You will find a treasure trove of desync reports and crash logs on FactorioMMO discord in
#clusterio-desyncs channel - screens and screens of those, some with more context than others.
Let us know if you would like us to collect those and post in a forum thread or you are fine with looking into Discord - it's a dedicated channel for desyncs/crashes only.
As far as we understand, it has actually little to do with mods and everything to do with long commands
Code: Select all
/silent-command call_function('{sizeable json here}')
Factorio complaints, basicallly, that there are broken UDP packets and all signs point to some kind of network traffic processing issue, mangled packets and all that jaz.
Server never fails, it executes all commands without a hitch - I think we haven't had a server crash the whole weekend for all 25 instances we are running and tons of people desyncing, some to a point that they can't join any of 25 servers without an immideate desync.
We tried mitigating the issue by reducing command lengths to various sizes, right now with a max of 400 symbols, but that still has issues.
You can see attempts to fix it here
https://github.com/Danielv123/factorioC ... ommits/dev - MCRcortex'es commits for the past weekend. It has decreased the amount of desyncs and crashes overall, but now either people play just fine with maybe desyncing once an evening or they desyncing all the time or can't even join any servers due to immediate desyncs. Some servers are more prone to this than others due to different geo locations, wich probably points to a latency issue of some kind.
Full code block that handles the split starts here and is
https://github.com/Danielv123/factorioC ... nt.js#L832 11 lines long.
This affects a large part of people playing the Clusterio project we have going, and we have 500 people whitelisted on the servers, so it's a quite big audience.
I would also imagine your crashlog uploaded is gonna see a big spike
