    Join Date
    Aug 2010
    Location
    France
    Posts
    3,908

    Default

    Quote Originally Posted by moomooprincess View Post
    Do they always go through Cogent? I thought Turbine only switched to Cogent once the MCI Worldcom(dba Verizon) connection went wonky.

    In my grey matter, it seemed the Cogent was the good one but the Datacenter always switches back to the Verizon one after the problems have been solved. Obviously, Cogent must charge more money to be run through them than Verizon does.
    Yes they do... Except if their BGP peers are down through Cogent, then they will go though a secondary ( maybe Worldcom ) Carrier.
    But as a rule, when you select a carrier ( note : I know the selected Carrier is Cogent, though I don't know if it's a Turbine choice or if it's a PNAP choice ), you put a low weight to the BGP peers towards that carrier in the routing tables, so that traffic will always go that way... unless the neighbors are down.


    Quote Originally Posted by Faltout View Post
    Hi, while what you say is true and lag can indeed be caused by my computer (slow processing of graphics/data) or my connection (packet loss/high latency), those are not the only sources of lag.

    For example, a DDoS attack generates lag. But there are many types of DDoS attacks:
    - An attack that uses the attackers' superior bandwidth to flood the single point failure of the server's connection. That means that the server's incoming network is down and also that the hardware can be fried if it's not properly configured. Solution against this is to simply have greater bandwidth than the attackers (and that's where DDoS protection companies come in).
    - An attack that uses the server's large responses to flood the outgoing network with low incoming traffic. Again the network is down. But this time you can use clever algorithms to determine that this is an attack and send no responses to the attackers' messages.
    - An attack that uses the server's memory usage per connection to cause memory crash. Again, with clever memory allocation and similar algorithms to detect attacks, you can defend yourself against this kind of attack
    - An attack that uses the server's computational complexity per connection to cause the server to have to process more than it can handle with its processors. Again, with smart connection algorithms that require minimal computations per connection, you can avoid this kind of attack.
    Sadly for you I'm not even talking of attacks... They are another matter completely different of normal network operations.

    Cogent is well known for having issues as a Carrier under normal conditions. ( as all the links I put shows )...
    I work in that... A single card on a single router can just puts lots of players through lag hell until the card gets restarted ( or the router )... And nobody ( not even the telecom porator that owns the router ) will have a clue before they get in tough with the right person.

    To put it another way :
    I got called once for a problem ( mobile network ( Packet ) gone wonky on a part of the country ) at 6 AM... The operator and lots of other people were working on the issue since noon the previous day. I did a quick check up of the router involved, and noticed nothing wrong... Then I made the right printout... It showed something wonky ( no other term, it wasn't a failure ), I recommended a card restart.
    Within minutes of that card restart all the problems they had went away. If somebody playing DDO was on that card, they would have had lag, rubberbanding, disconnects, and so forth... Until the card restart.
    On G-Land : Flavilandile, Blacklock, Yaelle, Millishande, Larilandile, Gildalinde, Tenalafel, and many other...

  7. #7
    Community Member
    Join Date
    Sep 2009
    Posts
    62

    Default Something has changed over the past week....can someone look at it?

    The lag is double if not worse the past week to the point of total stand still, then wipe. It acts the same in public areas as well. It is server wide on Kyber for sure. Bank, bags, vendors all hang at times for long periods. Don't write it off as normal lag, been around too long and I know how much lag I am paying for. This is the give me a refund amt.

  8. #8
    Community Member
    Join Date
    Sep 2013
    Posts
    350

    Default

    Quote Originally Posted by Flavilandile View Post
    A server reboot is a one hour and a half downtime.

    I'm sure Turbine can do it between 8PM and 9:30 PM every day, Boston Time... Every day.

    And it will not solve lag...
    Since most of the lag is caused by either your own computer ( though I expect you know that already )
    or the network between the computer and Turbine's server since they go through Cogent as a Carrier :

    http://forums.na.leagueoflegends.com....php?t=3688185
    http://sto-forum.perfectworld.com/sh....php?t=1280131
    http://forum.square-enix.com/ffxiv/t...lease-research
    http://www.dslreports.com/forum/r288...vice-requested
    http://blues.wowraider.net/blue/us2/...th-pictures%29

    The only case where Turbine might be the source of lag is when a whole server lags out at the same time.
    ( and even in that case, PNAP might be the guilty party, since they have a few Single Points of Failure in their Boston Datacenter for all Turbine's stuff. )
    Odd I did not realize everyone in the raid group are also using my laggy computer. Are people on different instances on the server also use my laggy computer? With zero packet loss and 120 ms latency?

