Author Topic: bridgewater-brunel.me.uk - Simutrans-Experimental - Pak128.Britain-Ex 0.9.1 (Read 185859 times)

prissi · « **Reply #665 on:** May 13, 2014, 08:45:08 AM »

Loading a game at log level 3 easily produces several MBs of logging information. The highest possible level for servers (at least not ultra high speed ones) was 2. Maybe James can tune this a little.

jamespetts · « **Reply #666 on:** May 13, 2014, 11:46:38 AM »

Once the desyncs problem is solved, I plan to turn down the logging again. Apologies everyone for the slowness.

Sarlock · « **Reply #667 on:** May 13, 2014, 02:19:57 PM »

The speed is tolerable... all in the name of bug squashing!

TurfIt · « **Reply #668 on:** May 13, 2014, 07:58:46 PM »

The loading speed of the server appears unchanged. Many of the excessive level 3 messages that show during loading were removed.
Now the file transfer speed, that's dropped. I'm seeing a very erratic transfer with speeds frequently dropping below 2Mbps. The changes to the logging have nothing to due with this.

ӔO · « **Reply #669 on:** May 13, 2014, 11:05:46 PM »

There is a train attempting to go the wrong way at Burnwell and it's caused a major traffic jam and pax pileup.

jamespetts · « **Reply #670 on:** May 14, 2014, 12:05:13 AM »

Quote from: TurfIt on May 13, 2014, 07:58:46 PM

The loading speed of the server appears unchanged. Many of the excessive level 3 messages that show during loading were removed.
Now the file transfer speed, that's dropped. I'm seeing a very erratic transfer with speeds frequently dropping below 2Mbps. The changes to the logging have nothing to due with this.

That is extremely odd. I have no idea what might cause this. My provider has unlimited bandwidth, and I do not think that anything in the network code has changed that might impact on the transfer speed itself. Can anyone who knows more about the intricacies of networking than I suggest what might be happening here and how to look into dealing with it? I am rather at a loss.

Sarlock · « **Reply #671 on:** May 14, 2014, 04:29:43 AM »

Seems okay to me right now... getting about 1.5-2.0 MB/s which is about normal for me.

VOLVO · « **Reply #672 on:** May 14, 2014, 07:32:36 AM »

Quote from: ӔO on May 13, 2014, 11:05:46 PM

There is a train attempting to go the wrong way at Burnwell and it's caused a major traffic jam and pax pileup.

A very funny prank from the train driver there.
I didn't even have replacing vehicle schedule on the line..
No reason for it to go backwards.

edit : Put it on the next stop, still wants to go backwards, you are fired!

jamespetts · « **Reply #673 on:** May 15, 2014, 11:40:03 PM »

I have just restarted the server with TurfIt's build of 11.31 to see whether this makes any difference to the desyncs - please report any difference in the frequency of desyncs from before the restart and any desync logs that you are able to muster.

VOLVO · « **Reply #674 on:** May 19, 2014, 10:51:00 AM »

I'm now in Hong Kong.

Very long loading time plus instant dissynchronise..

I have a 400MB Broadband here by the way.

Sarlock · « **Reply #675 on:** May 19, 2014, 06:29:02 PM »

It may not be you... I just had the same experience with several checklist mismatch desynchs instantly occurring right after connection. Check your logs to see if you are getting checklist desynchs or something else.

jamespetts · « **Reply #676 on:** May 19, 2014, 11:03:55 PM »

I have split the last two posts from the "How to get Simutrans-Experimental" thread and added them here since they appeared to be about this server rather than about how to download and install Simutrans-Experimental.

My apologies that you are having difficulties: I do not have time to look into it to-night, but please keep me updated with how things are going. TurfIt is in the process of writing an improved system for logging that should enable us to track down the issues rather better than is possible at present. An issue has already been traced to something relating to cities (we cannot be more specific yet), and Kevin/A-Train has identified a further possible issue relating to using the forest tool, although this is probably not the main cause of the problems.

Sarlock · « **Reply #677 on:** May 20, 2014, 01:49:28 AM »

I am hopeful that with the latest issue, you'll both be able to hone down on the core problems and address them-it seems much easier to recreate than previous intermittent issues, and hopefully a lot easier to zero in on the issue. The desynchs were nearly instantaneous.

Thanks for moving the thread.

jamespetts · « **Reply #678 on:** May 20, 2014, 10:04:14 AM »

What is interesting is that, having downloaded the server game locally, I cannot reproduce this with a server and client on my local system.

Sarlock · « **Reply #679 on:** May 20, 2014, 06:15:32 PM »

It really does point towards how different systems perform the math... could this be in the rand function itself? I don't know enough to be able to add much to the discussion and I'm sure you've both looked in many different places, but perhaps there is something in the rand function that causes the rare differential to occur and the sheer size of our map makes it statistically more likely to happen? When running a client/server on the same machine, you're getting the same method of computation occurring which makes the mismatch not happen.

When the checklist is made and compares rand, how does it account for temporary timing differences between server and client? ie: If the client is 1 second behind the server and perhaps a rand call that the server made 1 second ahead and has already made multiple calls later, how does the checklist reconcile this time differential? And how frequently is the rand function being called?

