ทวีตที่ปักหมุด
Have you ever felt captivated inside yourself? Captivated inside the cage of wrong habits, indiscipline, failure. Trying hard to come out like a bee trapped in spiderweb, but failing to do so. I am in the same position rn.
English
Ashmit JaiSarita Gupta
2K posts
Ashmit JaiSarita Gupta
@ashmitjsg
Post on Software Development, Quantum Computing and College Experience. | GSoC 24 @AsyncAPISpec | Ex-Quantum Computing Intern at @Creed_Bear Dubai | NITH'25
Azamgarh, India เข้าร่วม Mart 2021
579 กำลังติดตาม715 ผู้ติดตาม
Ashmit JaiSarita Gupta รีทวีตแล้ว
60 days of crude gives some breathing space.
But risks sit beyond storage.
➡️Shipping routes can get hit or blocked
➡️Insurance and freight costs can jump overnight
➡️Any disruption in the Gulf hits us first
We need to push harder on supply side.
✅Faster clearances for domestic exploration
✅Scale up offshore drilling on east and west coast
✅Take more equity in oil fields abroad
✅Build stronger naval cover for key sea lanes
✅Expand strategic reserves beyond current levels
📌Energy security needs constant expansion, not comfort.
📌Are we moving fast enough on extraction and access?
English
Crude supplies have been secured for the next 60 days, say Indian officials
English
Ashmit JaiSarita Gupta รีทวีตแล้ว
Ashmit JaiSarita Gupta รีทวีตแล้ว
Every Kafka serializer I write follows the same pattern:
JsonMapper.builder().findAndAddModules().build()
The findAndAddModules() call picks up JavaTimeModule automatically. This handles Instant and LocalDateTime serialization without manual registration.
Ashmit JaiSarita Gupta@ashmitjsg
Chose to write custom Kafka serializers instead of using Spring's built-in JsonSerializer. Spring Boot 4 uses Jackson 3 (tools.jackson.*), which has API differences from Jackson 2. Custom serializers are explicit, predictable, and debuggable. No magic. No version conflicts.
English
Chose to write custom Kafka serializers instead of using Spring's built-in JsonSerializer.
Spring Boot 4 uses Jackson 3 (tools.jackson.*), which has API differences from Jackson 2.
Custom serializers are explicit, predictable, and debuggable. No magic. No version conflicts.
Ashmit JaiSarita Gupta@ashmitjsg
Setting up the Kafka producer today. Key config decisions: - acks=all: wait for leader + replicas to confirm write - enable.idempotence=true: no duplicate messages on retry - retries=3: transient failures handled automatically This is the minimum config for a reliable producer.
English
Setting up the Kafka producer today.
Key config decisions:
- acks=all: wait for leader + replicas to confirm write
- enable.idempotence=true: no duplicate messages on retry
- retries=3: transient failures handled automatically
This is the minimum config for a reliable producer.
Ashmit JaiSarita Gupta@ashmitjsg
Used Lombok @Builder + final fields for the Kafka event DTO. Builder pattern avoids constructor explosion as the schema evolves. Immutability prevents accidental field mutation between creation and publishing. For request DTOs (user input), I keep them mutable for Jackson.
English
Ashmit JaiSarita Gupta รีทวีตแล้ว
Used Lombok @Builder + final fields for the Kafka event DTO.
Builder pattern avoids constructor explosion as the schema evolves. Immutability prevents accidental field mutation between creation and publishing.
For request DTOs (user input), I keep them mutable for Jackson.
Ashmit JaiSarita Gupta@ashmitjsg
Designing the RawLogEvent DTO that gets published to Kafka. Fields: eventId (UUID), tenantId, serviceName, environment, receivedAt (server time), logTimestamp (client time), level, message, trace. Two timestamps on purpose. The discrepancy tells you how delayed a log was.
English
Designing the RawLogEvent DTO that gets published to Kafka.
