Questions about performance

FemtoZetta

Yeah, either this is described poorly and it's not just observers that can do it but anyone who died, or it's really not a problem at all, because observers always could cheat in many ways (like VC) and they really never did that by using pings.

Jip

The decision to disable the markers was made by two council members (if I recall correct) last year December, see also the PR where the change was made from that time:

https://github.com/FAForever/fa/pull/3240

If you do not agree with this change, please make a new topic. It is unrelated to this topic about performance (and stability).

If people want to cheat they can cheat. Observers using a console command to place markers in the game in real-time would leave evidence of cheating in everyone's replay file. Doing it through voice chat would be basically undetectable.
This is not the consequence of the exploit. I'm leaving it vague on purpose because I do not want to give it more attention than it has to have.

Crotalus

Played a fafdevelop game where ACU upgrade after it should finish got stuck and acu was constantly -7 mass.

(I was assisting ACU with cybran t1 assault bot)

This error popped up in the log:

warning: Error running OnStartBuild script in Entity url0107 at 25034d08: ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(901): attempt to call method IsUnitState' (a nil value) stack traceback: ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(901): in function UpdateAssistersConsumption'
...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(971): in function UpdateConsumptionValues' ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(852): in function SetActiveConsumptionActive'
...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(2570): in function <...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua:2533>

Jip

@crotalus said in Questions about performance:

Played a fafdevelop game where ACU upgrade after it should finish got stuck and acu was constantly -7 mass.

(I was assisting ACU with cybran t1 assault bot)

This error popped up in the log:

warning: Error running OnStartBuild script in Entity url0107 at 25034d08: ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(901): attempt to call method IsUnitState' (a nil value) stack traceback: ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(901): in function UpdateAssistersConsumption'
...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(971): in function UpdateConsumptionValues' ...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(852): in function SetActiveConsumptionActive'
...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua(2570): in function <...gramdata\faforever\gamedata\lua.nx5\lua\sim\unit.lua:2533>

Hi Crotalus, can you share the entire log and the replay ID?

edit: the error you provided is from a Mantis.

Crotalus

@jip game_15190526.log

Jip

And timestamp as to when it roughly happened? And to what player?

Jip

Ah, I think I understand what happened here. Thank you for sharing.

edit: found and understood the issue.

BlackYps

I'm still not sure if this applies to games with observers off (i.e. most of the games) at all.

Edit: woops, meant to post this in the other thread, sorry

Jip

@BlackYps You can continue the other discussion here:

https://forum.faforever.com/topic/2461/ping-after-death-could-be-removed

zwierzak

@Jip In the same game that Crota reported, MaximusXXX said at 38:21 that he was able to build t2 navy despite the HQ being dead.

game_15190526.log

Jip

oh no

I'll look into that too - thank you for mentioning it.

Jip

@crotalus It is fixed - thank you for reporting it! This was also the issue @excelsior (discord) described and @Crotalus experienced during that replay.

@zwierzak The replay desyncs for me and that happens to be MaximusXXX . Can you describe what happened?

Jip

@utopian Thanks to KionX we may have found out what caused this crash. It will not be part of the next patch because its deadline is too close - but we'll certainly look at it. Again, thank you for reporting it.

snoog

@jip said in Questions about performance:

@crotalus It is fixed - thank you for reporting it! This was also the issue @excelsior (discord) described and @Crotalus experienced during that replay.

Is the fix live on the dev branch right now? Just wondering if I should warn people or not when hosting games.

Jip

The fix with the mantis is live - yes

zwierzak

I failed to reproduce MaximusXXX's error and the game now desyncs for me, too. Probably because of the Mantis fix.
However, while testing just now, I found new problems:

Game: #15195914
Timestamp: 20:00
Issue: UEF T1 Attack Bomber seems to do very little damage

Game: #15195914
Timestamp: 28:00
Issue: Cybran T2 Naval Support Factory auto-repairs VERY quickly

Log isn't much help, as I didn't have debug enabled.

Jip

@zwierzak

Game: #15195914
Timestamp: 20:00
Issue: UEF T1 Attack Bomber seems to do very little damage

This is a known issue, it has to do with a change unrelated to performance. The Janus is hit hard too by that same change. Keyser is going to look at it this weekend.

Game: #15195914
Timestamp: 28:00
Issue: Cybran T2 Naval Support Factory auto-repairs VERY quickly

I'll look at it - thanks for reporting it.

Saver

Hello Jip, thank you very much for your work in the forum and the game. Thanks also to the team. I think it's great how the community helps with every question. That's what I wish for in every game.

Regarding the question, I have noticed a drop in performance in the game so far (1 - 2 players vs 4 AIs)

Jip

There is an issue with the current faf develop that was introduced yesterday. It causes a lot of warnings. It is fixed in about an hour .

Jip

@Saver it should be stable now again .