Some times we only get something like below from stap:
WARNING: too many pending (warning) messages
ERROR: too many pending (error) messages
WARNING: /opt/stap-plus/bin/staprun exited with status: 1
Pass 5: run failed. [man error::pass5]
It's not helpful at all since we have no idea what messages are actually flooding the control message buffers.
We should make stap output the first few and also the last few messages in the current control buffer when it thinks that there are "too many". Otherwise there is no way to debug things here.
These messages are printed when the rate of the messages is so high that there wasn't room to store any more of them. Maybe the suggestion is to drop -earlier- ones rather than new ones? Not sure how easy that would be.
As long as we output some warning messages from the buffer at least. Right now I don't see any warnings at all generated, which makes debugging near impossible if without tracing stap itself...
I wonder if you have a reproducer handy. A few attempts here always resulted in some real payload WARNINGs' before the "too many" ones.
Yeah, it seems only reproducible in our big examples. I'll try digging a bit myself...