Maniaplanet Forum

As posted already here: http://forum.maniaplanet.com/viewtopic. ... 74#p158374

The event TrackMania.PlayerFinish is also fired twice when the maps is loaded after a map change, on respawn or finish it is fired only once.

Still not fixed.

The server has still a memory leak: http://i.imgur.com/99GJz1h.jpg

I updated the post here with the official bug report on this: http://forum.maniaplanet.com/viewtopic. ... 0&p=154091

It is much better than before, but the server is still leaking memory as you can see on the image I posted at the top.

I can confirm that for TrackMania² Stadium and ShootMania. Please tell us, if you can confirm the memory issue.

Cheers,
Bueddl

Bueddl wrote:The server has still a memory leak: http://i.imgur.com/99GJz1h.jpg

I updated the post here with the official bug report on this: http://forum.maniaplanet.com/viewtopic. ... 0&p=154091

It is much better than before, but the server is still leaking memory as you can see on the image I posted at the top.

Cheers,
Bueddl

I can confirm that, in Royal veeeeery slow, but still leaking.

Still a bit of a prob with events in TM.. with the newest API (2012-06-19), works almost perfectly -- but I'm not seeing any BeginRound/EndRound events. In the last API (2011-10-06), it'd be perfect -- but it throws EndMap along with EndMatch at the podium, instead of waiting until the exit status. (not even going to test the '2011-08-01' api anymore.. too many 'overlaps', it'd be too hard to bother coding for)

Still.. new server with the '2012-06-19' API yields the best results so far.. As functional as the 2013-02-27 server with the same API -- but now I can code for callvotes again, hopefully!

Edit: btw.. about the above comparison, both servers yield the same results with the exception of the change to the BeginMap timing (the 2013-04-09 and 2013-04-11 servers.. that's where I had the huge problems with EndMatch timing).

Bueddl wrote:

slugish wrote:I am assuming this update only fixed the memory issues with Trackmania? Still seeing the same issues with Shootmania.

Can you be more specific with that? Post some details etc.

I didn't provide any details because nothing appears to have changed.

I run two ellite servers on a VPS with Ubuntu 12.04. Each server starts with roughly 95MB ram usage. After cycling through the entire maplist, the ram usage grows to (estimating here) 125-150MB. The higher the ram usage climbs, the slower the inflation, but it does continue to grow. One of these servers is currently sitting empty with 207MB ram usage after having been online for 37 hours. Previously, I have had to restart the server after 3-4 days after ram usage climbs to 350MB or so due to limited ram on my setup. I'm not sure how much higher the usage will climb now, perhaps this has improved... but one of my servers is already using more than double the ram it was using initially and the other is not far behind. Maybe things have been improved, but something is grabbing memory and refusing to let it go.

slugish wrote:

Bueddl wrote:

slugish wrote:I am assuming this update only fixed the memory issues with Trackmania? Still seeing the same issues with Shootmania.

Can you be more specific with that? Post some details etc.

I didn't provide any details because nothing appears to have changed.

I run two ellite servers on a VPS with Ubuntu 12.04. Each server starts with roughly 95MB ram usage. After cycling through the entire maplist, the ram usage grows to (estimating here) 125-150MB. The higher the ram usage climbs, the slower the inflation, but it does continue to grow. One of these servers is currently sitting empty with 207MB ram usage after having been online for 37 hours. Previously, I have had to restart the server after 3-4 days after ram usage climbs to 350MB or so due to limited ram on my setup. I'm not sure how much higher the usage will climb now, perhaps this has improved... but one of my servers is already using more than double the ram it was using initially and the other is not far behind. Maybe things have been improved, but something is grabbing memory and refusing to let it go.

That's not at all bad.. especially considering it's an FPS game server. The 'real' mem leak problems, the server would continuously eat up memory far beyond a few gigabytes.. A few hundred MB, count yourself well within the realm of reason.

Really, though.. the true test would be how a server manages after a few weeks of uptime, not just a day or two.. Back in TMF, we could leave a server running indefinitely for months at a time. Now, when we have to monitor it's memory consumption day to day -- that's a real problem.

nocturne wrote: Really, though.. the true test would be how a server manages after a few weeks of uptime, not just a day or two.. Back in TMF, we could leave a server running indefinitely for months at a time. Now, when we have to monitor it's memory consumption day to day -- that's a real problem.

Aggree here 100%

TMF did rock that! I think the ram usage cames mostly from the dedicated.pak ( ~300MB as a file) in the packs folder of the server. It is temporarily partially unpacked into the memory, which would cause the high (350-400MB) memory from startup.
However in TM² (at least Stadium, no details about Canyon here) we can say that there is still something not correctly deleted out of the memory.

nocturne wrote:Back in TMF, we could leave a server running indefinitely for months at a time.

No, just a month and a half. The uint32_t timer bug meant that a restart was still needed after 49.7 days.

But indeed, that's still a lot better than now with the memory leaks.

Xymph wrote:

nocturne wrote:Back in TMF, we could leave a server running indefinitely for months at a time.

No, just a month and a half. The uint32_t timer bug meant that a restart was still needed after 49.7 days.

But indeed, that's still a lot better than now with the memory leaks.

Had never that short uptime in TMF, had run the servers for 3 month or longer without restart

49.7 days means a overflow in uint32_t with milliseconds stored... Maybe it had been fixed that time^^

Just some bugs ?

Podium not working anymore.

Replay plugin also not working, track get played after 2e track.
And some how it does not register the dedi times when track played 1e time.

Example :

track 1 played.
player pays for replay track 1.
now track 2 gets played.
after track 2 track one gets played.
but deditimes are gone.



Host updated the server to the newest version, could it be caused by new dedicated server ?