    If it's their ISP issue then they need to get better SLA or get another provider.

  9. #9
    Community Member
    Join Date
    Apr 2013
    Location
    Athens
    Posts
    2,105

    Default

    Quote Originally Posted by Flavilandile View Post

    Sadly for you I'm not even talking of attacks... They are another matter completely different of normal network operations.

    Cogent is well known for having issues as a Carrier under normal conditions. ( as all the links I put shows )...
    I work in that... A single card on a single router can just puts lots of players through lag hell until the card gets restarted ( or the router )... And nobody ( not even the telecom porator that owns the router ) will have a clue before they get in tough with the right person.

    To put it another way :
    I got called once for a problem ( mobile network ( Packet ) gone wonky on a part of the country ) at 6 AM... The operator and lots of other people were working on the issue since noon the previous day. I did a quick check up of the router involved, and noticed nothing wrong... Then I made the right printout... It showed something wonky ( no other term, it wasn't a failure ), I recommended a card restart.
    Within minutes of that card restart all the problems they had went away. If somebody playing DDO was on that card, they would have had lag, rubberbanding, disconnects, and so forth... Until the card restart.
    Wow, totally not what I wrote. Replying in order:
    - The attacks were an example (stated in the first sentence) of lag produced not by connections, but by holes in the system design. And then the next thing I said, was not about attacks (you didn't quote that part, I assume you did not read it?), but suggested that the code of the game was creating the lag also backed up by an example where no connections were involved. A monster having a buggy effect created from lag. This event doesn't need to get out of the server.

    - Wow, I guess since most of my connections to US go through Cogent I shouldn't be having internet at all let alone play flawlessly in the Sarlona game world (Now you're gonna suggest that Khyber goes through Cogent but Sarlona goes through.... what? Tata communications?)

    - A single card on a single router failing will cause the end of the world... *** is this? Are we in the 20th century? Today, abstraction in network systems has reached a whole new level. If your card in your home fails, you're done. But you're talking about a major organization that will spend money to ensure this thing does not happen. We have load balancing policies to make sure that there is not a single point of failure or traffic. We also have smart algorithms to detect such problems and automatically redirect traffic.

    - Did you miss the part where we said that our connections are showing no packet loss?

    -- something personal: In the example above, you didn't even find the cause of this "something wonky"? You just recommended a card restart? Wow, the guys working at the card from noon must have been really stupid to not think of that one earlier. I wonder what you would have done if the restart hadn't worked.
    My main server is Khyber. Have toons in almost every server for favor purposes. The Faltouts

  10. #10
    Community Member Flavilandile's Avatar
    Join Date
    Aug 2010
    Location
    France
    Posts
    3,908

    Default

    Quote Originally Posted by Fallout_Zero View Post
    Odd I did not realize everyone in the raid group are also using my laggy computer. Are people on different instances on the server also use my laggy computer? With zero packet loss and 120 ms latency?

    If it's their ISP issue then they need to get better SLA or get another provider.
    And I didn't talk about your computer specifically... your computer is just one potential source of lag...
    Now if you want I can give you exactly how a given computer in a party can lag out the whole party... that's called the Disconnect Lag.
    And that's caused by packet retransmission ( UDP ones ) from the server to the DCing computer... It locks down all the other people in the party, because
    they can't get a location update from that DCing player.

    Now Since you seem to have about zero clue about network complexity and have a head full of merchandizing terms...

    Quote Originally Posted by Faltout View Post
    - Wow, I guess since most of my connections to US go through Cogent I shouldn't be having internet at all let alone play flawlessly in the Sarlona game world (Now you're gonna suggest that Khyber goes through Cogent but Sarlona goes through.... what? Tata communications?)
    AT&T, NTT, Level 3, Worldcom... just to name a few... As for TataCom... well I have no clue if they are L2/L1 Carriers.

    Most of your connections to the US go through Cogent because your ISP chose to send them through Cogent and/or the ISP/website/whatever chose to have it's traffic routed by Cogent. It takes both end if you want to fully avoid a given carrier...

    Quote Originally Posted by Faltout View Post
    - A single card on a single router failing will cause the end of the world... *** is this? Are we in the 20th century? Today, abstraction in network systems has reached a whole new level. If your card in your home fails, you're done. But you're talking about a major organization that will spend money to ensure this thing does not happen. We have load balancing policies to make sure that there is not a single point of failure or traffic. We also have smart algorithms to detect such problems and automatically redirect traffic.
    Lol, you really have no clue. Abstraction is a nice commercial word... But in reality there is nothing abstract.
    Two switches straight from the box connected at the wrong place and in the wrong way can bring down a telecom operator backbone. ( or ISP )
    ( been there, saw it done and had to clean up the mess ).
    Now, I didn't say a single card on a single router causing the end of the world... I said a single card on a single router can cripple part of a telecom operator ( or ISP ) activity.

    Lets do some ASCII art :

    PC --- Router ( Box ) ( Called CPE ) ---- ( DSLAM ) ---- ( SWAG ) ---- ( PE ) --- ( LAC/LNS ) --- Backbone

    That's just your ISP diagram. ( they all look more or less like that )

    An Issue on the DSLAM will make you ( and any other people connected to said DSLAM ) have LAG and connection problems.
    Now the fun part about DSLAM is that they also have cards... so a given card can have a problem and it only affects the people connected on the card.

    The SWAG is an Agregation Switch, it just aggregate VLANS ( well PPPOE/PPPOA tunnels ) and send them to the PE on a single link.

    The PE is the end of the PPPOE/PPPOA tunnels so your IP traffic gets unencapsulated and sent to the LAC/LNS. That's where a lot of things happen.

    LAC/LNS is where you get authentified ( I won't go into that, it involves a few more nodes ) and where your actual link gets ended.
    After you're in the Backbone of your ISP... ( Aka P Routers ) Until you reach a Peering Node ( Aka a node that face another node in another ISP/Carrier backbone. ).
    Lets Say Cogent... Then you'll jump from Nodes to Nodes through the MPLS backbone of Cogent. Lets say that a Node near Boston suffer some congestion, and has to drop some traffic ( they have their own QOS rules... MMOs are not in the High Priority queues.... but that's the case for most ISP/Carriers anyway. )... so DDO traffic will get queued/dropped... to get that High Priority traffic through. You'll get Lag ( at best ) and dropped packets ( at worst ). Now Rinse Repeat for all the hops between you and PNAP Border router...

    Last do you have any clue how long it takes for an ISP/Carrier network to reconfigure when a node goes belly up ? At best it's a matter of 15 to 20 minutes.
    At worst it takes hours for all the routing tables to converge. And no redundancy/loadbalancing, whatever will change that. As for Smart Algorithms...
    Sure... they detect the problem, and force the routing table reconfiguration... See above for the time it takes to converge...

    Quote Originally Posted by Faltout View Post
    - Did you miss the part where we said that our connections are showing no packet loss?
    Nope... See above. MMO Traffic = Web Traffic = Low Priority. Congestion in a Node on a link = Low Priority Traffic gets queued, introducing lag. ( if the queue is not deep enough, the packets gets dropped eventually ).

    Quote Originally Posted by Faltout View Post
    -- something personal: In the example above, you didn't even find the cause of this "something wonky"? You just recommended a card restart? Wow, the guys working at the card from noon must have been really stupid to not think of that one earlier. I wonder what you would have done if the restart hadn't worked.
    I was dealing with an emergency, in that case, to hell with the cause, what is important is to fix the problem.
    In Telecom it's how it is, because some of the nodes carries emergency services....
    ( you really want to be left hanging with your 911 call because the exchange has a problem and people are investigating to find the cause while a system restart would solve the issue in 5 minutes ? )
    You find a way to get back the traffic, and then you investigate ( if still possible ) to find the cause... I didn't say I didn't find the cause afterwards... The cause is known, and yes they didn't think of a card restart... Because they weren't sure of the location of the issue in the network, so they were going up the chain one node at a time, once they had determined that a given node couldn't be the source of the issue, they moved to the next one.
    On G-Land : Flavilandile, Blacklock, Yaelle, Millishande, Larilandile, Gildalinde, Tenalafel, and many other...

