You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I use an RPC with "unreliable" or "unreliable_ordered" mode. The channel doesn't matter.
The bug is that sometimes, some RPCs aren't received (aka packets are lost)
Now, that is normal because of UDP, but I am uncertain if this is the case because this is localhost.
As in, I have no LAN or wifi connected, and send just integers. I searched if UDP packets can drop at localhost and https://stackoverflow.com/questions/7968566/what-would-cause-udp-packets-to-be-dropped-when-being-sent-to-localhost#7968907 users claim yes at MB/s to overflow the buffer, but this isn't even 1 KB/s
Granted, packets dropping at localhost are rare, but it got me to wonder and at least want to report it in case there is some ENET initialization bug or something, given it often happens near the start of joining.
So feel free to close this issue without warning, as it is very likely a false flag.
As for my testing, this is a tldr video
unreliable-losing-packets.mp4
of the MRP. And I have to admit it takes like 5+ boots to get it to happen.
Steps to reproduce
Open the MRP
Press F5/F6
Host as server
Join as client
Await a few seconds, if no errors, F8, repeat step 2 until you get an error.
Minimal reproduction project (MRP)
(The MRP is a tweak of a very old MRP made by @Mathis-Z)
I'm not an expert in this area. But aren't unreliable and unreliable_ordered by definition "unreliable"? They are not "unreliable-if-not-localhost" 😛
Packets are not acknowledged, no resend attempts are made for lost packets. 1
I don't think it makes sense to explore the conditions required to meet the no-packet-loss case in unreliable mode. The documentation should prevail. If you do want to avoid packet loss, reliable mode should be used instead.
I know but I want a confirmation because these packets drop near the start of the launch/boot and it seems weird for localhost to drop packets, especially because its just an integer spam being ~1KB/s, so it could be a bug
Tested versions
System information
Godot v4.4.dev (0a9ad8f) - Artix Linux #1 SMP PREEMPT_DYNAMIC on Tty - X11 display driver, Multi-window, 1 monitor - OpenGL 3 (Compatibility) - NVIDIA GeForce GTX 1050 Ti (nvidia; 560.35.03) - AMD Ryzen 5 2600 Six-Core Processor (12 threads)
Issue description
I use an RPC with
"unreliable"
or"unreliable_ordered"
mode. The channel doesn't matter.The bug is that sometimes, some RPCs aren't received (aka packets are lost)
Now, that is normal because of UDP, but I am uncertain if this is the case because this is localhost.
As in, I have no LAN or wifi connected, and send just integers. I searched if UDP packets can drop at localhost and https://stackoverflow.com/questions/7968566/what-would-cause-udp-packets-to-be-dropped-when-being-sent-to-localhost#7968907 users claim yes at MB/s to overflow the buffer, but this isn't even 1 KB/s
Granted, packets dropping at localhost are rare, but it got me to wonder and at least want to report it in case there is some ENET initialization bug or something, given it often happens near the start of joining.
So feel free to close this issue without warning, as it is very likely a false flag.
As for my testing, this is a tldr video
unreliable-losing-packets.mp4
of the MRP. And I have to admit it takes like 5+ boots to get it to happen.
Steps to reproduce
Minimal reproduction project (MRP)
(The MRP is a tweak of a very old MRP made by @Mathis-Z)
mrp.zip
Its code is just 1 class with this code, posted below by filtering away the host/join code:
The text was updated successfully, but these errors were encountered: