|
On June 26 2012 11:01 Sufficiency wrote:Show nested quote +On June 26 2012 10:58 obesechicken13 wrote:On June 26 2012 10:47 Sufficiency wrote:On June 26 2012 09:23 Goragoth wrote:On June 26 2012 09:07 Sermokala wrote:
if you look at the client when you login they have a new stamp for it from the 15th that's around when the problem started. There are new fonts for everything which suggests a major change to the client probably server side.
Its pretty funny beacuse all the patch was is a bunch of quality of life changes to some champs and other small business that could have waited until the next champ came out but instead they introduced it now and everything went to shit.
They changed a lot of backend/under-the-hood stuff with this patch, hence the breakage. Patch notes only ever contain stuff relevant to players and never things to do with the internals. That's why it's been wreaking havoc (specifically the main problem, or one of the main problems, was that they introduced a new bottleneck on the database side of things which didn't show up in testing but destroyed the servers when it went live). Here's the problem. Riot has almost certainly modified certain services on the server during the 15th patch. Afterwards the server ran into problems which basically is still happening now. The issue, as far as I can see, is a stress-load issue; we have a long login-queue right now because the server can no longer handle the same amount of players as it could before. In the end I think it's still caused by bugs in their code which is causing higher server load. If there are any bugs in their code. Any threads or processes that run for longer than they should, then they will continue to run and mess with how much load the machines are experiencing. I don't see why Riot hasn't just restarted their servers if it were a hanging process issue. And since they're so heavily loaded, I'd expect them to have someone to monitor cpu usage by process type. They could also theoretically revert their changes since they probably use version control. Does anyone have any idea of what's going on with the riot servers? Could it be hardware related? I hypothesize that all the "processes" are running fine, but they don't run as efficiently as they used to because of the recent changes. This also makes it difficult for them to buy new hardware, because the hardware is fine and sufficient but the software has issues.
They would've just restarted the servers
On June 26 2012 11:06 Goragoth wrote:Show nested quote +On June 26 2012 10:58 obesechicken13 wrote:On June 26 2012 10:47 Sufficiency wrote:On June 26 2012 09:23 Goragoth wrote:On June 26 2012 09:07 Sermokala wrote:
if you look at the client when you login they have a new stamp for it from the 15th that's around when the problem started. There are new fonts for everything which suggests a major change to the client probably server side.
Its pretty funny beacuse all the patch was is a bunch of quality of life changes to some champs and other small business that could have waited until the next champ came out but instead they introduced it now and everything went to shit.
They changed a lot of backend/under-the-hood stuff with this patch, hence the breakage. Patch notes only ever contain stuff relevant to players and never things to do with the internals. That's why it's been wreaking havoc (specifically the main problem, or one of the main problems, was that they introduced a new bottleneck on the database side of things which didn't show up in testing but destroyed the servers when it went live). Here's the problem. Riot has almost certainly modified certain services on the server during the 15th patch. Afterwards the server ran into problems which basically is still happening now. The issue, as far as I can see, is a stress-load issue; we have a long login-queue right now because the server can no longer handle the same amount of players as it could before. In the end I think it's still caused by bugs in their code which is causing higher server load. If there are any bugs in their code. Any threads or processes that run for longer than they should, then they will continue to run and mess with how much load the machines are experiencing. I don't see why Riot hasn't just restarted their servers if it were a hanging process issue. And since they're so heavily loaded, I'd expect them to have someone to monitor cpu usage by process type. They could also theoretically revert their changes since they probably use version control. Does anyone have any idea of what's going on with the riot servers? Could it be hardware related? My guess as to why they haven't reverted back to Draven again (they already did once so clearly they could do it again) is that they had a hard time identifying the bug(s) that were causing the problems without having the system running at full capacity. By keeping the new code live and restricting log ons they are no doubt trying to pinpoint the problem(s) and fix them as they go. I'd do the same, rather than revert the patch again and stab in the dark hoping to fix the issue. Despite what people might think testing a system to be stable with millions of users without actually exposing it to millions of users is kinda hard.
They're using us for stress testing?!
I agree. it is hard to stress test without the actual conditions. Even if you record user actions and replay them back on the servers and just simulate how many users are online, you need to stress all the servers to simulate load balancing or to see if the bottleneck is in the network. + Show Spoiler +Is this the issue or am I completely off?
It's possible the summer's just brought them more users than they could handle. It'd be stupid for a company with such stressed servers, but perhaps they don't have anyone doing capacity planning. And when servers are overloaded, they just have everyone working on that stuff instead of working on bugs with Mao, Lee, or Baron. I really hope this isn't the problem. I've heard a few times that large backend machines can take months to order and install.
They're might be losing money every second. People are playing other games while servers are down. They'll come back, but during that time they aren't buying champions.
Also of interest is Amazon's Elastic Cloud. I've heard that Amazon has had (for a few years) a service where they offer companies cloud based database machines. They deal with server maintenance so you don't have to. Amazon has a lot of seasonal load during Christmas when people are buying presents, so they decided that rather than buy more than they needed for the rest of the year, they would create a shared database where people would only use and pay for as much as they needed. Different companies have different seasonal peaks. Riot's probably are over the break periods (winter, spring, summer). I doubt Riot uses Elastic Cloud though. I doubt Riot would use the cloud since they might want more control over their backend and it might actually be cheaper to own rather than lease but it is interesting. Some small-medium sized companies use Amazon Web services. The only names I really recognize are IMDB, SEGA, and PBS http://aws.amazon.com/solutions/case-studies/
I have no idea whether Amazon was the first to use this service or even if the service is reliable. Again, the only companies I see using it are small ones. Just something interesting.
...
so. How bout that riven.
|
|
|
When they split the servers they claimed it was the only way they could handle the load because of how everything was set up. It's not particularly surprising given how much they continue to build on what was a pretty bad platform initially. Who knows what kind of steps they've taken since that time. Every software developer is familiar with the line "get it working now and we'll fix it later." Best intentions, etc., etc.
Throwing more hardware at it will almost certainly have a very small effect at this point. Clearly something in the latest patch fucked a lot more things up than just in-game bugs.
They could also theoretically revert their changes since they probably use version control.
Version control is only a VERY small part of rolling back changes. You have to aggregate all the data that's occurred since the patches, create processes that can translate it back into the old formats, somehow save the states of whatever data needs to be kept in the new format, and then down the road create a process for re-integrating the two.
This is independent of whatever other behind-the-scenes changes have also taken place that now rely on whatever new things they've implemented, too.
Rolling back code literally only takes a few button presses, but rolling back a production system without loss of integrity is hard.
|
On June 26 2012 11:48 Craton wrote:When they split the servers they claimed it was the only way they could handle the load because of how everything was set up. It's not particularly surprising given how much they continue to build on what was a pretty bad platform initially. Who knows what kind of steps they've taken since that time. Every software developer is familiar with the line "get it working now and we'll fix it later." Best intentions, etc., etc. Throwing more hardware at it will almost certainly have a very small effect at this point. Clearly something in the latest patch fucked a lot more things up than just in-game bugs. Show nested quote +On June 26 2012 11:06 Goragoth wrote:
They could also theoretically revert their changes since they probably use version control. Version control is only a VERY small part of rolling back changes. You have to aggregate all the data that's occurred since the patches, create processes that can translate it back into the old formats, somehow save the states of whatever data needs to be kept in the new format, and then down the road create a process for re-integrating the two. This is independent of whatever other behind-the-scenes changes have also taken place that now rely on whatever new things they've implemented, too. Rolling back code literally only takes a few button presses, but rolling back a production system without loss of integrity is hard.
Gorogath didn't say that. I did.
But thanks.
|
Riot has said a rollback is pretty much impossible at this point, since the new patch changed how data was stored internally. They'd have to rollback the databases to their old states too, which means players would lose all progress since the last rollback.
|
You watch Nhat + aphro because it's entertaining and it's the 2 of them having fun.
They rarely tryhard in ranked games
|
On June 26 2012 05:36 mcimba42 wrote:Show nested quote +On June 26 2012 04:53 RogerX wrote:On June 26 2012 04:38 TheYango wrote:On June 26 2012 03:44 barbsq wrote:
i think something like sc2's mapmaker would be really cool for stuff like dota2/lol/etc Two words: Pudge Wars 3 words, Luthers Party Ultima. ...luther?
Arthur,
DONT BLAME ME, i haven't played War3 mods for ages but i miss them.
|
On June 26 2012 12:32 RogerX wrote:Show nested quote +On June 26 2012 05:36 mcimba42 wrote:On June 26 2012 04:53 RogerX wrote:On June 26 2012 04:38 TheYango wrote:On June 26 2012 03:44 barbsq wrote:
i think something like sc2's mapmaker would be really cool for stuff like dota2/lol/etc Two words: Pudge Wars 3 words, Luthers Party Ultima. ...luther? Arthur, DONT BLAME ME, i haven't played War3 mods for ages but i miss them.
