• Dear forum reader,
    To actively participate in our forum discussions or to start your own threads, in addition to your game account you need a forum account. You can
    REGISTER HERE!
    Please ensure a translation in to English is provided if your post is not in English and to respect your fellow players when posting.

Cannot Reproduce: Memory leak-o-rama

Andi47

Overlord
The behaviour towards (excessiely increasing) memory useage has been improved a little (!) bit since the last few attempts to fix memory leaks, but it seems that the game is still full of memory leaks.

Reproduction Steps:
1.) ermmmm. play and keep an eye at the memory useage (windows task manager).

ok, seriously and a bit more detailed:
* Memory useage seems to increase (and never decreases - I have the impression, that unused memory is almost never purged) every time when you do something else than just having the city open. (ok, and I am not sure if the memory leak when having the city open and doing nothing for a a few hours is completele fixed, although this one has been largely improved with the fix in autumn(?).). So try to do (and keep an eye at the memory useage) any of these steps (numbering is just for identifying the steps, you can do any of these alone or in any order you want):
1.) open and close army management
2.) go to research tree and back to city (at least a small memory leak here)
3.) go to continent map (fight if you want - or just take a look) and go back to city (I can't test fighting right now - my scout takes still one day to reach the next sector)
4.) go to GvG map, click through the maps of the ereas (from iron age to tomorrow in any order you want). Do this repeatedly to check whether one of your guild's sector is under attack, and keep an eye on memory useage (memory leak definitely still present with current version on beta, I just checked a few minutes ago). Note: Memory useage was increasing during clicking through the eras, and there was a significant increase when I went back to city
5.) collect and restart production of buildings (just tested with a robot factory)
6.) open and close event history (jsut tested - memory use increases every time when you open and close it from townhall)
7.) open and close global ranking (memory use increases every time when you open and close it)

Issue Description (x/10): (12/10 - or so....)

Technical info:
1.47.29968 (12.03.2015 10:20), Andi47 (73349), zz1, en_US, WIN 17,0,0,134, Windows 7, Firefox/36.0, 1920x939, DirectX9Ex (336 MB VRAM)

I have performed a quicksearch of the forums using a select few keywords relating to my bug to see if it has already been reported: Hmmm... maybe these long-standing memory-leaks have been reported (without the details given above) throughout the servers, including the "performance unter aller S..." thread on .de
 
Last edited:

Amy Steele

Regent
Hi Andi47,

Thanks for reporting this issue in such detail. Firstly, the changes that were made last year (in the Changelog 1.39 Announcement) apply only to visiting other players cities, to polish, motivate, attack or plunder. My understanding of what happens is that any new elements in those cities are temporarily cached. This cached data is cleared when you do any of the following:

  • return to your own city
  • visit the next player to pol/mot
  • pvp (where returning to the city in between battles acts like visiting another city)
  • Reload the game
  • Log out

Cached data is not cleared in any other circumstances, such as when you visit several provinces/sectors on the gvg map consecutively. Additionally, when you return to your city, only city related cached data, such as building graphics, is cleared. Any other cached data will only be cleared when you reload the game or log out.
 

Andi47

Overlord
I think, at least these are true memory leaks, because memory usage increases every time when you repeat the action, so I disagree with "not a bug" in the following cases:

ow - my scout takes still one day to reach the next sector)
- go to GvG map, click through the maps of the ereas (from iron age to tomorrow in any order you want). Do this repeatedly to check whether one of your guild's sector is under attack, and keep an eye on memory useage (memory leak definitely still present with current version on beta, I just checked a few minutes ago). Note: Memory useage was increasing during clicking through the eras, and there was a significant increase when I went back to city
- open and close event history (jsut tested - memory use increases every time when you open and close it from townhall)
- open and close global ranking (memory use increases every time when you open and close it)

If it is just normal caching, I think memory usage should increase the first time, and than stay more or less the same when you repeat the action.

BTW: The first of the issues mentioned in this posting eventually leads to freezing during loading of the GvG era maps. Just try to repeat loading the era maps one after the other and returning to overview 3 or 4 times (maybe it is necessary to return to city somewhere in between).

