2-Second lag spikes consistently occuring roughly after 30 minutes

For the past few weeks, I’ve been regularly getting a 2-second ish lag spike after extended play (about 30 minutes). I finally managed to catch the lag spike with the microprofiler active, and the issue seems to be an abnormally long Net PacketReceive (~20ms) followed by an insanely long Write Marshalled (~2s)
This happens to me on all games, including a regular empty baseplate


I’ve attached some html microprofiler dumps in the private content

A private message is associated with this bug report

2 Likes

Thanks for the report and the profiler captures! Did you test with the empty baseplate in studio, or did you have to publish it to see the issue?

Also, do you happen to have the client/studio logs from when you were able to reproduce this?

I’ve not tested this within studio, only while playing on the client. I’ll edit this message once I’ve had time to test if it happens in studio too It doesn’t seem to happen in studio (couldn’t cause a lag spike nor see the relevant log after 30 minutes)

I do still have the logs for both of the dumps, and I do see something relevant in them. I’ve attached both of them in a reply to the private message of this post

1 Like

Same here. When I stay inside a game for EXACTLY 30 minutes, a few seconds of lag occur in the same manner you’ve described here. It happens with ALL types of games, whether it is a baseplate or a full-fledged game.

I tried a few times on client and was unable to repro the lag spike. Are you still seeing this?

I don’t have the time to do a perfect repro test right now, but I’ve definitely not noticed this issue at all the past few weeks, despite extended play sessions. I suspect this resolved itself not too long after this post initially got made

I can confirm that neither the log message nor the lagspike happen after being in a game for 30 minutes

Thank you for following up, I’m going to close this as “Can’t Reproduce” for now, but if you see this again, please submit another report! We also have planned an improvement to these microprofiler labels that will help pinpoint similar issues. The label right now is a generic one for a task queue in the engine, so we don’t know which task is actually taking all that time without being able to reproduce it ourselves.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.