Its Uther......
|
On June 26 2012 12:58 BarneyEX wrote:Show nested quote +On June 26 2012 12:32 RogerX wrote:On June 26 2012 05:36 mcimba42 wrote:On June 26 2012 04:53 RogerX wrote:On June 26 2012 04:38 TheYango wrote:On June 26 2012 03:44 barbsq wrote:
i think something like sc2's mapmaker would be really cool for stuff like dota2/lol/etc Two words: Pudge Wars 3 words, Luthers Party Ultima. ...luther? Arthur, DONT BLAME ME, i haven't played War3 mods for ages but i miss them. Its Uther......
I was thinking the same thing.
|
how is vvv skilly not banned lol, he's like the only person i've seen at this elo that cusses out everyone for everything literally every game
also is this guy srsly on a pro team lol
|
On June 26 2012 13:21 Xevious wrote:
how is vvv skilly not banned lol, he's like the only person i've seen at this elo that cusses out everyone for everything literally every game
also is this guy srsly on a pro team lol
Depends on how loosely you define "pro team."
|
You seem to fail to realize that a lot of US "pro" players are high schoolers.
|
On June 26 2012 11:21 obesechicken13 wrote:Show nested quote +On June 26 2012 11:01 Sufficiency wrote:On June 26 2012 10:58 obesechicken13 wrote:On June 26 2012 10:47 Sufficiency wrote:On June 26 2012 09:23 Goragoth wrote:On June 26 2012 09:07 Sermokala wrote:
if you look at the client when you login they have a new stamp for it from the 15th that's around when the problem started. There are new fonts for everything which suggests a major change to the client probably server side.
Its pretty funny beacuse all the patch was is a bunch of quality of life changes to some champs and other small business that could have waited until the next champ came out but instead they introduced it now and everything went to shit.
They changed a lot of backend/under-the-hood stuff with this patch, hence the breakage. Patch notes only ever contain stuff relevant to players and never things to do with the internals. That's why it's been wreaking havoc (specifically the main problem, or one of the main problems, was that they introduced a new bottleneck on the database side of things which didn't show up in testing but destroyed the servers when it went live). Here's the problem. Riot has almost certainly modified certain services on the server during the 15th patch. Afterwards the server ran into problems which basically is still happening now. The issue, as far as I can see, is a stress-load issue; we have a long login-queue right now because the server can no longer handle the same amount of players as it could before. In the end I think it's still caused by bugs in their code which is causing higher server load. If there are any bugs in their code. Any threads or processes that run for longer than they should, then they will continue to run and mess with how much load the machines are experiencing. I don't see why Riot hasn't just restarted their servers if it were a hanging process issue. And since they're so heavily loaded, I'd expect them to have someone to monitor cpu usage by process type. They could also theoretically revert their changes since they probably use version control. Does anyone have any idea of what's going on with the riot servers? Could it be hardware related? I hypothesize that all the "processes" are running fine, but they don't run as efficiently as they used to because of the recent changes. This also makes it difficult for them to buy new hardware, because the hardware is fine and sufficient but the software has issues. They would've just restarted the servers Show nested quote +On June 26 2012 11:06 Goragoth wrote:On June 26 2012 10:58 obesechicken13 wrote:On June 26 2012 10:47 Sufficiency wrote:On June 26 2012 09:23 Goragoth wrote:On June 26 2012 09:07 Sermokala wrote:
if you look at the client when you login they have a new stamp for it from the 15th that's around when the problem started. There are new fonts for everything which suggests a major change to the client probably server side.
Its pretty funny beacuse all the patch was is a bunch of quality of life changes to some champs and other small business that could have waited until the next champ came out but instead they introduced it now and everything went to shit.