Steps to reproduce the freeze:
1.) go to GvG and load e.g. iron age, then go back to overview
2.) load FMA map, go back to overview.
3.) etc..... - go to all the other eras and go back to overview

4.) go back to city if you want....
6.) repeat steps 1 to 3 - do this a few times if necessary (if you are a hostile attack every moment you are doing exactly this a few times to check your sectors...)

---> at some point the game freezes during loading one of the era maps and has to be reloaded.
 

Andi47

Overlord
Are you really sure that this behaviour is 100% caused by caching? See also here: http://forum.beta.forgeofempires.com/showthread.php?4923-The-plugin-container-file-(Waterfox)

I think that at least issues 4, (5), 6, 7 (see first posting of the thread) are not only caused by caching, but also by memory leaks (or do you need double, triple, quadruple, quintuple, .... copies of the same items in the cache when you repeat these steps twice, or 3, 4, 5,... times??). Issue 4 even causes a CRASH if done repeatedly (for patrouilling all the ages when you expect attacks of other guilds) - so causing a crash is a feature??

Edit: According to the experients of magnus555 (http://forum.beta.forgeofempires.co...n-container-file-(Waterfox)&p=23312#post23312) also issue 3 is affected by at least one true memory/cache file leak

Note: Changelog 1.48 says that memory leaks with battles should be fixed now. But there is still at least one with GvG, which STILL causes a CRASH if you are doing ~20 consecutive battles in GvG.
 
Last edited:

DeletedUser5800

Guest
There is obviously a problem as many of us experience it and I do not think allocating this as 'NOT A BUG' helps.

I think the 'Not a bug' and 'will not be fixed' need to be separated. Memory leaks have been a problem for a long time and to be fair the developers have been working to solve them and also to minimise the reason for them by introducing AID so we don't actually load so many cities.

However, memory leaks are still a serious problem with the running of the game and the animations for some reason cause some issues. So instead of trying to hide it by stating it isn't a bug face the fact there is a problem and it needs further work to resolve.
 

DeletedUser

Guest
@mods: Are you still reading here?

Yes we are, we always are.

Personally I have been waiting for everyone to offer their feedback before posting anything myself.

I don't want to get into a discussion on the broad aspects of the cache handling of the game, as I don't see it being productive in any way at this time. What I would like to comment on is this specific issue in relation to what we consider is or is not a bug.

So what is a bug? Simply put it is unexpected or unwanted behavior in the game that has not only been reported, but has also been consistently reproduced under the exact same conditions. This does not mean that something is not a bug if we only reproduced similar results 20% of the time. But if we can't produce consistent results, then we have not yet identified the exact trigger, and until we can, we cannot forward the issue as a bug. The devs can't fix things if we can't even pinpoint the cause.

Now with that said, the issue of cache handling and memory leaks is a very broad, and complex thing to confirm, and one of the reasons is the sheer scope of variables that can affect the results when we attempt to reproduce even parts of it.
We understand that some players have done some very in-depth testing on their own, and all of that information is very helpful as we try to gather enough data to see a larger pattern.
Unfortunately, just because a player has produced some consistent results of their own, with issues relating to cache and memory, this does not mean that any of the support staff will necessarily get the same results, in fact, I personally have not gotten the same results.
Does that mean I don't believe this issue is having significant impact on some players? Absolutely not. It only means on my system, with my given set of variables, I can't reproduce what someone else has on their system, with their own unique set of variables.

Now I realize that many of you are not going to like this post, and I understand that. Believe me I would like nothing more than to tell you all about the piles of data we have to support any type of conclusion on this issue, but the fact is that we just do not have that conclusive data at this time. This in no way means we have given up on testing this, and trying to finally determine the exact trigger or triggers that are influencing these outcomes, it just means we can't call it a bug yet.

So I propose we work together to try to get more consistent results.
It is clear that many of you feel this is very important and needs to be addressed, and many of you have extensive technical knowledge that could be invaluable to us in this process.
So if any of you really want to get involved and become part of the process that we use to determine consistent results, then I encourage you to send me a message here on the forum. I will then take the list of players interested in assisting us and discuss it with the rest of the Beta Team, and we will find a way to integrate you into a team effort as we try to determine if there is indeed something here that can be corrected in the coding, or if it is still something that is happening in such random ways that no single coding solution will resolve this for everyone.

