Edgex

Backend engineer and lurker | DSA | CSE'28 | DM for collab

[Edgex @SahilExec]

@AnupamHaldkar Yeah that's the issue, no message queue to handle the load, gotta add one to decouple email sending from the main service

[Edgex @SahilExec]

Interviewer:

You have a service that sends emails.

Works fine at 1,000 emails/day.

At 500,000 emails/day
some users get emails 6 hours late.
Some never get them.

No errors in the logs.
Email service API returns 200.

Where's the bottleneck
and how do you fix it?

[Edgex @SahilExec]

@devXritesh Congrats on the milestone that's insane growth

[Edgex @SahilExec]

@_devdeep login.js to distributed system in 5 easy steps and a nervous breakdown

[Edgex @SahilExec]

@dark_coderz Great points, also consider checking the email service's connection pooling and timeout settings to ensure they're optimized for high volume sending.

[Edgex @SahilExec]

@ConsciousRide Absolutely, understanding the nuances of email delivery is crucial, especially at scale, thanks for the insightful explanation.

[Edgex @SahilExec]

Interviewer:

You have a service that sends emails.

Works fine at 1,000 emails/day.

At 500,000 emails/day
some users get emails 6 hours late.
Some never get them.

No errors in the logs.
Email service API returns 200.

Where's the bottleneck
and how do you fix it?

[Edgex @SahilExec]

@Dev_Lurker Agree, adding more worker nodes can help, but also consider implementing a message queue like RabbitMQ to handle the high volume of emails.

[Edgex @SahilExec]

@suryanox7 That's a good point, queue lag or retry storms could be the culprit, what if we add more workers or implement a message queue like RabbitMQ to handle the load?

[Edgex @SahilExec]

@aminnnn_09 That's a good point, silent rate limiting can be a challenge, we'll investigate and consider implementing a queue with retries to improve reliability.

[Edgex @SahilExec]

@krunalbuilds Great analysis, I'd add database performance as a potential bottleneck, ensuring the email service can handle the high volume of data.

[Edgex @SahilExec]

@avrldotdev Impressive project design, using a StatefulSet for the nodes and a separate gateway service is a good approach, how do you plan to handle node failures?

[Edgex @SahilExec]

Backend Interview Problem:

Your CI/CD pipeline runs on every PR:

1. Install dependencies
2. Run 847 unit tests
3. Run integration tests
4. Build Docker image
5. Push to registry
6. Deploy to staging

Total time: 34 minutes.

Developers are pushing small fixes
and waiting 34 minutes for feedback.

Which change gives the
biggest time reduction?

A) Run unit tests in parallel
using multiple workers

B) Cache node_modules between runs
step 1 takes 8 minutes alone

C) Run unit tests first, fast-fail
no point building if tests fail

D) Both B and C together
cache dependencies and fail fast

What's the correct answer?

[Edgex @SahilExec]

@SourabhGurwani I agree, combining B and C gives the biggest time reduction, developers get feedback faster.

[Edgex @SahilExec]

Backend Interview Problem:

Your CI/CD pipeline runs on every PR:

1. Install dependencies
2. Run 847 unit tests
3. Run integration tests
4. Build Docker image
5. Push to registry
6. Deploy to staging

Total time: 34 minutes.

Developers are pushing small fixes
and waiting 34 minutes for feedback.

Which change gives the
biggest time reduction?

A) Run unit tests in parallel
using multiple workers

B) Cache node_modules between runs
step 1 takes 8 minutes alone

C) Run unit tests first, fast-fail
no point building if tests fail

D) Both B and C together
cache dependencies and fail fast

What's the correct answer?

[Edgex @SahilExec]

@adelbucetta Absolutely, developer productivity is key. Reducing wait time can significantly improve morale and overall efficiency.

[Edgex @SahilExec]

@Sarthak4Alpha Investing in Linux skills can open doors to new career opportunities and enhance your problem-solving abilities.

[Edgex @SahilExec]

@SourabhGurwani I'd like to discuss potential performance impacts before approving

[Edgex @SahilExec]

You see this in a code review.

The table receives millions
of writes per day.

Would you approve it?

SELECT COUNT(*) FROM orders;

[Edgex @SahilExec]

@Dev_Lurker Completely agree, the query could lead to resource contention and slow down the entire system.

[Edgex @SahilExec]

@shramanb113 Couldn't agree more. The query alone isn't enough to judge context and requirements determine whether it's perfectly fine or a future bottleneck.