I'm a full-stack developer, and lately I've become really interested in understanding how complex software systems work under the hood.I'd like to learn how to build things like message brokers, log and trace collectors, databases, and other infrastructure tools that developers use every day but don't usually think much about internally.I'm wondering what I should learn to get into this kind of development. Is this considered systems programming, or does it fall more into areas like distributed systems and infrastructure engineering? I'd also appreciate any recommendations for books, courses, or practical projects that could help me build the necessary knowledge from the ground up. I'm currently learning C, so I'm curious whether systems programming is the right direction and what topics I should focus on next.

If I use S3 for storing images, should I store only the object keys in the database and generate a presigned url every time I fetch the records to return to the client? How efficient is this approach for something like a product catalog where each product can have multiple images and I would need to generate several presigned urls every time the catalog is requested?

I've seen some developers say that we should stop using js on the server, but I rarely see the same criticism directed at python.Honestly, js is pretty fast, especially with node.js, and the backend ecosystem is really solid. There are great frameworks like nest and fastify, and typescript gives you static typing, which makes larger projects much easier to maintain.I'm not saying node is perfect or the best choice for every backend, but I don't really understand why js gets so much criticism while python seems to get a pass.What am I missing?

The longer I use linux, the more I appreciate it. I switched from windows a while ago, and I honestly haven't regretted it for a second.I'm a developer, and every time I learn a new command or discover another tool that makes my workflow easier, I realize how powerful and well-designed this operating system is. It feels like there's always something new to learn, and instead of making me frustrated, it just makes me enjoy using it even more. At this point, going back to windows for my daily development work would feel like a huge step backward.

One of the reasons I'm thinking about leaving web development is AI. I know it's a tool, not a replacement, but I can't shake the feeling that it will have a much bigger impact on web development than on areas like network security.I'm 26 with a full stack background, mostly backend, and I'm considering switching to cybersecurity. My plan is to start with network security and eventually move into DevSecOps or Cloud Security.Does this seem like a reasonable path? Has anyone here made a similar transition, especially for similar reasons?

I'm a software developer, not planning to become a system administrator, but I'd like to get a much deeper understanding of Linux. I noticed that the latest edition covers modern topics like systemd, containers, and cloud infrastructure, so it doesn't seem completely outdated.Would this book still be a good investment for a developer who wants to better understand how Linux systems work, or is it more useful for people working as full-time sysadmins? I'm looking for knowledge that will make me a better developer rather than preparing for a system administration career.

https://preview.redd.it/cstt2h1s7zgh1.png?width=367&format=png&auto=webp&s=16345dfec5aa74bb4baea1e4a4f8ff246b7a8445

I've noticed that I get really tired after eating oatmeal or buckwheat for breakfast or lunch. I thought they were slow-digesting carbs, so they shouldn't cause a big blood sugar spike. I decided to remove them from my meals, and the tiredness completely went away. What could explain this? Has anyone experienced something similar?

I've been taking 5g of creatine every morning, and I've noticed that my sleep has gotten much worse. Some nights I struggle to fall asleep, and on other nights I barely sleep at all. Everything else in my routine has stayed the same, so creatine is the only thing that stands out. Has anyone else experienced this, or could it just be a coincidence?

I'm currently taking Now Omega 3 and realized it's in ethyl ester form rather than triglyceride form. Should I be concerned about that, or is the difference mostly theoretical?Would you keep using Now, or is it worth paying more for a triglyceride or re-esterified triglyceride product? I'm interested in what the research says and whether anyone has noticed a real difference after switching.

https://preview.redd.it/uxug8wkuh48h1.png?width=2194&format=png&auto=webp&s=dd5200084b7300517bbee0b35c35ba1d52c46130

https://preview.redd.it/x82n01bwh48h1.png?width=1562&format=png&auto=webp&s=58abe2e8c1eee3202befbc66f4462213e680ac1f

Got a mapping issue with EF Core and PostgreSQL. I’m loading an Order entity using FromSqlInterpolated with FOR UPDATE to lock the row, and the entity includes owned/complex types like billing and delivery address.

