bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 We spent the day optimizing the code a touch, and trying to figure out which feature could be causing the kicks from BE. We've tweaked several things and are testing it out this evening. Haych, Google™, CriticalTug and 2 others like this Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, bamf said: We spent the day optimizing the code a touch, and trying to figure out which feature could be causing the kicks from BE. We've tweaked several things and are testing it out this evening. that's good, it's honestly the best thing you can do at this point man. i like that the updates are more constant and more shit is being pushed out, (huge difference from back then), but ALWAYS make sure that things are running smoothly before adding anything more to the cake. Link to comment
JIMBO Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, bamf said: We spent the day optimizing the code a touch, and trying to figure out which feature could be causing the kicks from BE. We've tweaked several things and are testing it out this evening. Cap 2 servers to 70 players until you find a fix and see if that'll help, I only seem to get kicked when server has > 70 players, it's lag free until then, can play all morning EU times without a problem, soon as we hit more than 70 players server takes a shit, kicks a bunch, fine again.. Link to comment
bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 3 minutes ago, Jaxon said: that's good, it's honestly the best thing you can do at this point man. i like that the updates are more constant and more shit is being pushed out, (huge difference from back then), but ALWAYS make sure that things are running smoothly before adding anything more to the cake. Well things always run fine in the dev environment before getting rolled out (at least we are testing all of the features before pushing them live). The issue we have now is that there is something that only crops up during heavy load - and is only on Altis. We narrowed down a good deal of the things that sit only on the one map today and are testing things out. That being said. The federal reserve is currently NOT spawning gold bars. It is one of those features that is on Altis but not on Tanoa. Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 1 minute ago, bamf said: Well things always run fine in the dev environment before getting rolled out (at least when I'm doing the testing). The issue we have now is that there is something that only crops up during heavy load - and is only on Altis. We narrowed down a good deal of the things that sit only on the one map today and are testing things out. That being said. The federal reserve is currently NOT spawning gold bars. It is one of those features that is on Altis but not on Tanoa. as i'm not the best at .sqf, please forgive me for any stupidity that may go into my next statement you could make dev-server only scripts that basically runs stress tests on different scripts you make, pushing them to the limit (simulate the amount of average players) on the server at once, but then again i don't know if this is possible but i know typically game companies have done this in the past Link to comment
JIMBO Posted September 19, 2016 Report Share Posted September 19, 2016 1 minute ago, bamf said: Well things always run fine in the dev environment before getting rolled out (at least we are testing all of the features before pushing them live). The issue we have now is that there is something that only crops up during heavy load - and is only on Altis. We narrowed down a good deal of the things that sit only on the one map today and are testing things out. That being said. The federal reserve is currently NOT spawning gold bars. It is one of those features that is on Altis but not on Tanoa. We noticed alot of server lag towards the end of prison breaks and banks for a while (usually when swat deployed), haven't tested this an awful lot since new prison/bank though. Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) 6 minutes ago, Gnashes said: Another thing there is that Bamf, Speed, and myself are usually working on new things / changes / bug fixes, but aren't necessarily the best people when it comes to optimization. (I'll admit I know absolutely fuckall about optimization; I'm just persistent enough to make something work.) The entire 7.5.5 series of hotfixes (the 6? of them) were Bamf and Myself just going ham on a lot of bugs and small changes that needed to be made. i know before i was banned things actually ran pretty smooth. but this was before the huge bounty hunter update + everything else came along - i can't really suggest anything to do in this case because i can't imagine you guys have a saved copy of that specific version before the bounty hunter update.. but if you do, it wouldn't be a bad idea to put up a beta server for a couple days and allow the general public to try that version and see if anything changes - that way it could determine if it's a script issue or a hardware / datacenter issue. Edited September 19, 2016 by Jaxon .Sean likes this Link to comment
