Not a Bug, But new Limitation? InfoMessage seems to now have a hardcoded limit

2023.1.8 (Do not know how far back this occurs.)

I was using the system last night, and was using a dashboard that has an option to send me back a “slice” of textual data as an InfoMessage, and it kept failing.

It used to work properly as coded, which was to send back 4096 bytes, or less.

Now it seems to have the following behavior:

1754 characters - Message comes back successfully → Success

1755 characters - Message comes back → successfully but then …

  • Error After: Unable to read beyond the end of the stream.

1756 characters - Error: Unknown REST exception:

<html>
<head><title>502 Bad Gateway</title></head>
<body>
<center><h1>502 Bad Gateway</h1></center>
<hr><center>nginx/1.23.2</center>
</body>
</html>

Imagine how fun that was to track down :rofl:

Thoughts?

I wonder if the message is being passed through the http header, which is limited. The 1.7K plus other cookies, etc. might have pushed it over a limit. :thinking:

It is, and that has lowered the practical limit to significantly lower that when they were using WCF, however, it has been up until recently still been supporting at least up to 10k characters, which is as far as I bothered to test before.

Hmmmm…

I decided to change environments, and I believe it’s isolated to my AKS CAAB instance.
They must have lower limits set, for what reason, I don’t know why. I will have to ask.

I moved into my Pilot environment, which is also 2023.1.8 and just sent 40k bytes back.

Please Disperse Naked Gun GIF

Issue was specific to AKS environments. Cloud services applied a fix, and has corrected their setup process for the future.

Seeing this error in our event logs we on on Prem with 2022.1 Just can’t narrow down to what the problem is.

I would hazard a guess something is putting more data in the call context than what is normally allowed/configured.

You getting the gateway error, or end of stream error?

@klincecum We a are getting