bugfix(network): Reset network command id to keep commands in chronological order#2736
Open
Caball009 wants to merge 3 commits into
Open
bugfix(network): Reset network command id to keep commands in chronological order#2736Caball009 wants to merge 3 commits into
Caball009 wants to merge 3 commits into
Conversation
This reverts commit 52d5e82.
Caball009
force-pushed
the
fix_network_commandID_overflow
branch
2 times, most recently
from
ee6bcd8 to
1f31061
Compare
|
| Filename | Overview |
|---|---|
| Core/GameEngine/Source/GameNetwork/NetworkUtil.cpp | Moves commandID from a function-local static to a file-level static and adds ResetCommandID(). First generated ID after reset is 1 (pre-increment). Starting value changed from 64000 → 0, intentional and correct with the new periodic reset. |
| Core/GameEngine/Include/GameNetwork/networkutil.h | Adds ResetCommandID() declaration. Header already uses #pragma once. No issues. |
| Core/GameEngine/Source/GameNetwork/Network.cpp | Adds m_lastFrameResetCommandID member, calls ResetCommandID() in init(), and adds a periodic reset every MAX_FRAMES_AHEAD (128) frames during active gameplay in update(). Also caches TheGameLogic->getFrame() into a local frame variable. |
| Core/GameEngine/Source/GameNetwork/NetCommandList.cpp | Adds an informational comment explaining why the insert-sort does not handle commandID overflow — the retail-compatible fix is to reset the counter rather than change the sort logic. |
Sequence Diagram
sequenceDiagram
participant GL as GameLogic
participant N as Network::update()
participant NU as NetworkUtil (commandID)
participant NCL as NetCommandList::addMessage()
GL->>N: getFrame() → frame
Note over N: Check reset condition:<br/>frame + runAhead > lastExecutionFrame<br/>AND frame ≥ lastFrameResetCommandID + MAX_FRAMES_AHEAD
N->>NU: "ResetCommandID() → commandID = 0"
N->>N: GetCommandsFromCommandList()
N->>NU: "GenerateNextCommandID() → ++commandID (= 1, 2, …)"
NU-->>N: unique ID per command
N->>NCL: addMessage(cmd with ID)
Note over NCL: Insert-sort by type → playerID → commandID<br/>(no overflow check needed — ID reset prevents wrap)
N->>GL: RelayCommandsToCommandList(frame)
Reviews (2): Last reviewed commit: "Added code to reset command id." | Re-trigger Greptile
Caball009
force-pushed
the
fix_network_commandID_overflow
branch
from
1f31061 to
ebf1fa2
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Synchronized network commands are given a unique id so that they can be sorted chronologically. The id needs to be unique per execution frame. It's a uint16 value so it'll overflow quite fast, especially because it's never reset and the starting value is 64000 by default.
When the network runahead decreases it's expected that multiple frames are executed right after each other. I noticed an issue with the overflow when I set the CRC interval to once per frame. When the overflow happened for a client combined with a decrease of the runahead, the order of the CRC messages was no longer guaranteed to be in chronological order for that execution frame. That would cause a mismatch if the overflow didn't happen for all clients at the same time.
With a starting value of 64000 and a CRC message every frame the first overflow happens after ~25 seconds. You can see it mismatch here because of the overflow (this was with a special test build);
https://www.youtube.com/live/V_l-q9Y-DmA?t=621s
https://www.youtube.com/live/V_l-q9Y-DmA?t=9499s
I've added some test code in the first commit. If that's used without this fix, it becomes very apparent that the messages are not in the right order when the command id overflows. On top of that, it's very likely that the disconnect bug (disconnect screen with everyone at ~59XXX msec) occurs frequently.
I intend to do a follow up PR to make the
static commandIDvariable a part of theNetCommandMsgclass.