When I run the query, PostgreSQL throws a 42703 undefined column error for something like BillingAddress_Apartment, even though everything works correctly with normal LINQ queries.

It looks like EF is applying its own expected column mapping for owned types during materialization, and the SQL result from SELECT * doesn’t match that internal shape. I’m not sure why this only breaks with raw SQL or what the correct way is to combine row locking with full entity loading without hitting these mapping issues.

Hi, I’m building an e-commerce backend and ran into a tricky case.

When a user creates an order, I immediately reserve stock by decrementing product sku quantity and store order items with quantity + price snapshot. Each order has an expires_at timestamp. If the order is not paid in time, a background job runs every minute, finds expired orders, marks them as cancelled, and returns reserved stock back to inventory.

The issue is a race condition between payment and expiration. A user can click “pay” at the last moment while the order is still valid, but right after payment succeeds, the background job might already have marked the order as cancelled and restored stock. So payment succeeds, but the order becomes cancelled and inventory was incorrectly released.

I considered re-activating the order or recreating it as paid when the webhook arrives, but then there’s another problem: what if the stock was already taken by someone else after it was released during expiration processing?

What is the correct way to handle this kind of flow? Is the proper solution strict atomic state transitions in the database, or is there a standard pattern for combining stock reservation, payment confirmation, and expiration safely?

Hi, I’m building an e-commerce backend and ran into a tricky case.

When a user creates an order, I immediately reserve stock by decrementing product sku quantity and store order items with quantity + price snapshot. Each order has an expires_at timestamp. If the order is not paid in time, a background job runs every minute, finds expired orders, marks them as cancelled, and returns reserved stock back to inventory.

The issue is a race condition between payment and expiration. A user can click “pay” at the last moment while the order is still valid, but right after payment succeeds, the background job might already have marked the order as cancelled and restored stock. So payment succeeds, but the order becomes cancelled and inventory was incorrectly released.

I considered re-activating the order or recreating it as paid when the webhook arrives, but then there’s another problem: what if the stock was already taken by someone else after it was released during expiration processing?

What is the correct way to handle this kind of flow? Is the proper solution strict atomic state transitions in the database, or is there a standard pattern for combining stock reservation, payment confirmation, and expiration safely?

Hi, I’m building an e-commerce backend and ran into a tricky case.

When a user creates an order, I immediately reserve stock by decrementing product sku quantity and store order items with quantity + price snapshot. Each order has an expires_at timestamp. If the order is not paid in time, a background job runs every minute, finds expired orders, marks them as cancelled, and returns reserved stock back to inventory.

The issue is a race condition between payment and expiration. A user can click pay at the last moment while the order is still valid, but right after payment succeeds, the background job might already have marked the order as cancelled and restored stock. So payment succeeds, but the order becomes cancelled and inventory was incorrectly released.

I considered re-activating the order or recreating it as paid when the webhook arrives, but then there’s another problem: what if the stock was already taken by someone else after it was released during expiration processing?

What is the correct way to handle this kind of flow? Is the proper solution strict atomic state transitions in the database, or is there a standard pattern for combining stock reservation, payment confirmation, and expiration safely?

The longer I work in backend development, the more I think databases are the area worth studying the deepest. Languages, frameworks, and architectural trends come and go, but almost every backend system ultimately depends on how data is stored, queried, and managed. Understanding raw SQL is important, but so is learning indexing, query optimization, transactions, locking, schema design, normalization, denormalization, data modeling, partitioning, replication, and consistency trade-offs. I've noticed that many performance, scalability, and reliability problems often lead back to database decisions made early in a project. In contrast, a well-designed database can make the rest of the system significantly simpler. If a backend developer had limited time to become an expert in just one area, databases would be a strong candidate. Curious whether others feel the same or would choose something different.

The longer I work in backend development, the more I think databases are the area worth studying the deepest.