Fields: eventId (UUID), tenantId, serviceName, environment, receivedAt (server time), logTimestamp (client time), level, message, trace.
Two timestamps on purpose. The discrepancy tells you how delayed a log was.
Ashmit JaiSarita Gupta@ashmitjsg
Built a RequestHeaderExtractor utility today. Instead of reading X-TENANT-ID in every controller method, a single utility extracts all required headers into an immutable RequestContext object. One place to add new required headers. One place to fix validation. Clean. #java
English
Having some fun in exploring ascii art , gotta say these are my favourite
English
Built a RequestHeaderExtractor utility today.
Instead of reading X-TENANT-ID in every controller method, a single utility extracts all required headers into an immutable RequestContext object.
One place to add new required headers. One place to fix validation. Clean.
#java
Ashmit JaiSarita Gupta@ashmitjsg
One thing I was deliberate about: the ingestion endpoint returns 202, not 200. 202 means "accepted for processing" — the log is in Kafka, but downstream processing hasn't happened yet. Semantically honest. Don't tell the client something is done when it isn't. #api
English
One thing I was deliberate about: the ingestion endpoint returns 202, not 200.
202 means "accepted for processing" — the log is in Kafka, but downstream processing hasn't happened yet.
Semantically honest. Don't tell the client something is done when it isn't.
#api
Ashmit JaiSarita Gupta@ashmitjsg
Designed the ingestion API contract today. POST /api/v1/logs Required headers: X-TENANT-ID, X-SERVICE-NAME, X-ENVIRONMENT, X-API-KEY Body: timestamp, level, message, trace (optional) Response: 202 Accepted (async — we don't wait for Kafka confirmation) #api #backend
English
Designed the ingestion API contract today.
POST /api/v1/logs
Required headers: X-TENANT-ID, X-SERVICE-NAME, X-ENVIRONMENT, X-API-KEY
Body: timestamp, level, message, trace (optional)
Response: 202 Accepted (async — we don't wait for Kafka confirmation)
#api #backend
Ashmit JaiSarita Gupta@ashmitjsg
Starting the first service: ingestion-gateway. Responsibility: accept logs from client services via HTTP, validate them, and publish to Kafka. That's it. Single responsibility. No business logic, no storage, no auth complexity — just reliable ingestion. #microservices #java
English
Starting the first service: ingestion-gateway.
Responsibility: accept logs from client services via HTTP, validate them, and publish to Kafka.
That's it. Single responsibility. No business logic, no storage, no auth complexity — just reliable ingestion.
#microservices #java
Ashmit JaiSarita Gupta@ashmitjsg
One of the first design decisions: what should the Kafka partition key be? Options: eventId (random), serviceName, or tenantId. Chose tenantId. This guarantees all logs from one customer are processed in order. Critical for time-window clustering later. #kafka #systemdesign
English
One of the first design decisions: what should the Kafka partition key be?
Options: eventId (random), serviceName, or tenantId.
Chose tenantId. This guarantees all logs from one customer are processed in order. Critical for time-window clustering later.
#kafka #systemdesign
Ashmit JaiSarita Gupta@ashmitjsg
Set up the local Kafka stack today using Docker Compose. Confluent's cp-kafka + Zookeeper. Single broker, 1 replication factor for local dev. The first topic I created: raw-logs. Every log from every service on the platform will flow through this topic first. #kafka #backend
English
Set up the local Kafka stack today using Docker Compose.
Confluent's cp-kafka + Zookeeper. Single broker, 1 replication factor for local dev.
The first topic I created: raw-logs. Every log from every service on the platform will flow through this topic first.
#kafka #backend
Ashmit JaiSarita Gupta@ashmitjsg
Naming the project: log0 - An intelligent incident copilot that turns raw logs into actionable incidents. It ingests high-volume logs from distributed services, normalises and clusters similar errors, and automatically creates incidents when recurring failures are detected.
English