Analytics: Real Time Error Reports and Client Metrics (Crash Rates, Frame Rates, and Memory Usage)

[NoobFragged]

Even though it comes at a time when I’m struggling with another update, this one is greatly appreciated.

I hope this feature continues to be built on. It would be nice if we could find data to help source isolated errors to. Like how others have mentioned exploiters spamming errors that obviously will not line up with normal users.

[NoobFragged]

Roblox has said if you don’t do the questionnaire your game age rating defaults to 13+.

I would suggest doing the questionnaire.

Edit: It looks like you’ve done it, at least now, and your game is 13+
At least now players will know and see it.

[signal_zzz]

Thank you all for your excitement! We’re so glad that you’re enjoying these updates.

Here’s what we’ll prioritize as fast follows:

1. Add filter to server or client errors only
2. Add custom start and end date support

P.S. If you’re part of a group, you should be able to see both the error report and performance page if you have group analytics permissions.

We’ll be monitoring this thread - keep the feedback coming

[JoyfulFlowerMary (Mary)]

@BackspaceRGB

But whats just stopping exploiters from executing this script

@PablitoTheChicken

Btw, do errors caused by people running scripts using exploit software like Electron or Synapse also get logged here if they make an error by mistake?

I’m assuming any client errors will appear in this analytics section.

I’m not sure how Roblox would prevent exploit scripts from spamming invalid errors and warnings, flooding the analytics with useless data, unless they somehow noted which scripts are a part of the game itself then didn’t show any errors that weren’t written by those scripts.

They probably should filter which scripts are allowed to add their outputted warnings and errors here (basically ignoring everything in CoreGui, since that’s also where exploiters place their scripts). Otherwise, is an exploiter intentionally makes a lot of errors about invalid animations appear, they could send any message “to the developers”, or make them think that their experience isn’t programmed well.

If I made an experience that a lot of people “experienced”, I know I wouldn’t appreciate exploiters basically sending me messages through errors, cluttering my error log with junk, when I’d rather see real errors that I could try fixing.

Edit: I just experimented and found that errors written by exploits do not appear in analytics, so I don’t think I need to be afraid of them messing up my error logs.

[signal_zzz]

Thanks Mary this is good feedback - we’ll find some way to prevent this from happening.

[iOwn_You]

Isn’t that a recommendation though?
I thought It doesnt actually prevent players from playing the game

[BackspaceRGB]

Adding onto the incorrect error logs, I literally found the example error that was shown in the original post, in my game (among a lot of others which are clearly not from my game)

Something is definitely fishy. Really hoping this gets fixed soon because its quite annoying, not knowing which bug is actually in your game

[jan12000p]

This is a great update, but:

Please add a toggle for Core GUI errors :

That big one is a core gui error. The picture was taken at 33% normal browser sizing.

Also, Exploiters? (I don’t have anything in ReplicatedFirst in my game, and that unknown symbol thing is suspicious not to mention the fact it is trying to call the command bar.) :

Suspicious

Lastly, I use adonis in my game, so it would be nice to have a keyword exclusion filter.

Actually lastly, having a player count line would be nice to know if less errors at a certain time was just because there were less players.

[vikboy666 (VIK)]

I’ve also seen this in my games. With today’s tool, I know how many Exploiters the roblox platform is

[Baysights (Gaius Maecenas)]

This is insanely helpful and has already proven useful for performance issues in games I work with.

I do have one question… how do the percentiles work in this setting? I am aware of what a percentile is but I do not understand what percentile it is necessarily calculating. Any further information on this?

[7z99 (cody)]

I love this update.

Are there any plans for allowing these to be posted to a webhook or even an API to manually do it through an external server? In my opinion it’d be much easier to go through the issues if they’re posted in a dedicated channel rather than through the creator dashboard, but it’s more of a minor thing.

[Timelens]

Amazing quality-of-life update… however, as others have rightfully pointed out… what about errors forced from script executors? They’re clearly being logged and are clearly coming from unknown scripts.

Wasn’t Byfron activated not too long ago? Unless it was disabled, it is clearly entirely useless since the script source is not coming from the game.

We should be able to see the usernames of the players who encounter the logged client errors.

[IsNotJail (Jail)]

So, just curious what % is regular for everyone when it comes to crash rates. I am hovering under 1%, which does not seem good, but most of it is coming from singular devices. For example, the IPad Air has a 31% crash rate for whatever reason.

Now, these rates may not be super accurate as I am only at around 50 concurrent players, and only so many are on an IPad Air, but still, it is concerning to see a 1% crash rate, so I was wondering what everyone else has got.

[lolmanurfunny (lolman)]

@BackspaceRGB

But whats just stopping exploiters from executing this script:

while true do
    error("This is a fake error!")
    task.wait()
end

And flooding the error log?

Anyone who’s explicitly called error, or even assert for that matter wouldn’t ask that question.

error kills the thread it is called in…

[apenzijncoolenleuk1 (Apenz1)]

Incredibly useful.
Very interesting to see the amount of Roblox corescript errors/warnings. Most of them are probably not critical, but can we expect the corescript errors to reduce now that there’s proper analytics for it?
Or perhaps there were always hidden analytics?

[Blafert]

This is a convo I had with a dev directly, I’ll just put it here in its entireity:

Blafert
Do the P1, P5…P99 give info on a single server? A changing one of course, but a single one? Or does it group multiple servers around that percentile together into an average?
Dev
Is average of percentile of groups of servers. Allow me to explain: servers are grouped into smaller groups this depends on a hash of a random ID so then we calculate percentiles on the observations of the metric for all servers in the same group, then we average the different groups.
Blafert
How large are these groups generally?
Dev
very small ~10
Blafert
I assume they’re smaller when there are fewer servers?
Dev
Sorry not 10 servers per group I meant 10 groups, and yes is a max so small amount of servers might even use 1 group
Blafert
Okay, that makes sense. Much appreciated for the answers!
Dev
Np

I hope this clears it up?

[Baysights (Gaius Maecenas)]

I understood that a bit. They should probably include an explanation on the create hub for others.

Thanks for the response!

[BackspaceRGB]

Alright, what about warn then, or just running the same script with error a lot of times?

[xPoloBear (Polo)]

This is an excellent and a long awaited analytics tool for us to check in-game errors! Thank you so much!!
One thing I would suggest though is to add a sort button for client/server errors.

[AerooDev (Aero)]

Why is this great feature not accessible to the developers who work on a place that is under someones profile? Would make our lives much easier.