Servers crashed.

[ENO]

No problem Blaze I think we got it sorted! There was a bit of delay between a couple of the crashes since I was transitioning between locations.

[ENO]

There are too many individual functional issues to count which could be causing crashes. I've yet to see someone positively ID one single issue that they can reliably recreate a crash with. I will however say that the more users on the server, the more it appears to crash and this would point towards windows session handling and the way the server handles user threads. C++ applications can have issues with memory allocation and freeing memory up if developers do not de-allocate all the memory they assign's. Having run data collectors on the s77th server and observed whats looks like a memory leak, I suspect this is the problem causing most of the instability. I get the impression ED have not quite grasped this problem, until they do, it may change a little but not much.

 

I wanted to address this also, looking back. I watched the ram use as well as recorded with the performance monitor- and throughout the crashcapade there was no fluctuation in memory, network, or CPU.

Edited February 1, 2014 by ENO

[Frostie]

1 more try with gun at the start this time?

 

If people gun running and crashing (never done that before personally :music_whistling: ) causes it- it makes sense. That stackhash error is very familiar.

We only ever get a stackhash error on the 51st server on odd occasions immediately when we connect by remote, for some reason that crashes the server and gives the stackhash error. Could be something to do with selecting the DCS window.

[ENO]

I've had 6-8 in the past few days where I was either also playing or already watching when it crashed- granted a couple (few) of them were forced by mustang. I would not be inclined to limit the cause to only access by remote but it's clearly a factor.

 

Now that you mention it though I remember I used to crash a lot more using logmein- which remote software do you guys use?

[Frostie]

We use Radmin.

 

We only ever get stackhash when logging in through radmin and the crash is immediate, this is only when the server has been running for some time, 4+ hours.

Weapon.dll seems to be usually trigger related.

DCS.exe is the usual.

Edited February 1, 2014 by Frostie

[MadCow]

Wasn't there a patch a little bit ago to fix significant slowdown when firing and the spent casings (particularly from the mustang) were causing significant slow downs in performance? Is it possible that while the patch fixed the slowdowns via what's being rendered, it may not have fixed the game recognizing their existence, and therefore the more you use the gun, the more memory you'll take?

[ENO]

Wasn't there a patch a little bit ago to fix significant slowdown when firing and the spent casings (particularly from the mustang) were causing significant slow downs in performance? Is it possible that while the patch fixed the slowdowns via what's being rendered, it may not have fixed the game recognizing their existence, and therefore the more you use the gun, the more memory you'll take?

 

 

I wanted to address this also, looking back. I watched the ram use as well as recorded with the performance monitor- and throughout the crashcapade there was no fluctuation in memory, network, or CPU.

 

Wasn't a factor in this case.

[Pman]

I log in via RDP, our server runs Windows Server 2012

 

Pman