.Sean Posted September 19, 2016 Report Share Posted September 19, 2016 Remove skiptracer see if that helps nobody uses it unless it's buying hunters (which you could make available at the bounty hunter shop) and nobody wastes their time to go up there to strip Link to comment
.Sean Posted September 19, 2016 Report Share Posted September 19, 2016 2 minutes ago, Gnashes said: Things that exist but never run won't impact performance.... They are still in the mission file... Link to comment
Hanzo/Dirty Scrubz Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) 23 minutes ago, bamf said: We spent the day optimizing the code a touch, and trying to figure out which feature could be causing the kicks from BE. We've tweaked several things and are testing it out this evening. Can you please consider using a different host as a test? Just one server and see if it fares better than OVH. I'm willing to bet it will during peak times if you pick a good host. Also, why not open up a dev-test server with 100 slots and give everyone millions in cash to do what they want to stress test? I bet many players will hop on just to fuck around and do test fed/bank etc and you guys get instant feedback. Edited September 19, 2016 by Hanzo/Dirty Scrubz Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 8 minutes ago, Jaxon said: i know before i was banned things actually ran pretty smooth. but this was before the huge bounty hunter update + everything else came along - i can't really suggest anything to do in this case because i can't imagine you guys have a saved copy of that specific version before the bounty hunter update.. but if you do, it wouldn't be a bad idea to put up a beta server for a couple days and allow the general public to try that version and see if anything changes - that way it could determine if it's a script issue or a hardware / datacenter issue. also to add on to this, you could probably compile a list of every change that's happened since then and determine what COULD impact performance and what's probably safe to leave in - from there just play around with things Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 1 minute ago, House. said: They are still in the mission file... it doesn't matter. just because i put " // lol this is an epic comment" doesn't mean it's gonna effect the performance in any way Link to comment
bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 1 minute ago, Hanzo/Dirty Scrubz said: Can you please consider using a different host as a test? Just one server and see if it fares better than OVH. I'm willing to bet it will during peak times if you pick a good host. At this point, we do not believe it is the host. Just now, Jaxon said: also to add on to this, you could probably compile a list of every change that's happened since then and determine what COULD impact performance and what's probably safe to leave in - from there just play around with things Yes, we've done that. There is a repository that keeps track of all changes, but the sheer volume of changes that have happened since an unknown date is tough. Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, Hanzo/Dirty Scrubz said: Can you please consider using a different host as a test? Just one server and see if it fares better than OVH. I'm willing to bet it will during peak times if you pick a good host. OVH is not bad. OVH is actually one of the best hosts out there right now, with the right specs the server can run fine. although if you are pushing the server to it's limits with performance heavy scripts then that can effect it - this goes with any host. Link to comment
Hanzo/Dirty Scrubz Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) 3 minutes ago, Jaxon said: OVH is not bad. OVH is actually one of the best hosts out there right now, with the right specs the server can run fine. although if you are pushing the server to it's limits with performance heavy scripts then that can effect it - this goes with any host. OVH is okay if you are looking to save money and get free DDOS protection. I ran a 100 player server on it for Altis and other games and had issues w/them. Moved to NFO and all the problems went away. I've been watching that Chinese Asylum clone and they're using an old build of Asylum with around 30-50 players on there and so far I haven't seen a single yellow chain on that clone. Yet lately on Asylum I see yellow chain and memory leaks with 30 people or less. Edited September 19, 2016 by Hanzo/Dirty Scrubz Link to comment
bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 1 minute ago, Jaxon said: OVH is not bad. OVH is actually one of the best hosts out there right now, with the right specs the server can run fine. although if you are pushing the server to it's limits with performance heavy scripts then that can effect it - this goes with any host. The bottom line with Arma is that until BIS takes the small amount of time and effort to compile the server.exe as a 64bit application, it's always going to be sub-optimal. Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, bamf said: At this point, we do not believe it is the host. Yes, we've done that. There is a repository that keeps track of all changes, but the sheer volume of changes that have happened since an unknown date is tough. if you're using github you can view any change history, although i'm not sure if it actually saves any type of backup though. in that case, i don't really have any suggestions as you'd have to go through every change and re-configure any database saving scripts and all that bullshit. the forum updates are a good way to view those dates specifically, though - or a close range around the specific date Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, Gnashes said: I've had Contributor access for about two and a half weeks, and have pushed through just a hair over 50 commits personally. Going through the changelog is easier said than done. ah, that i didn't imagine haha. I agree that makes things harder.. lol. are you able to search commits? Link to comment
