Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update CI and improve discard_results #500

Merged
merged 7 commits into from
Feb 25, 2023
Merged

Update CI and improve discard_results #500

merged 7 commits into from
Feb 25, 2023

Conversation

larskanis
Copy link
Collaborator

This PR updates CI on Appveyor and Github Actions and adds a test of the binary gems for x86-mingw32.

Adding the platform x86-mingw32 showed that there can be issues with discard_results getting into starvation under memory pressure. So the second half of commits improve discard_results behavior, documentation and specs.

larskanis and others added 7 commits February 24, 2023 12:32
On Windows ruby-3.1 is needed.
Just to be safe, since x86-mingw32 is a bit slower.
It takes approx 30 seconds on my Laptop with Windows in qemu-kvm.
.. in the busy loop and after PQputCopyEnd.

This blocked on ruby-3.1 on x86-mingw32 like so:

  1) PG::Connection in nonblocking mode rejects to send lots of COPY data
     Failure/Error: raise("scheduler timeout in:\n#{th.backtrace.join("\n")}")

     RuntimeError:
       scheduler timeout in:
       D:/a/ruby-pg/ruby-pg/spec/pg/connection_spec.rb:666:in `discard_results'
       D:/a/ruby-pg/ruby-pg/spec/pg/connection_spec.rb:666:in `block (4 levels) in <top (required)>'
       D:/a/ruby-pg/ruby-pg/spec/helpers.rb:645:in `block in run_with_gate'
       D:/a/ruby-pg/ruby-pg/spec/helpers.rb:602:in `block in thread_with_timeout'
     ./spec/helpers.rb:607:in `thread_with_timeout'
     ./spec/helpers.rb:641:in `run_with_gate'
     ./spec/pg/connection_spec.rb:645:in `block (3 levels) in <top (required)>'
     ./spec/helpers.rb:55:in `block in included'

Experimentation showed that under memory pressure PQflush, PQconsumeInput and PQgetResult must be called in combination to not block infinitely.
Separate loops for PQflush+PQconsumeInput and PQgetResult+PQconsumeInput don't work and often result in starvation on Github-Actions on x86-mingw32.
@larskanis larskanis merged commit cb249b4 into ged:master Feb 25, 2023
@larskanis larskanis deleted the ci branch February 25, 2023 14:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant