bridgewater-brunel.me.uk - Simutrans-Experimental - Pak128.Britain-Ex 0.9.0

[asaphxiix]

Alas, I can hardly play on the server for a while now. I can connect sometimes until the end of the month, but then I desync and try to connect until I give up.

[dustNbone]

Yeah I'm in a similar boat.  The game is running my CPU fairly hard during the month, and at month end it just falls too far behind to remain connected.  Also another player joining will generally result in a desync.  My CPU is an AMD Athlon x4 840 running at 3.3Ghz, so not stellar but also not low end.  There aren't many CPUs in the AMD family with better single core performance, so I think the "practical" limits of CPU load have been reached and somewhat exceeded. 

[jamespetts]

I am sorry that you are having difficulties - thank you very much for reporting them. Bernd Gabriel has been doing a lot of work on improving performance on his 112.x-merge branch, but these will take a while to test and integrate fully. This game has been a very useful testing ground for these performance issues, however, so your feedback is extremely valuable.

[chicken]

I think it crashed when I tried to replace my trains, earlier.

[jamespetts]

Has it reverted or stopped completely?

[ӔO]

I tried replacing a few hundred buses, and it never reverted.

Replacing up to 200 at once should not cause a desync, but over 300 at once does.


btw, when do cities start demanding electricity? I thought they started demanding it in 1910.

[jamespetts]

Looking at electricity.tab in my Pak128.Britain-Ex 0.8.4 directory, it looks as though you will have to wait until 1925. I must have amended the date back to 1910 after distributing 0.8.4. Apologies!

[ӔO]

lindley lines has 3 stuck trains at finsland. Luckily, it's still working with one functional platform.


P.S. Capacity problems between Enthorne and Yerlbridge.
P.P.S Deanpool tram and bus line also has capacity problems.


Also Capacity problems around Coatsand and Rainleigh for NLT. I can't pinpoint it exactly, but it seems to overflow once in a while.

[wlindley]

Ah, ...I replaced the stub-end terminal at Finsand with a one-way loop.  Even with curves and a tunnel, the elimination of reversing -- plus the hassle of whatever in the game is fouling the choose block reservation -- makes that station a much happer place.  Will look ar the others.

[ӔO]

ah, now a lot of buses are getting no route because they are too heavy for cobblestone roads...

Cities really ought to upgrade their roads more aggressively, rather than rely on players doing the job.

I'll try fixing up the roads as best I can, but I can't do anything about the buses that were sent to depots.

[jamespetts]

Hmm - the current rule is that roads upgrade when the buildings in front of them upgrade. What other strategy would you suggest?

[ӔO]

would it be possible to upgrade/downgrade all roads (that are not bridges or tunnels) within city borders to X road type after X date?


Bridge at Enthorne City Hall Railway Station is too light (9t) for the fully loaded AEC Q2 buses

[jamespetts]

Hmm - this seems a little arbitrary, and will require a lot of manual intervention to set up for each specific road type.

[ӔO]

or just automatically upgrade/downgrade all roads built within city borders to be of the road type for the given age, although that's also a bit unrealistic.

Ideally, inside cities, well used roads with X convoys traveling over them should upgrade themselves.

[jamespetts]

How would we decide what X is...?

[ӔO]

There should be a theoretical limit of convoys a stretch of road can handle in a month if it has Y speed.

But it may be more practical to look at the scheduling. I would consider anything over 64 convoys/month to be a fairly busy bus line for a 6h:24m month. 32 to 48 co/mo is what I use on medium usage lines.


---
For giggles, I tested out theoretical limits, when both ways are used, and it looks like this:

around 685co/mo for 15km/h road.
around 950co/mo for 25km/h road.
around 1100co/mo for 35km/h road.

Which is far more convoys than possible inside cities that have many intersections.

[ӔO]

NLT has stuck trains at wingbury.

probably because convoy 4663 has reverse route enabled.

[asaphxiix]

strange, I don't see any stuck trains, and convoy 4663 doesn't have reverse route...

[ӔO]

yeah, I sent 4663 to the depot with some effort. The game was quite unplayable otherwise, with the +300k pax constantly rerouting.

ah, maybe it wasn't 4663, it was 4660.
It should be in the depot at wingbury.

[ӔO]

dustNtrans has a stuck train.


All player profits are going down and some players are losing money, although not badly yet.

[ӔO]

convoy 4773 is broken. It keeps trying to go to the bus depot at durdon, but the road underneath the depot is too light for the convoy.
End result is the schedule has 254 stops and the game stutters every time it tries to send the convoy to the depot by adding another line.

the road underneath the depot needs to be upgraded or convoy 4773 needs to be removed to fix the stuttering.


P.S.

I notice that the server runs smoother after restarting from crashes and keeps that way until 3~4months later

[wlindley]

I have relocated cities and will be unable to play for anohter (physiological) month or two, if someone could please deactivate lindley lines (assuming it's still around).

[ӔO]

lindley lines is still doing quite well. It has been in financial equilibrium for the past 5 or so years.

There are only two points that bleed away money.

- Lack of Mail road vehicles causing extensive backlogs
- electrified branch lines have inadequate capacity


Other than that, it's not in bad shape. Nothing extensively old, but nothing new either.

It's 1954 right now, so there won't be anything good to update road and rail vehicles with for another 5 to 8 years.


If you would like, I can fix the problematic points.

There's nothing that needs to be done (more like nothing to do until 1959) with my own network, so my hands are free.



---

P.S.

If anyone has trouble getting in, it may require at least 3 tries.

login 1 may result in a no connection error, if the server had no activity for some time.
login 2 may work, but the server takes some time to get moving
login 3 should work after 5~10mins after login 1

[jamespetts]

This server is now offline, as I try to upgrade to the latest Release Candidate to see how the latest version copes with a large network game. Unfortunately, I am having some trouble making the command line server version work properly: for some reason, it seems to compile as a version that demands graphics and refuses to run on my Linux server; the server will remain down until these issues have been fixed.

[jamespetts]

The server has been restarted with the 11.9007 RC for testing purposes - comments on how this RC works in network mode would be most appreciated!

[asaphxiix]

crash and revert now, for no obvious reason this time - I was just looking around, last action I think was open a station window - Eveingham Fountain Railway Station. This does not reproduce.

[jamespetts]

Hmm - perhaps somebody else created a line?

[japadapa]

I get disconnected when creating a line.

EDIT: But I seem to be able to create line if I first set up individual schedule and then promote it to line.

[jamespetts]

Hello and welcome to the forum! Thank you for your feedback. The issue with creating a line crashing the server (but not the client) is a known bug and will be fixed in the next version. In the meantime, your workaround is very useful - thank you.

[Junna]

At Limburg Junction station we have an example of a train stuck due to a new bug with reservation blocks, I think. I've had this in single-player, too.

[jamespetts]

I cannot find any town called "Limburg". What are the co-ordinates of this station?

[Junna]

Apologies - it is Lumbury. Pos. 1935/721.

[jamespetts]

Is this not perhaps an instance of this bug from Standard (and as yet unfixed there)?

[Junna]

Is this not perhaps an instance of this bug from Standard (and as yet unfixed there)?

It might be, though I have had it happen without long-block signals, but maybe that can happen with that bug?

[jamespetts]

I don't know - can you reproduce the occurrence of the problem that you describe without a long-block?

Edit: It seems on further testing that the problem is more that the long-block signal does not recognise the implicit signal of a station as a further signal: adding a further actual signal releases it.