Ok, well please let us know ASAP if 1) the leak is fixed in non-threaded mode and 2) if threaded-mode works properly (or better) now. This goes for anyone helping to test!
1. I am currently running the game, only using 1.6% of my 16GB of total RAM, seems good, but we'll have to wait a few days until I can really see any difference.

2. Windows servers crash after it reboots and then initializes, unsure if that's replicate-able. Other than that, I've had some trouble with the server, which is most likely related to my own code, and for stability, I disabled threading for the time being.
In response to Laser50
Thanks for the update. If you get another crash on Windows, please get the crash details if you can (module and offset of crash). Knowing where it's crashing could help. I haven't had reboot issues in testing. Also, was that crash with threads on or off?
I'd like to report that I -also- crash on round restart in 1236. Also, 1236 is the FIRST version that has allowed me to use thread AND/OR mapthreads on my Windows host. Since I run the most popular public SS13 server, I'm very happy to have finally gotten threading runnong, but I am getting an inordinate amount of crashing during round changes.
I'm glad to see more testing of this on larger games. The first thing I'd like to know is if you run this with "threads on", but "map-threads off", if it is stable. That would isolate the crashes to the map-threads, and that's something Lummox JR can test with you next week.
Map-threads off, regular threads on has proven to be successful in the last 3 restarts I did just now.

Any way, I've patched up my server (Reverted to stable until I did), so I've gone back to the beta build to help out with some more testing.

Percentages report 176MB of RAM currently in use, we'll see how that is at the end of the day.
Pardon, to correct myself there, server crashed with map-threads off, normal threads on, no core file, no logs. I've enabled the -core parameter, should've done that, any way.

Will try with map-threads on and normal threads off. See if there is a difference.
From what I understand if normal threads are off, it AUTOMATICALLY means map-threads are ALSO off.
In response to ZomgPonies
ZomgPonies wrote:
From what I understand if normal threads are off, it AUTOMATICALLY means map-threads are ALSO off.

You are correct.
That's correct about threads off. I'm disturbed that it is still having issues with map threads off.

With threads off entirely, is it stable? It should be.
The original CPU leak issue seems to have gone away. No crashes since updating to the beta build.
Are you hosting with threads, MisterPerson?
<img src='http://i.imgur.com/huyl2cW.png
'>
These are my results after a day has passed. Not much RAM issues. The CPU is still crazy.. But that's just their games. :/

Threads:
The first one has both threads on.
The second one has threads off (but map threads on even though it still does nothing).
3rd has both on.
4th has both.
5th has both.
6th has none on.
7th has both.
8th has both.
9th has both.
10th has both.
11th has none.
12th has none.
13th has both.
14th has threads on but map threads off.
With threads off, I haven't had a crash all day.

With threads on (map threads on/off doesn't matter), I crash AT least at the end of every round, and sometimes mid-round.
Ok that is very interesting and probably easier to isolate.
In response to Tom
Our host tried running with threads on and I'm guessing DD froze midway through running the dmb. Don't know if map threading works or not. Sorry, I don't have a ton of info here and I know you guys are trying hard. I keep asking people to run experiments with threading and nobody does.
To come back on the memory problems, server has been running for about 6 hours so far, sitting on a good 225 MB. Which probably means that the issue is solved. Good job, guys!
In response to Laser50
Yeah, mine are still the same as well.
Thanks to this update I have saved a good 2GB of RAM that was being sucked up for no reason by multiple games! Thank you!
Eternia's CPU is still increasing over time after updating to the latest build. This is with threads off.

Going to see if there's a memory increase and will update accordingly.


Yesterday morning:




This morning:




The increase seems kind of minor compared to the actual CPU increase, which is about 2x if not more after 24 hours. Could they be separate issues, or is there more that I'm unaware of?
Page: 1 2 3 4