Jaxon Posted September 19, 2016 Report Share Posted September 19, 2016 7 minutes ago, Hanzo/Dirty Scrubz said: OVH is okay if you are looking to save money and get free DDOS protection. I ran a 100 player server on it for Altis and other games and had issues w/them. Moved to NFO and all the problems went away. I've been watching that Chinese Asylum clone and they're using an old build of Asylum with around 30-50 players on there and so far I haven't seen a single yellow chain on that clone. Yet lately on Asylum I see yellow chain and memory leaks with 30 people or less. NFO is a good host, yes, but i could perform the same performance on an OVH box. the difference between NFO is that they have a custom built automated firewall script that is brilliantly built by some very smart guys. 8 minutes ago, bamf said: The bottom line with Arma is that until BIS takes the small amount of time and effort to compile the server.exe as a 64bit application, it's always going to be sub-optimal. and yeah, obviously arma is also one of the key factors for lag - bethesda has never been good at optimization. honestly i just think that a list needs to be compiled of what could be causing performance issues and what isn't, like stated above. it could help pin some things down and from there you could test those individual things or even think about different ways to do them again, i don't know sqf but i'm going off personal experience of other projects i've done Link to comment
CriticalTug Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) Here is suicide vest changes for any one wanting to see destruction: Edited September 19, 2016 by CriticalTug Google™ and Budbringer like this Link to comment
they took the name eazy Posted September 19, 2016 Report Share Posted September 19, 2016 1 minute ago, OLL13 said: This is a sign from the gods to covert back to 6.0!!! Go practice shogun 2 ollie u scrub Haych and Corgi like this Link to comment
JIMBO Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) Hey @bamf, you should take a look at crafting / discounts. We just crafted a bunch of MK1's and the price was increasing everytime we crafted, I'll upload pictures/video soon. Edited September 19, 2016 by Jimbo! Link to comment
Phil. Posted September 19, 2016 Report Share Posted September 19, 2016 1 hour ago, Gnashes said: Muahahaha. Look what I have created!!!! That's pretty awesome..lmao Link to comment
Haych Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) Just found a pretty weird bug that has a massive impact on weapon prices. So as I was crafting weapons, I told Jimbo and Tiger to keep a eye on the Rebel store. We noticed that the more I crafted, the more expensive a gun got. The MK1 Sand which had around 5 on the Market was much cheaper than the other Camo's which had 14+ on the market. I carried on crafting the one with 14, and the price just increased more and more, the more I crafted. @bamf @speed @Gnashes Here's Jimbo's recording while it was happening Edited September 19, 2016 by Haych Link to comment
.Sean Posted September 19, 2016 Report Share Posted September 19, 2016 9 hours ago, Gnashes said: Right; but not everything in the mission file is being evaluated 100% of the time. Skiptracer is only called when a player actually does something there. Otherwise it's being used a much as your toilet when you don't need to take a shit. Now I understand, maybe try removing day night cycle see if it helps? Link to comment
Paratus Posted September 19, 2016 Report Share Posted September 19, 2016 35 minutes ago, House. said: Now I understand, maybe try removing day night cycle see if it helps? Already did that last night. .Sean likes this Link to comment
Paratus Posted September 19, 2016 Report Share Posted September 19, 2016 40 minutes ago, Haych said: Just found a pretty weird bug that has a massive impact on weapon prices. So as I was crafting weapons, I told Jimbo and Tiger to keep a eye on the Rebel store. We noticed that the more I crafted, the more expensive a gun got. The MK1 Sand which had around 5 on the Market was much cheaper than the other Camo's which had 14+ on the market. I carried on crafting the one with 14, and the price just increased more and more, the more I crafted. @bamf @speed @Gnashes Here's Jimbo's recording while it was happening Thank you for the heads up on this. That's crazy. Will look into it today! JIMBO likes this Link to comment
