RBXCRASH "Out of job nodes" joining development game

[lewisakura (lewi)]

We’ve recently started encountering an RBXCRASH with the following message:

2025-06-16T17:28:16.061Z,16.061840,5a68,6 [LOGCHANNELS + 1] RBXCRASH: HardAssert (Out of job nodes. Check for job leaks and/or increase the FInt::TaskSchedulerMaxNumOfJobs value)

This crash only occurs in our development game and doesn’t seem to affect any other games that we play (so we can tie it to something our game is doing, but we’re not really sure what), and is prefaced with what I assume is a job dump showing the following:

[snip, repeats the same three lines 20 or so times]
2025-06-16T17:28:16.061Z,16.061840,5a68,6 [FLog::Output] ReplicationDataCacheTaskQueue
2025-06-16T17:28:16.061Z,16.061840,5a68,6 [FLog::Output] MegaReplicatorTaskQueue
2025-06-16T17:28:16.061Z,16.061840,5a68,6 [FLog::Output] MegaReplicatorPPRTaskQueue

Of note, I see this above the crash logs:

2025-06-16T17:28:15.511Z,15.511263,5a68,7 [DFLog::MegaReplicatorLogDisconnectCleanUpLog] Destroying MegaReplicator.
[repeats 40 or so times]

Privately attached is the last known bad copy of the place (where we can reproduce this), the full log, and a link to the affected experience.

A private message is associated with this bug report

[Rockraider400]

(I work on the same game as lewi)

We discovered a workaround – if we place task.wait() in between every client controller KnitStart function, the client no longer crashes.

This is what it looks like in Knit/KnitClient.

-- Start:
for _, controller in controllers do
   if type(controller.KnitStart) == "function" then
   	task.spawn(function()
   		debug.setmemorycategory(controller.Name)
   		controller:KnitStart()
   	end)
   	task.wait() -- added task.wait() prevents client crash
   end
end

Link to Knit API in case you’re wondering how it works: Knit | Knit

[newtonmetre]

Thank you for the report, we’ll take a look at this.

[asdfiji123]

Hey! Some time ago, we released a change such that this should no longer repro. Please let us know if the issue persists!