  11. #11
    Community Member
    Join Date
    Sep 2013
    Posts
    350

    Default

    Another evening wasted because of unplayable lag.
    Its like Turbine doesn't want our money.

  12. #12
    Community Member Xionanx's Avatar
    Join Date
    Nov 2006
    Location
    Standing Behind You!
    Posts
    1,408

    Default

    So... I am just wondering if all these people who "Blame the Player" for having a ****** connection are suffering from Stockholm Syndrome. Just saying, I have only see two types of people on these forums.. people who have lag and understand its the fault of Turbine, and those who have lag and seem to think its because of some guy in their party and/or raid. (I suppose there is a 3rd type who apparently NEVER experiences lag of any kind and have no idea what any of us are talking about..)

    http://en.wikipedia.org/wiki/Stockholm_syndrome

    Just saying...

  13. #13
    Community Member
    Join Date
    Apr 2013
    Location
    Athens
    Posts
    2,105

    Default

    This is really fun
    Quote Originally Posted by Flavilandile View Post
    And I didn't talk about your computer specifically... your computer is just one potential source of lag...
    Now if you want I can give you exactly how a given computer in a party can lag out the whole party... that's called the Disconnect Lag.
    And that's caused by packet retransmission ( UDP ones ) from the server to the DCing computer... It locks down all the other people in the party, because
    they can't get a location update from that DCing player.
    The purpose of UDP against TCP is that UDP does not care about packets being dropped or packets arriving in the wrong order. So, even if you lose some of the packets, you don't care as long as the next ones come in. I don't know what DDO does (do you?), but a well thought out engine would ignore the packets lost. So, a DCing computer can neither send, nor receive any game update packets. So, the computer thinks the game does not have any updates to send and the game thinks the computer is sitting idly doing nothing. No reason to tell the other players "Look, this computer is DCing. Hold on while I try to get in touch with him". I don't see your reasoning for saying that a player's crappy connection can destroy others as well. Next thing you'll say that we should all play at exatly the same connection speed because the game will only send equal number of packets to each party member. Crazy talk.


    AT&T, NTT, Level 3, Worldcom... just to name a few... As for TataCom... well I have no clue if they are L2/L1 Carriers.

    Most of your connections to the US go through Cogent because your ISP chose to send them through Cogent and/or the ISP/website/whatever chose to have it's traffic routed by Cogent. It takes both end if you want to fully avoid a given carrier...
    Wow, you still don't read what I say?
    Just now you said
    Your Khyber connections pass through Cogent, but your Sarlona ones pass through another T1 provider. Do you read what you write? Jesus.

    Lol, you really have no clue. Abstraction is a nice commercial word... But in reality there is nothing abstract.
    Two switches straight from the box connected at the wrong place and in the wrong way can bring down a telecom operator backbone. ( or ISP )
    ( been there, saw it done and had to clean up the mess ).
    Now, I didn't say a single card on a single router causing the end of the world... I said a single card on a single router can cripple part of a telecom operator ( or ISP ) activity.

    Lets do some ASCII art :

    PC --- Router ( Box ) ( Called CPE ) ---- ( DSLAM ) ---- ( SWAG ) ---- ( PE ) --- ( LAC/LNS ) --- Backbone

    That's just your ISP diagram. ( they all look more or less like that )

    An Issue on the DSLAM will make you ( and any other people connected to said DSLAM ) have LAG and connection problems.
    Now the fun part about DSLAM is that they also have cards... so a given card can have a problem and it only affects the people connected on the card.

    The SWAG is an Agregation Switch, it just aggregate VLANS ( well PPPOE/PPPOA tunnels ) and send them to the PE on a single link.

    The PE is the end of the PPPOE/PPPOA tunnels so your IP traffic gets unencapsulated and sent to the LAC/LNS. That's where a lot of things happen.

    LAC/LNS is where you get authentified ( I won't go into that, it involves a few more nodes ) and where your actual link gets ended.
    After you're in the Backbone of your ISP... ( Aka P Routers ) Until you reach a Peering Node ( Aka a node that face another node in another ISP/Carrier backbone. ).
    Lets Say Cogent... Then you'll jump from Nodes to Nodes through the MPLS backbone of Cogent. Lets say that a Node near Boston suffer some congestion, and has to drop some traffic ( they have their own QOS rules... MMOs are not in the High Priority queues.... but that's the case for most ISP/Carriers anyway. )... so DDO traffic will get queued/dropped... to get that High Priority traffic through. You'll get Lag ( at best ) and dropped packets ( at worst ). Now Rinse Repeat for all the hops between you and PNAP Border router...
    A bunch of mostly unrelated information, till the last paragraph that actually reiterates what you've said in previous posts.
    First of all, I used the word "abstraction" because you keep talking like each ISP is using one connection, one card, one router, one DSLAM, etc. NO! They have redundant resources to minimize the single point of failure threat. And that's where abstraction comes in: You say "passes through the router" but in reality you mean "It passes through a series of routers".

    The problems you're actually mentioning will cause total packet loss, not partial. And that's not what we're complaining about.

    Last do you have any clue how long it takes for an ISP/Carrier network to reconfigure when a node goes belly up ? At best it's a matter of 15 to 20 minutes.
    At worst it takes hours for all the routing tables to converge. And no redundancy/loadbalancing, whatever will change that. As for Smart Algorithms...
    Sure... they detect the problem, and force the routing table reconfiguration... See above for the time it takes to converge...
    I can wait for 15 to 20 minutes Did you miss the part where we said "It's been going on for 2 weeks"? Now tell me that between those 2 weeks, noone has repaired that problem they supposedly have.



    Nope... See above. MMO Traffic = Web Traffic = Low Priority. Congestion in a Node on a link = Low Priority Traffic gets queued, introducing lag. ( if the queue is not deep enough, the packets gets dropped eventually ).
    You: packet loss, connection loss, etc.
    Me: But I'm noticing no packet loss. (meaning the game updates are coming in just fine, but they are not what they're supposed to be - they should tell me "your spell went off" but they're telling me "you're doing what you did last minute")
    You: the packets get dropped eventually
    Me: DO YOU SEE ME SAYING THE PACKETS ARE DROPPED???????? HOW THE HELL DO YOU KNOW THE PACKETS ARE BEING DROPPED? ON WHAT GROUNDS?
    My main server is Khyber. Have toons in almost every server for favor purposes. The Faltouts

  14. #14
    Community Member Flavilandile's Avatar
    Join Date
    Aug 2010
    Location
    France
    Posts
    3,908

    Default

    You still don't get it... and you don't understand what I write.

    Quote Originally Posted by Faltout View Post
    This is really fun

    The purpose of UDP against TCP is that UDP does not care about packets being dropped or packets arriving in the wrong order. So, even if you lose some of the packets, you don't care as long as the next ones come in. I don't know what DDO does (do you?), but a well thought out engine would ignore the packets lost. So, a DCing computer can neither send, nor receive any game update packets. So, the computer thinks the game does not have any updates to send and the game thinks the computer is sitting idly doing nothing. No reason to tell the other players "Look, this computer is DCing. Hold on while I try to get in touch with him". I don't see your reasoning for saying that a player's crappy connection can destroy others as well. Next thing you'll say that we should all play at exatly the same connection speed because the game will only send equal number of packets to each party member. Crazy talk.
    DDO uses UDP and TCP/HTTP with a hint of TCP/HTTPS for some specific things... Yes I know that as a fact... Wireshark comes handy...
    I agree that UDP has no resilience/retransmit coded into it, but there's nothing making it impossible to actually code something like that on top of UDP.
    Actually for semi-real time transmission where speed is more important it's better to use UDP and to code a handshaking system on top of it. DDO has something like that coded on top of the UDP transmission bit. There's issues with that handshaking system [ the issues have always been there ] that are most obvious when a player goes through a PPPOE tunnel reset ( DSL link reset ).
    That's due to how the RADIUS at the player end behave - 2-3 minutes timeout before it ( and the router ) consider the session dead and go through all the stuff to bring up a new session... The Handshake timer in DDO is lower than that ( 1 minute more or less ).
    A player disconnecting can induce lag to all the other players in the party, it's a known fact. And when it happens during a fight it's Not Fun.