All of us on the Beta Team appreciate all of your input, but if we really want to make this productive, then we need to work together to try and be part of a solution. Otherwise we are just making points, and counterpoints, that are not really getting any of us closer to a desired outcome.

As always, myself and the rest of the team appreciate everyone's efforts and feedback.

Keep up the great work Beta :)
Zarok Dai
 

DeletedUser5149

Guest
I have been following this thread since it started, and I would like to give my experience. Battling in GvG 16 times always results in the loss of audio. Four battles more, and I get the black screen, and have to reload the browser (latest chrome with the latest version of flash). This has been happening as long as I can remember. When we fight as a guild in GvG, all of us there experience the exact same thing, with several different browsers be used. In addition, when I look in "task manager" under the "processes" tab, memory usage of chrome (FOE) starts somewhere around 700,000k, and then grows to 2,000,000k or more, which starts to affect performance. Also, I used to experience this same thing when mot/pol-ing a large amount of players, but since the introduction of the aid button, the problem has gone away, probably because aiding does not require the constant loading of cities.

So as far as being reproducible, I am positive it is, and very consistently. As far as it being classified a bug, all I can say is that in my experience, this behavior is unique to FOE (I don't play any other large browser-based games, however).
 

Andi47

Overlord
@Zarok Dai:
Thanks for your posting.

So what is a bug? Simply put it is unexpected or unwanted behavior in the game that has not only been reported, but has also been consistently reproduced under the exact same conditions. This does not mean that something is not a bug if we only reproduced similar results 20% of the time. But if we can't produce consistent results, then we have not yet identified the exact trigger, and until we can, we cannot forward the issue as a bug. The devs can't fix things if we can't even pinpoint the cause.

So if I understand correctly, according to your definition, any unexpected / unwanted behaviour, which can be consistently reproduced under exact the same conditions is a bug. And from what you said (quoted above) I would derive, that a "confirmed bug", which is ready to be reported to the devs, is an unwanted behaviour, that can be consistently reproduced under known conditions.

So what is "not a bug"? I would say, an issue is not a bug, if
* either the observed behaviour is wanted / intended (or at least not unwanted)
* or it can be proven (...or there is strong evidence) that there is no such issue (there are cases where it turns out to be a mistake made by the user or a neutrino hit the CPU or whatever)

What do we have here:
* there are issues with caching / memory useage, which - under certain circumstances even lead to a CRASH.
* many players are experiencing this issue, and it is at least annoying enough (i.e. affecting performance and stability(!) that many players consider it an important issue)
* as you said, at the moment there are too many variables so the issues have not yet been tracked down.

This means: There is an unwanted issue which affects many players, but the circumstances have not yet been tracked down.

According to the above definitions, it might be one or more bugs, but it is not a confirmed bug. But it is also not a "not a bug".

As you said, there might be much work to do, to track it down, so why not setting the issue to "investigating"?

I have been following this thread since it started, and I would like to give my experience. Battling in GvG 16 times always results in the loss of audio. Four battles more, and I get the black screen, and have to reload the browser (latest chrome with the latest version of flash). This has been happening as long as I can remember. When we fight as a guild in GvG, all of us there experience the exact same thing, with several different browsers be used. In addition, when I look in "task manager" under the "processes" tab, memory usage of chrome (FOE) starts somewhere around 700,000k, and then grows to 2,000,000k or more, which starts to affect performance. Also, I used to experience this same thing when mot/pol-ing a large amount of players, but since the introduction of the aid button, the problem has gone away, probably because aiding does not require the constant loading of cities.

So as far as being reproducible, I am positive it is, and very consistently. As far as it being classified a bug, all I can say is that in my experience, this behavior is unique to FOE (I don't play any other large browser-based games, however).

I am experiencing the same with firefox as long as I can remember (i.e. all versions of FoE, firefox and flash since GvG has gone live!). I did not count the battles, but for me the issue is also 100% reproducible: After ~15 battles (no matter whether you fight or surrender) there is no more audio, and after 2-3 more battles I get a blackscreen (or the GvG map doesn't fully load and freezes) and I have to reload. This ALSO happens with FoE version 1.48 despite the changelog says that the issues with memory leaks in batteling should be solved.

