Closing RPE app in Windows leaves the backend server running in some instances causing clashes on new rpe instance

[shivaahir158]

Motivate of the pull request

• To address an existing issue. If so, please provide a link to the issue:
• Breaking new feature. If so, please describe details in the description part.

Describe the technical details -

1. Added signal handlers in restapi_server.py to handle SIGINT and SIGTERM shutdowns for the Flask server.
2. Implemented a server_shutdown function to log and execute the backend shutdown process cleanly upon receiving a shutdown signal.
3. Created a Playwright test to launch and close the Electron app 10 times (can do 100 times as well, but job run would be time consuming on CI, while testing I do an iteration of 100), checks that no lingering processes remain.
4. Commented out dsp.test.js and clocking.test.js for now, I will uncomment them again when we shall have proper testing suites, for now doing e2e for app launch, restart, and checking if the app process is killed.
5. Backend PR Closing RPE app in Windows leaves the backend server running in some instances causing clashes on new rpe instance #263

What is currently done? (Provide issue link if applicable)

What does this pull request change?

Which part of the code base require a change

• Frontend:
• Backend: restapi_server.py
• Library: <Specify the library name, e.g. npm packages>
• Plug-in: <Specify the plugin name, e.g. Webpack, jtest>
• Documentation
• Playwright tests restart.test.js, dsp.test.js, clocking.test.js
• Continuous Integration (CI) scripts

Impact of the pull request

• Require a change on Quality of Results (QoR)
• Break backward-compatibility. If so, please list who may be influenced.

User can verify if the process has shut down, in the log file, fyi @NadeemYaseen :

[codecov]

Codecov Report

Attention: Patch coverage is 50.63291% with 39 lines in your changes missing coverage. Please review.

Project coverage is 68.76%. Comparing base (f5dd78a) to head (9aef427).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/utils/serverAPI.js	10.71%	25 Missing ⚠️
backend/api/shutdown.py	50.00%	11 Missing ⚠️
backend/restapi_server.py	89.65%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #263      +/-   ##
==========================================
- Coverage   68.83%   68.76%   -0.07%     
==========================================
  Files         110      111       +1     
  Lines        9022     9052      +30     
  Branches      402      398       -4     
==========================================
+ Hits         6210     6225      +15     
- Misses       2812     2827      +15     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests
• JS Bundle Analysis - Avoid shipping oversized bundles

[ravikiranchollangi]

Let's not comment out the e2e tests. You can re-factor later.
How long does the 10times re-launch test case take?

[shivaahir158]

Let's not comment out the e2e tests. You can re-factor later. How long does the 10times re-launch test case take?

@ravikiranchollangi , 10 iterations takes around 45s, also, I will push all the e2e test cases in the next PR, it needed some fixes which is been done, requesting to for your approval to this PR so we have a solution for Shutdown issue, and I shall raise one more PR for the e2e test cases.

[shivaahir158]

Let's not comment out the e2e tests. You can re-factor later. How long does the 10times re-launch test case take?

@ravikiranchollangi , 10 iterations takes around 45s, also, I will push all the e2e test cases in the next PR, it needed some fixes which is been done, requesting to for your approval to this PR so we have a solution for Shutdown issue, and I shall raise one more PR for the e2e test cases.

@ravikiranchollangi SIGTERM is not a forced shutdown, to my knowledge, it’s a polite request for the process to terminate. When a process receives SIGTERM, it has the opportunity to perform any cleanup tasks before it exits, such as closing files, saving state, or logging. This makes SIGTERM a preferred way to smoothly (not forcibly) stop a process. SIGKILL is a forced shutdown.

I believe SIGTERM is not essential if SIGINT is the only signal sent by the frontend, but keeping both signals allows our backend to smoothly handle termination requests.

[NadeemYaseen]

Adding a screenshot confirms that the backend honors the kill signal sent by the front end.

[shivaahir158]

@NadeemYaseen , check the lastest changes, as discussed, I have created shutdown.py API, called the same api in restapi_server.py, and in cleanup.js I have called the shutdown api, please check the changes once, so we can discuss further. the push will fail because of yml changes that we did previously.

[shivaahir158]

@NadeemYaseen , fetch the latest changes, just removed the need to have axios and called the shutdown function in serverapi.js, and after that called n cleanup.js.

[NadeemYaseen]

@NadeemYaseen , fetch the latest changes, just removed the need to have axios and called the shutdown function in serverapi.js, and after that called n cleanup.js.

@shivaahir158 I am facing the below issue, am I missing something,

[shivaahir158]

@NadeemYaseen , fixed. check now.