    I never said that everybody should play with the sme connection speed... Connection speed doesn't matter, it's not as if the traffic for DDO was going to generate a congestion on a DSL/Cable link.
    Heck you can play DDO through a V92 Modem, it won't have any problem nor generate lag.
    Now yes, the game will send more or less the same amount of packets to all the players in the same instance... to keep the clients updated with all the things that happens in that instance.

    Quote Originally Posted by Faltout View Post

    Wow, you still don't read what I say?
    Just now you said
    Your Khyber connections pass through Cogent, but your Sarlona ones pass through another T1 provider. Do you read what you write? Jesus.
    I read what you say... But what you say has no meaning since I don't live where you live... ALL my connections to DDO go through Cogent.
    As i said... which Carrier you use for what is going to depend on what choices have been made at both end ( your ISP and PNAP/Turbine ).

    To make it simpler for you do a traceroute ( I hope you know how to do that ) here are the server IPs :

    Code:
    74.201.106.53 Wayfinder [DE]
    74.201.106.64 Ghallanda
    74.201.106.56 Argonnessen
    74.201.106.21 Thelanis
    74.201.106.22 Sarlona
    74.201.106.23 Khyber
    74.201.106.59 Cannith
    74.201.106.60 Orien
    Quote Originally Posted by Faltout View Post
    A bunch of mostly unrelated information, till the last paragraph that actually reiterates what you've said in previous posts.
    Actually they are quite related... since Lag can appear at any point beyond the DSLAM... due to QOS.


    Quote Originally Posted by Faltout View Post
    First of all, I used the word "abstraction" because you keep talking like each ISP is using one connection, one card, one router, one DSLAM, etc. NO! They have redundant resources to minimize the single point of failure threat. And that's where abstraction comes in: You say "passes through the router" but in reality you mean "It passes through a series of routers".
    Oh they don't have one connection... but the Peering Centers are not that numerous... And when a router/card/link goes down on a given center rerouting all the traffic through another center takes time... ( if the failure is clear... if the failure is just a link flapping... it can goes on for days before being actually solved, and in the meantime the routing tables will keep updating one way or another... Causing black holes, congestion and other Fun Stuff in the network. )

    I say passes through a router... I mean passes through a router... not a series of router... Once in the MPLS backbone passing through routers is supposedly transparent ( don't forget packet priority though )... but before the MPLS backbone of an ISP/Carrier/Operator and between ISP/Carriers/Operators, it's one router not a series of routers.

    Quote Originally Posted by Faltout View Post
    The problems you're actually mentioning will cause total packet loss, not partial. And that's not what we're complaining about.
    Not always, packets can still being sent and received while the network reconfigure... In some case there will be packet loss, in other case a huge latency will be induced ( read : LAG spikes from HELL ).


    Quote Originally Posted by Faltout View Post
    I can wait for 15 to 20 minutes Did you miss the part where we said "It's been going on for 2 weeks"? Now tell me that between those 2 weeks, noone has repaired that problem they supposedly have.
    Sorry, forgot to mention Link Flapping that generates routing reconfiguration over and over and over and over... generally when those occurs, it generates huges latency spikes ( and not that many packet loss )... Link Flapping can go on for months... especially when you have long distance links... before the cause of the issue is localized... ( oh since you love Abstraction : Link Flapping is something abstract : it can be BGP neighbors flapping, it can be an actual fiber that has a problem, it can be the transmission system on which the fiber is connected that has an issue, it can be the SFP/XFP on the card that has an issue, and so on... In the end it generate a routing reconfiguration. )



    Quote Originally Posted by Faltout View Post
    You: packet loss, connection loss, etc.
    Me: But I'm noticing no packet loss. (meaning the game updates are coming in just fine, but they are not what they're supposed to be - they should tell me "your spell went off" but they're telling me "you're doing what you did last minute")
    You: the packets get dropped eventually
    Me: DO YOU SEE ME SAYING THE PACKETS ARE DROPPED???????? HOW THE HELL DO YOU KNOW THE PACKETS ARE BEING DROPPED? ON WHAT GROUNDS?
    Yup, and you don't see any packet loss... does it matter ? no... as what is reported in DDO is not the real traffic.... you don't know how many retransmit were sent/received, so you don't know if there was any real packet lost... You even don't know if it's the incoming packets that are lost or if it's the outgoing packets that are lost with the Icon... ( when you have packet loss )