1.48.30412 (07.04.2015 10:53), Andi47 (868769), de5, de_DE, WIN 17,0,0,134, Windows 7, Firefox/37.0, 1920x939, DirectX9Ex (336 MB VRAM)
 
Last edited:

DeletedUser5800

Guest
Yes we are, we always are.

Personally I have been waiting for everyone to offer their feedback before posting anything myself.

I don't want to get into a discussion on the broad aspects of the cache handling of the game, as I don't see it being productive in any way at this time. What I would like to comment on is this specific issue in relation to what we consider is or is not a bug.

So what is a bug? Simply put it is unexpected or unwanted behavior in the game that has not only been reported, but has also been consistently reproduced under the exact same conditions. This does not mean that something is not a bug if we only reproduced similar results 20% of the time. But if we can't produce consistent results, then we have not yet identified the exact trigger, and until we can, we cannot forward the issue as a bug. The devs can't fix things if we can't even pinpoint the cause.

Now with that said, the issue of cache handling and memory leaks is a very broad, and complex thing to confirm, and one of the reasons is the sheer scope of variables that can affect the results when we attempt to reproduce even parts of it.
We understand that some players have done some very in-depth testing on their own, and all of that information is very helpful as we try to gather enough data to see a larger pattern.
Unfortunately, just because a player has produced some consistent results of their own, with issues relating to cache and memory, this does not mean that any of the support staff will necessarily get the same results, in fact, I personally have not gotten the same results.
Does that mean I don't believe this issue is having significant impact on some players? Absolutely not. It only means on my system, with my given set of variables, I can't reproduce what someone else has on their system, with their own unique set of variables.

Now I realize that many of you are not going to like this post, and I understand that. Believe me I would like nothing more than to tell you all about the piles of data we have to support any type of conclusion on this issue, but the fact is that we just do not have that conclusive data at this time. This in no way means we have given up on testing this, and trying to finally determine the exact trigger or triggers that are influencing these outcomes, it just means we can't call it a bug yet.

So I propose we work together to try to get more consistent results.
It is clear that many of you feel this is very important and needs to be addressed, and many of you have extensive technical knowledge that could be invaluable to us in this process.
So if any of you really want to get involved and become part of the process that we use to determine consistent results, then I encourage you to send me a message here on the forum. I will then take the list of players interested in assisting us and discuss it with the rest of the Beta Team, and we will find a way to integrate you into a team effort as we try to determine if there is indeed something here that can be corrected in the coding, or if it is still something that is happening in such random ways that no single coding solution will resolve this for everyone.

All of us on the Beta Team appreciate all of your input, but if we really want to make this productive, then we need to work together to try and be part of a solution. Otherwise we are just making points, and counterpoints, that are not really getting any of us closer to a desired outcome.

As always, myself and the rest of the team appreciate everyone's efforts and feedback.

Keep up the great work Beta :)
Zarok Dai

Zarok Dai I appreciate the sentiments behind your post but this is an issue that has been reported a number of times across many servers. Memory leaks need the developers to track down as it is slightly unique to each computer and the way a player plays the game. It can depend on the specification of the computer, how long they are logged in and what they are doing while logged in. What can be monitored is the increase in memory usage in task manager. They already know animations and fighting causes problems.

So, the onus is on the developers to find and cure it. We do not work for Innogames, we are here to report the problems and there is a problem as well you know because its not just in beta it is being reported.

So, time to cut to the chase and just acknowledge there is a problem and get on with solving. Denial, however nicely you want to portray it with your post doesn't help anyone. It just cause frustration for players and reflects poorly on the developers.

So stand up and say, "we have a memory leak issue, it has been happening for years, we will find it and cure it".

and get on with it......
 
Last edited by a moderator:

DeletedUser

Guest
I have not suggested there is not an issue affecting many players. Nor have I implied in any way that this does not need to be investigated further, so solid data can be passed on to the dev team.
But that's the real issue here, what would you have me pass on to them other than 'cache issues and memory leaks'? Which I don't personally think is going to produce any type of resolution for anyone.

I have read what some of you have reported, the problem is even I can't duplicate them. I am aware we will never identify everything relating to this issue, but we have to start with some baseline information to work off of.