They changed a lot of backend/under-the-hood stuff with this patch, hence the breakage. Patch notes only ever contain stuff relevant to players and never things to do with the internals. That's why it's been wreaking havoc (specifically the main problem, or one of the main problems, was that they introduced a new bottleneck on the database side of things which didn't show up in testing but destroyed the servers when it went live). Here's the problem. Riot has almost certainly modified certain services on the server during the 15th patch. Afterwards the server ran into problems which basically is still happening now. The issue, as far as I can see, is a stress-load issue; we have a long login-queue right now because the server can no longer handle the same amount of players as it could before. In the end I think it's still caused by bugs in their code which is causing higher server load. If there are any bugs in their code. Any threads or processes that run for longer than they should, then they will continue to run and mess with how much load the machines are experiencing. I don't see why Riot hasn't just restarted their servers if it were a hanging process issue. And since they're so heavily loaded, I'd expect them to have someone to monitor cpu usage by process type. They could also theoretically revert their changes since they probably use version control. Does anyone have any idea of what's going on with the riot servers? Could it be hardware related? My guess as to why they haven't reverted back to Draven again (they already did once so clearly they could do it again) is that they had a hard time identifying the bug(s) that were causing the problems without having the system running at full capacity. By keeping the new code live and restricting log ons they are no doubt trying to pinpoint the problem(s) and fix them as they go. I'd do the same, rather than revert the patch again and stab in the dark hoping to fix the issue. Despite what people might think testing a system to be stable with millions of users without actually exposing it to millions of users is kinda hard. They're using us for stress testing?! I agree. it is hard to stress test without the actual conditions. Even if you record user actions and replay them back on the servers and just simulate how many users are online, you need to stress all the servers to simulate load balancing or to see if the bottleneck is in the network. + Show Spoiler +Is this the issue or am I completely off? It's possible the summer's just brought them more users than they could handle. It'd be stupid for a company with such stressed servers, but perhaps they don't have anyone doing capacity planning. And when servers are overloaded, they just have everyone working on that stuff instead of working on bugs with Mao, Lee, or Baron. I really hope this isn't the problem. I've heard a few times that large backend machines can take months to order and install. They're might be losing money every second. People are playing other games while servers are down. They'll come back, but during that time they aren't buying champions. Also of interest is Amazon's Elastic Cloud. I've heard that Amazon has had (for a few years) a service where they offer companies cloud based database machines. They deal with server maintenance so you don't have to. Amazon has a lot of seasonal load during Christmas when people are buying presents, so they decided that rather than buy more than they needed for the rest of the year, they would create a shared database where people would only use and pay for as much as they needed. Different companies have different seasonal peaks. Riot's probably are over the break periods (winter, spring, summer). I doubt Riot uses Elastic Cloud though. I doubt Riot would use the cloud since they might want more control over their backend and it might actually be cheaper to own rather than lease but it is interesting. Some small-medium sized companies use Amazon Web services. The only names I really recognize are IMDB, SEGA, and PBS http://aws.amazon.com/solutions/case-studies/I have no idea whether Amazon was the first to use this service or even if the service is reliable. Again, the only companies I see using it are small ones. Just something interesting. ... so. How bout that riven.
as someone who flew down to riot and applied to be a platforming engineering intern (maybe not entirely knowingly but that's another matter), holy shit does this thread make my head hurt right now.
please stop
|
On June 26 2012 13:26 0123456789 wrote:
You seem to fail to realize that a lot of US "pro" players are high schoolers.
This, just look at team Dynamic lol
|
i finally picked up renekton and holy shit there's nothing more satisfying than letting enemy jungler gank you, running into your brush instead of to your tower, ulting, and collecting your double kill when they follow you in
fun champ
|
It's a matchup I so rarely see:
What AP mids can hold their own against talon?
And yeah gtrsrs, since playing renek a lot I can't count the number of double kills I've gotten from attempted jungle ganks.
|
Anyone know if JP from state of the game plays League.. Just played someone called CallmeJP
|
On June 26 2012 15:07 Shelke14 wrote:
Anyone know if JP from state of the game plays League.. Just played someone called CallmeJP
I think he's said before that he doesn't like it at all, so I doubt it.
|
On June 26 2012 14:22 gtrsrs wrote:
i finally picked up renekton and holy shit there's nothing more satisfying than letting enemy jungler gank you, running into your brush instead of to your tower, ulting, and collecting your double kill when they follow you in
fun champ
yeah renny is a qtpie 
|
On June 22 2012 14:24 Sufficiency wrote:
I played Sona on a smurf yesterday to take advantage of the IP boost. For some reason, I thought I locked in Rumble but the game gave me Sona instead. I felt Sona could be a good bruiser if her HP and HP per level was higher. She has everything going for her that makes her a suitable top laner.
Q: ranged poke with low CD/AD steroid
W: defense and sustain
E: movement steroid
R: teamfight ability
Not to mention she has good pokes through her passive, a slow, and she has 550 range.
One can only dream... perhaps one day we can finally see huge breasts on the top lane.
I saw voyboy play sona top :o
|
|
|
|
|
|
EPT