Haych Posted September 19, 2016 Report Share Posted September 19, 2016 3 minutes ago, Paratus said: Thank you for the heads up on this. That's crazy. Will look into it today! So after heading back to the Rebel store, the prices where fixed and working like intended. It was just weird at that moment the prices where going up as we monitored them, however the issue is no longer there when we headed back to rebel an hour later. Link to comment
Paratus Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, Haych said: So after heading back to the Rebel store, the prices where fixed and working like intended. It was just weird at that moment the prices where going up as we monitored them, however the issue is no longer there when we headed back to rebel an hour later. Ah ok, good to know. We were having issues with the database being DDoS'd which was causing all kinds of issues. It's better protected now (not on an IP that's connected directly to the internet) so it's possible that the server failed to pull tax data and went crazy. Haych likes this Link to comment
.Sean Posted September 19, 2016 Report Share Posted September 19, 2016 Finally cloud flare was added lol Link to comment
Heidelberg Posted September 19, 2016 Report Share Posted September 19, 2016 I see that the forums also had a DDOS protection upgrade, or is it just me? Forums have been on and off recently. Link to comment
Haych Posted September 19, 2016 Report Share Posted September 19, 2016 5 minutes ago, Paratus said: Ah ok, good to know. We were having issues with the database being DDoS'd which was causing all kinds of issues. It's better protected now (not on an IP that's connected directly to the internet) so it's possible that the server failed to pull tax data and went crazy. Yeah, it's not the first time i've experienced weapon pricing going up when capping something like a Turf. Good to know it's not a permanent issue Link to comment
bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 Were there any mass kicks last night? I played on S3 for a whole server cycle and did not see any. The server was 90+ the whole time as well. ChicoXL likes this Link to comment
The Orphan Posted September 19, 2016 Report Share Posted September 19, 2016 Just now, bamf said: Were there any mass kicks last night? I played on S3 for a whole server cycle and did not see any. The server was 90+ the whole time as well. I played server 1 and never had any mass kicks last night Link to comment
bamf Posted September 19, 2016 Author Report Share Posted September 19, 2016 17 minutes ago, Tom said: S1 and S2 had mass kicks last night around 10:00 - 10:30ish PM EST. That one was the memory allocator crashing. So I can at least point to the issue there. Haych, .Sean and 544heu like this Link to comment
ViiET Posted September 19, 2016 Report Share Posted September 19, 2016 3 minutes ago, bamf said: That one was the memory allocator crashing. So I can at least point to the issue there. I love you .Sean likes this Link to comment
Jesse Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) 12 hours ago, Gnashes said: Another thing there is that Bamf, Speed, and myself are usually working on new things / changes / bug fixes, but aren't necessarily the best people when it comes to optimization. (I'll admit I know absolutely fuckall about optimization; I'm just persistent enough to make something work.) The entire 7.5.5 series of hotfixes (the 6? of them) were Bamf and Myself just going ham on a lot of bugs and small changes that needed to be made. https://community.bistudio.com/wiki/Code_Optimisation Now you can know something about optimization/optimisation(cringe). Also, just looked at the v7.5.4 code for your addAction "Dictionary" You guys use very little lazy eval... Might wanna throw some in, could help potentially. Not sure as to the extent of "how much" it would help though... Maybe even @TheRealLethal would do it for you since ya know it's a large concern of his Quote You may expect the engine to stop reading the condition after the group has no knowledge about the object but that's false. The engine will continue evaluating the condition until the end even if any of the previous conditions evaluated false. ^^ per the optimization link posted above (near the "Fewer statements => faster code") Edited September 19, 2016 by Jesse CriticalTug, Haych and TheRealLethal like this Link to comment
Destrah Posted September 19, 2016 Report Share Posted September 19, 2016 (edited) On some drug runner runs I notice if you land at the same drug runners but in a different order it gives a different payment. Is that how its meant to be? On my google sheet run 14 and 15 for Altis are the same run but reversed and they paid differently. Same for Tanoa runs 3 and 4. @bamf @Paratus Edited September 19, 2016 by Destrah Link to comment
Vhantom Posted September 19, 2016 Report Share Posted September 19, 2016 @bamf Can we expect the return on the Asylum Exchange anytime soon? I personally didn't experience any issues from it and its removal hasn't seemed to improve performance. It is an amazing feature and I have things to be claimed from it. Rag likes this Link to comment
Phil. Posted September 19, 2016 Report Share Posted September 19, 2016 8 minutes ago, Gnashes said: Paratus changed some things regarding payouts this morning. It looks like some runs may not have been calculating 100% correctly. Did he change the time at all? Link to comment
Recommended Posts