My experience in GvG is not what is occurring with some of you. I do not experience loss of sound, I also don't crash after 15 to 20 battles, or 30 for that matter. That doesn't mean I don't believe that this is happening to some of you, I know it is or you wouldn't be taking the time to report it. Believe me I would like nothing more than to have at least some solid data to pass on to the devs, and as soon as I do I will.

I would also like to say that in my previous post, I offered anyone interested the chance to work with myself and the rest of the team to try and track down some of this information we need to effectively report it. And I think within the scope of my position I am showing nothing but a willingness to keep working on this. I would also assume that the support staff on the other live servers that some of you have referenced have not made this offer, or many of you would already be involved in helping those teams.

So I am here, I am listening, and I am willing to do anything I possibly can. And even after making my previous post, and offering any of you interested the chance to get more involved, I have so far received zero messages. So that means myself and the team will keep trying to test and reproduce things, in between all of our other duties. But I don't know what else I can do. There are only so many hours in the day, and a limited number of us to start really testing and retesting different scenarios until we can at the very least establish some type of pattern to give the devs.
This is not me denying there is a problem, this is simply me trying to convey to all of you what is needed to take this to the next level.

Zarok Dai I appreciate the sentiments behind your post but this is an issue that has been reported a number of times across many servers.

You are correct, this has been reported across many servers, many times. Yet there is still no clear pattern or reproducible data to go off of, which may be why there has been no resolution. We are going to need much more than what has been gathered so far, and I am trying very hard to work out a way to achieve this goal.

I realize you want to be heard, and you want the devs to do something. For myself and the team to be your advocates, we need more information, this is just a fact.

Zarok Dai
 

Andi47

Overlord
I have not suggested there is not an issue affecting many players. Nor have I implied in any way that this does not need to be investigated further, so solid data can be passed on to the dev team.

The thread is tagged with "[Not a bug]". It has been moved to the subforum "not a bug / will not fix". Both circumstances (seem to) imply that the issue is no longer investigated.
 

DeletedUser

Guest
I have moved this thread back to bugs and marked it as 'Investigating'.
 

Andi47

Overlord
I just triggered the "sound fails and it eventually crashes" bug, which I reported in another thread for GvG, on the "tomorrow" campaign map on .de8:

1.) I was at the point that I had conquered the first province (the island) and I had scouted the adjacent three provinces (accelerated with diamonds) and had the quest, which asks to build a military building of the current era or research a tech, finished.
2.) reload
3.) conquer one province after the other (I conquered all sectors of Villariba, insta-sent the scout to Bajatai with 50 diamonds and conquered 4 or 5 sectors of Bajatai) without reloading in between, and fullfilled the infiltrating and conquering quests if applicable, at the same time. (I infiltrated more sectors than I conquered, due to the quests)

Observed: Memory use of the Flash player successively increased. After several battles, sound failed. (no more driving and shooting sound) (I did not continue to see if it crashes)
expected: Memory use should stay more or less stable (I can't imagine that ~800 MB to 1GB (!) memory would be required for caching??), sound should be available even after doing 100 battles ;)

1.49.30799 (21.04.2015 9:35), Andi47 (868769), de8, de_DE, WIN 17,0,0,169, Windows 7, Firefox/37.0, 1920x939, DirectX9Ex (336 MB VRAM)
Note: I am using Windows 7 professional 64 bit, Firefox and an NVidia graphics card
 

DeletedUser

Guest
I did notice an improvement to the game loading the city, after visiting other areas of the game, when the change to the coin, supplies and goods graphics occurred. Maybe a thought on reducing all graphics in details would be a great plus in reducing the games memory usage. I myself have never been a huge fan of finely detailed graphics and like the content of games better than all the pretties.
Still I know developers get awards for such things and that is where it takes out, the real joy of playing (for me anyways) a game.
 

Andi47

Overlord
a few additional thoughts about this issue:
Is it possible, that the occurance of this issue depends on whether you have other browser tabs opened or not?

Normally (and so I did when reproducing the issue) I have three browser tabs open with FoE (de5, de8 and zz1), and maybe a few more with other sites (but no other online games).
 

DeletedUser

Guest
I would say having multiple tabs open will definitely have a negative impact on the overall performance.
 
Top