I wonder if a worthwhile test would be to just run the rand function a bunch of times on different machines known to cause mismatch and see if there are any variances.

I should get myself back in to coding, I haven't really done any for close to 20 years but I'm sure I could pick up C++ fairly quickly. (I used to do Pascal, some Perl (which has similarities to C) and some assembly). As always, it just comes down to finding the time in a busy life to devote to it

ӔO · « **Reply #680 on:** May 21, 2014, 12:13:09 AM »

The only thing I can think of is the server hardware is unstable and not calculating correctly, or the linux and windows builds calculates things differently.

---
to add, the server is currently extremely slow to send the save game over to me.
I think it is only doing about 500kb/s, which causes the transfer to take 4 minutes to complete.
---
Deep sea can still be edited in some directions, if the increase terrain height tool is used a certain level above sea level.
I think it was 3 high, then just drag left to right to increase water height.

AP · « **Reply #681 on:** May 21, 2014, 06:39:49 AM »

Quote from: ӔO on May 21, 2014, 12:13:09 AM

Deep sea can still be edited in some directions, if the increase terrain height tool is used a certain level above sea level.I think it was 3 high, then just drag left to right to increase water height.

This workaround has been around for a while. Expensive though

jamespetts · « **Reply #682 on:** May 21, 2014, 10:06:51 AM »

Thank you for that report - I will have to fix that before the next release.

Sarlock · « **Reply #683 on:** May 21, 2014, 11:03:10 PM »

Thank you. For the record, if you find a bug or exploit like this, please report it on the forums to James... don't use it in the game and leave it for someone else to discover that you've used it and how it's done. The whole purpose of this server is to identify such things.

jamespetts · « **Reply #684 on:** May 21, 2014, 11:11:00 PM »

I am in the process of fixing this now.

jamespetts · « **Reply #685 on:** May 21, 2014, 11:53:14 PM »

I have now restarted the server with the latest version, 11.32. I should be grateful for any feedback on how this works and the stability and prevalence or otherwise of desyncs, as well as any reports on whether the earlier reported instability of the passenger network has improved.

Sarlock · « **Reply #686 on:** May 22, 2014, 03:56:52 AM »

Did the logs get renamed or moved? I cannot seem to locate them in their usual spot, strangely.

jamespetts · « **Reply #687 on:** May 22, 2014, 07:36:46 AM »

No, there was no intentional change to logging locations. Odd.

ӔO · « **Reply #688 on:** May 22, 2014, 08:00:55 AM »

my logs are still in the same spot.

---

is it intentional that the game makes 100MB+ log files?

jamespetts · « **Reply #689 on:** May 22, 2014, 08:23:55 PM »

Yes - the log files are large because they contain a great deal of detail, especially with TurfIt's work on more detailed logging for the desyncs.

Sarlock · « **Reply #690 on:** May 22, 2014, 10:32:09 PM »

Error on my end, I figured it out... I had inadvertently added a random character to the -log command and didn't notice.

ӔO · « **Reply #691 on:** May 24, 2014, 03:32:25 PM »

Sarlock, you still have many many ships that are stuck in 'no load', possibly because they are 3 tile long schooners that you replaced with 2 tile long brigs.

Sarlock · « **Reply #692 on:** May 24, 2014, 03:35:20 PM »

Yes, I am cycling through them sequentially and resolving them as well as several other things that I am changing with those lines (to make them better mesh with the rail lines)... so it's a slow progressive process. Thanks for the heads up, however.

jamespetts · « **Reply #693 on:** May 24, 2014, 03:53:10 PM »

Server now restarted with new version 11.33 - TurfIt is hopeful that he has found the desync issue, which I have fixed according to his instructions. I should be grateful for feedback on how this works out.

wlindley · « **Reply #694 on:** May 25, 2014, 08:28:27 AM »

For the first time in months, it works for hours at a time. Hooray! Getting started at this late date (1850) and up against entrenched competitors is a bit of a challenge.

ӔO · « **Reply #695 on:** May 25, 2014, 01:21:49 PM »

I haven't desync'd thus far, so it looks like turfit nailed it.

A job well done.

jamespetts · « **Reply #696 on:** May 25, 2014, 02:25:30 PM »

Three cheers for TurfIt!

Sarlock · « **Reply #697 on:** May 25, 2014, 04:01:31 PM »

Indeed! I'm late to the party as I wanted to have a few hours of evidence before I cautiously declared anything, but it is running much, much better now. I have had one checklist mismatch desynch (and one socket disconnect) in about 6 hours of play, which is pretty amazing.

wlindley: starting with road transport is a hard way to make money in 1850. Use ships instead and find some cargo to truck around -- start by connecting it to one of our main freight hubs for some quick profits (and a guaranteed market). Find a couple of cattle farms and connect the milk to our networks.

wlindley · « **Reply #698 on:** May 25, 2014, 07:31:06 PM »

Sarlock: Good point. But now I'll have to wait until the bankruptcy reaper comes along

ӔO · « **Reply #699 on:** May 25, 2014, 11:50:24 PM »

I think there is a major deficiency in the western island with its north-south connection, which is possibly causing the backlog of pax there right now.
I don't know if volvo wants to work on that or not, but it is an opportunity.

I don't want to do it, because it's a hassle maintaining a massive pax network.