fix(kill): should not try to open a connection to a not running ReplSet

[arthursimas1]

I've a situation where I listen to kill signals and execute .stop() when needed. The replica set has already been killed because the Ctrl + C signal is propagated to the whole process group, then I waits forever to connect and send a graceful shutdown.

[codecov]

Codecov Report

Merging #811 (b58958a) into master (c1f2333) will increase coverage by 0.20%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #811      +/-   ##
==========================================
+ Coverage   90.94%   91.14%   +0.20%     
==========================================
  Files          15       15              
  Lines        1965     1965              
  Branches      494      494              
==========================================
+ Hits         1787     1791       +4     
+ Misses        169      165       -4     
  Partials        9        9              

Files	Coverage Δ
...ngodb-memory-server-core/src/util/MongoInstance.ts	98.61% <100.00%> (+1.84%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[hasezoey]

Looks good, though we should probably also add some options to the connection (like connectTimeoutMS which defaults to 30s)

[github-actions]

🎉 This PR is included in version 9.0.1 🎉

The release is available on:

• npm package (@latest dist-tag)
• npm package (@latest dist-tag)
• npm package (@latest dist-tag)
• npm package (@latest dist-tag)
• npm package (@latest dist-tag)
• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀

[arthursimas1]

hey @hasezoey.
I've tried the following options:

// 1)
con = await MongoClient.connect(uriTemplate(ip, port, 'admin'), {
  ...this.extraConnectionOptions,
  directConnection: true,
  socketTimeoutMS: 100,
  connectTimeoutMS: 100,
  waitQueueTimeoutMS: 100,
});

// 2)
con = await MongoClient.connect(uriTemplate(ip, port, 'admin'), {
  ...this.extraConnectionOptions,
  directConnection: true,
  serverSelectionTimeoutMS: 100,
});

Only the 2) case works. Right now I cannot think out of my head a test case for this alternative solution.

[arthursimas1]

Testing some cases, it is possible that the server is killed between the isAlive assert and the connection. I agree that it's better to use the connection options. I'll send another PR.

[hasezoey]

I agree that it's better to use the connection options. I'll send another PR.

no need to replace the change from the current PR, the options are only meant because it should not have much latency because it will always connect to localhost and so 30 second timeout is somewhat high