Languages, frameworks, and architectural trends come and go, but almost every backend system ultimately depends on how data is stored, queried, and managed. Understanding raw SQL is important, but so is learning indexing, query optimization, transactions, locking, schema design, normalization, denormalization, data modeling, partitioning, replication, and consistency trade-offs.

I've noticed that many performance, scalability, and reliability problems often lead back to database decisions made early in a project. In contrast, a well-designed database can make the rest of the system significantly simpler.

If a backend developer had limited time to become an expert in just one area, databases would be a strong candidate. Curious whether others feel the same or would choose something different.

The longer I work in backend development, the more I think databases are the area worth studying the deepest. Languages, frameworks, and architectural trends come and go, but almost every backend system ultimately depends on how data is stored, queried, and managed. Understanding raw SQL is important, but so is learning indexing, query optimization, transactions, locking, schema design, normalization, denormalization, data modeling, partitioning, replication, and consistency trade-offs. I've noticed that many performance, scalability, and reliability problems often lead back to database decisions made early in a project. In contrast, a well-designed database can make the rest of the system significantly simpler. If a backend developer had limited time to become an expert in just one area, databases would be a strong candidate. Curious whether others feel the same or would choose something different.

I keep seeing clean architecture recommended for almost every .NET project, but sometimes it feels like unnecessary complexity.For example, EF Core is already an abstraction over the database, yet many projects add another repository layer on top of it. At that point it feels like we're hiding useful EF Core features behind our own abstractions and writing extra code just for the sake of following a pattern.The same applies to other services. We often wrap everything behind interfaces even when there is only one implementation and no realistic chance of replacing it.Years ago the argument was usually testability and mocking dependencies. But today it's easy to spin up real databases and services with tools like testcontainers, so integration tests are often more valuable than mocking everything.

Another thing I've noticed is that real enterprise development is rarely as clean as architecture diagrams make it look. Deadlines, legacy systems, performance requirements, third-party integrations, and business demands often force you to bend or completely break some architectural rules. In many projects, the "perfect" architecture only exists on paper, while the actual codebase evolves around practical constraints.I'm not saying сlean architecture is useless, but sometimes it feels like developers spend more time maintaining layers and abstractions than solving business problems.

Do you think clean architectureis still worth the added complexity for most projects, or has it become something developers apply by default without questioning the trade-offs?

I'm a full-stack developer, and recently I've become curious about how low-level software is built.Things like runtimes, compilers, operating systems, databases, virtual machines, and similar projects seem really interesting to me. I use a lot of software every day, but I don't have a good understanding of how these systems actually work internally.I'd like to learn the fundamentals and eventually build some small projects myself to better understand these concepts.For someone coming from a web development background, what would be the best way to get started? Are there any books, courses, articles, or open-source projects that helped you learn this area of programming?

There is a situation where an entity has images stored in S3, but saving the entity in the database and uploading files to S3 are not atomic operations. This can lead to cases where files are successfully uploaded to S3, but the database operation fails while creating the entity. As a result, orphaned files remain without any reference in the database.The outbox pattern doesn’t really help here since the files themselves can’t be part of the database transaction. I’m trying to figure out the cleanest way to handle this kind of consistency issue. Maybe some form of compensation flow, delayed cleanup, or background reconciliation, but I’m not sure what the standard approach is in real systems.How do you usually solve this kind of problem?

I’ve written backend services with Node.js, Go, and C#/.NET over the last few years, and every stack had things I liked. Node feels fast to prototype with, go is simple and lightweight, but every time I come back to .net it just feels really enjoyable to work with.The tooling is excellent, the ecosystem feels mature, dependency injection is built in, and the overall developer experience feels very polished. There’s something about the structure and consistency of .net projects that makes development feel smooth instead of chaotic.Performance is also insanely good now, so it doesn’t feel like you sacrifice speed for developer productivity anymore.Didn’t expect to enjoy backend development in .net this much after spending time in other ecosystems. Anyone else had the same experience?