    If you're rubberbanding then you're having retransmit somewhere... And that means packets are lost somewhere.
    The client send your action result to the server, the server acknowledge it and sent it to all the other clients that are concerned by your action.
    If the server does not receive your action... and others do an action, the server will send you an update, that keeps your client informed of what others have done... along with where you are ( and what you're supposed to do ), if the received data does not match what the client has, the client updates to what he received from the server... Ding, you're back to where you were one minute ago.
    In the meantime the server might have received your lost update packet... but since it has been overruled by what has been sent, it just gets dropped.
    Last edited by Flavilandile; 12-30-2014 at 10:18 AM.
    On G-Land : Flavilandile, Blacklock, Yaelle, Millishande, Larilandile, Gildalinde, Tenalafel, and many other...

  15. #15
    2015 DDO Players Council
    Axel's DDO Channel
    axel15810's Avatar
    Join Date
    Nov 2010
    Location
    West Virginia
    Posts
    3,421

    Default

    I'd be all for this if it reduces lag. However, I doubt it will happen if it really takes an hour and a half to reboot the servers. Even if they did it in the middle of the night it would screw over people who are in non-USA timezones. Also, I assume someone would have to be on site to do it so not sure Turbine employees want to be working in the middle of the night on a regular basis.

  16. #16
    2015 DDO Players Council Seikojin's Avatar
    Join Date
    Feb 2006
    Location
    gamertown usa
    Posts
    7,435

    Default

    Quote Originally Posted by Kerria View Post
    The lag is double if not worse the past week to the point of total stand still, then wipe. It acts the same in public areas as well. It is server wide on Kyber for sure. Bank, bags, vendors all hang at times for long periods. Don't write it off as normal lag, been around too long and I know how much lag I am paying for. This is the give me a refund amt.
    Kyber? Didn't see problems doing my weekly tasks and dice on there.

  17. #17
    Community Member
    Join Date
    Sep 2009
    Posts
    62

    Default

    Quote Originally Posted by Seikojin View Post
    Kyber? Didn't see problems doing my weekly tasks and dice on there.
    We have gone for long stretches with no issues then it will be the lag from hell for about 10 to 30 min, random and server wide. But it is does happen repeatedly over any six hour stretch regardless. I don't want to know how turbine funnels the packets, what IP they use. If other servers are not seeing this then Kyber needs to be fixed. Any response from anyone at turbine would be nice right now.

  18. #18
    Community Member
    Join Date
    Sep 2013
    Posts
    350

    Default

    Quote Originally Posted by axel15810 View Post
    I'd be all for this if it reduces lag. However, I doubt it will happen if it really takes an hour and a half to reboot the servers. Even if they did it in the middle of the night it would screw over people who are in non-USA timezones. Also, I assume someone would have to be on site to do it so not sure Turbine employees want to be working in the middle of the night on a regular basis.
    Do it first thing they come in. An hour and half downtime is better than lag throughout the day. It's better than doing nothing at all.

  19. #19
    Community Member
    Join Date
    Nov 2013
    Posts
    91

    Thumbs up New hamster

    Quote Originally Posted by Fallout_Zero View Post
    Since they are unable to fix the lag, a band aid solution is daily reboot.

    Clears connections.
    Clears ghost LFMs.
    They need to pull the old Hamster out of there dark side before they can put it back on the sever wheel that they have now . But they are enjoying it to much so don't expect much from them .

  20. #20
    Community Member Keladon's Avatar
    Join Date
    Sep 2010
    Posts
    325

    Default

    You MIGHT be able to reduce lag, but I'm afraid it would cause a wave of complaints.

    It might be in the early morning or even at night for you, but for others this maintenance will be during prime time.

    It's a small price to pay.. unless you're in the wrong timezone

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

This form's session has expired. You need to reload the page.

Reload