Lag

by Estarra

Back to Common Grounds.

Ssaliss2011-09-29 21:50:21
I take a timestamp when I send the request, and calculate the difference when I get Core.Ping back.
Ssaliss2011-09-29 22:25:16
Well, yeah. The full docs regarding Core.Ping is:

QUOTE
Sent by client:
- Core.Ping
* causes the server to send a Core.Ping back
* message body is a number which indicates average ping time from previous requests, if available
* example: Core.Ping 120
Sent by server:
- Core.Ping
* Sent in reply to Core.Ping. No body.

Personally I just send the last ping time to the server, but I'm not sure that number is even used in Lusternia.
Vadi2011-09-29 22:34:46
It's probably a Nexus artefact. I'm not quite what's happening, but what we want mostly fixed is the connection lag that occurs, not the server one as much. To measure that well, I think several people would need to setup logging w/ ping, and then compare ping data, see where everyone got a spike, and look at the logs to see what was happening - maybe something can come out of it.
Ssaliss2011-09-29 22:38:40
Well, the thing is... I've got _no_ connection lag whatsoever (I've got a rock-solid 128-129-130ms connection to Lusty), yet I still hit lag with my GMCP-pings.

Also, according to post 7 and 8 (Xenthos and Estarra) the server-lag is what they're after. And GMCP is perfect for measuring that.
Unknown2011-09-30 01:39:17
Has it been fixed then that the Core.Ping replies would only come on the next prompt? That was causing a lot of timing confusion a while back.
Ssaliss2011-09-30 01:45:08
I get them right on time, at least. Never even noticed that I didn't, to be honest...
Vadi2011-09-30 02:00:59
They'll be on time if you have MCCP off, next prompt if MCCP is on.
Saqa2011-09-30 02:07:22
QUOTE (Zarquan @ Sep 30 2011, 02:39 AM) <{POST_SNAPBACK}>
Has it been fixed then that the Core.Ping replies would only come on the next prompt? That was causing a lot of timing confusion a while back.


That bug only manifests when MCCP is enabled.

On the topic of lag, from a few minutes ago:

CODE
You have recovered balance on your right arm.(02:47:29.02)
You have recovered balance on your left arm.(02:47:29.03)
You attempt the Kata form of huntingt.(02:47:29.05)
With a quick spin, you kick a hard-shelled, many-legged cave-fisher with your left foot.(02:47:29.07)
With a quick spin, you strike a hard-shelled, many-legged cave-fisher with a luminescent coral spire
tahto.(02:47:29.11)
You have scored a CRUSHING CRITICAL hit!(02:47:29.13)
With a quick spin, you strike a hard-shelled, many-legged cave-fisher with a luminescent coral spire
tahto.(02:47:29.16)
3908h 3105m 3105e 10p 2mo ex-(02:47:29.18)
You may drink another health, mana, or bromide potion.(02:47:29.21)
3908h 3105m 3105e 10p 2mo ex-(02:47:29.24)
Leaping forward through the air toward you, a hard-shelled, many-legged cave-fisher wraps his many
lanky legs about your body and gives a powerful squeeze.(02:47:30.36)
3019h 3105m 3105e 10p 2mo ex-(02:47:30.38)(drink health)
You down the last drop from a jade vial.(02:47:30.63)
The potion heals and soothes you.(02:47:30.64)
3754h 3105m 3105e 10p 2mo ex-(02:47:30.66): Resetting Health balance.
(drink health)(02:47:34.86)
(02:47:36.49)
(02:47:38.39)
You bleed 55 health.(02:47:45.46)
3699h 3105m 3105e 10p 2mo ex-(02:47:45.47)
You have recovered balance on your left leg.(02:47:45.51)
3699h 3105m 3105e 10p 2mo ex-(02:47:45.53)
You have recovered balance on your right arm.(02:47:45.57)
You have recovered balance on your left arm.(02:47:45.58)
The huntingtswift Kata form is too complex for you to execute at this time.(02:47:45.59)
Leaping forward through the air toward you, a hard-shelled, many-legged cave-fisher wraps his many
lanky legs about your body and gives a powerful squeeze.(02:47:45.60)
3887h 3105m 3105e 10p 1mo exL-(02:47:45.62)
You take a drink from a ruby vial.(02:47:45.82)
The potion flows down your throat without effect.(02:47:45.83)
3887h 3105m 3105e 10p 1mo exL-(02:47:45.86)
3887h 3105m 3105e 10p 1mo exL-(02:47:46.13)
3887h 3105m 3105e 10p 1mo exL-(02:47:46.32)
3887h 3105m 3105e 10p 1mo exL-(02:47:46.41)
You may drink another health, mana, or bromide potion.(02:47:46.51)
3887h 3105m 3105e 10p 1mo exL-(02:47:46.53)
A hard-shelled, many-legged cave-fisher leans forward and bashes you with his massive shell, causing
agonising pain to vibrate through your body.(02:47:48.72)
(drink health)(02:47:48.74)
3062h 3105m 3105e 10p 1mo exL-(02:47:48.75)
You take a drink from a ruby vial.(02:47:49.04)
The potion heals and soothes you.(02:47:49.06)


Chained forms, note the momentum drop, the 15s potion balance and the pause in the cave-fisher's attacks. Further note that the third 'drink health' went through promptly, demonstrating that my latency didn't suddenly jump to 12 seconds and stay there.
Vadi2011-09-30 04:34:21
Estarra: an easily implementable solution would be to accept responsibility for lag-based deaths and issue refunds, given that proper evidence is given (maybe a screenshot with timestamps).
Lerad2011-09-30 04:57:52
That's unlikely to happen anytime soon. There's a good reason for the administration for IRE games not offering exp refunds and all. The sheer number of complaints, valid and invalid both, will tie up more resources than anything else. For some reason, people are very attached to their exp.

Besides which, that opens up the floodgates to more and more unreasonable refunds. A rollback to refund hundreds and thousands of gold/exp? With exp refunds for lagdeaths in place, you can be sure there will be people clamouring for rollback refunds. And rollback affects 100% of the playerbase. So the admin has to, what, refund all players their exp/gold? If they could do that, there wouldn't be rollback losses. Yet you can't justify refunding only a few people for rollback losses - you'd have to do it for everyone who asks if you do it at all. Etc etc.
Everiine2011-09-30 21:56:34
And experience is so ridiculously easy to get back anyway (even comparably at Titan, which yes, will still take quite a long time) that it's not worth the hassle.
Ardmore2011-10-01 04:01:28
I'd say whatever you guys did, it's 10x worse. Or it's just the poorly coded event causing this ridiculous lag.