feat: derive all upload limits from single APP_MEMORY_MB config

[NotYuSheng]

Summary

• Replaces MAX_UPLOAD_SIZE_BYTES with a single APP_MEMORY_MB knob in .env that automatically derives every related limit at container startup
• Fixes the 500MB upload failure root causes: no JVM heap set, nginx body limit too tight, proxy timeouts too short

How it works

APP_MEMORY_MB	JVM heap (-Xmx)	Max upload	Nginx body limit	Proxy / analysis timeout
2048 (default)	1536 MB	512 MB	562 MB	900 s
4096	3072 MB	1 GB	1074 MB	900 s
8192	6144 MB	2 GB	2098 MB	900 s

Formulas (all computed in the entrypoint scripts):

• JVM heap = 75% of APP_MEMORY_MB
• Max upload = 25% of APP_MEMORY_MB
• Nginx body limit = max upload + 50 MB (multipart overhead buffer)
• Timeout = 45% of APP_MEMORY_MB, clamped to 300–900 s

Changes

• backend/docker-entrypoint.sh (new) — shell script that computes and exports all values, then exec java -Xmx... -jar app.jar
• backend/Dockerfile — switched from bare java -jar to the entrypoint script
• nginx/docker-entrypoint.sh — rewritten to derive body size and timeouts from APP_MEMORY_MB instead of raw bytes
• nginx/nginx.conf.template — proxy timeouts now templated (${NGINX_PROXY_TIMEOUT})
• application.yml — timeout-seconds reads ANALYSIS_TIMEOUT_SECONDS env var set by the entrypoint
• SystemController (new) — GET /api/system/limits returns { maxUploadBytes, maxUploadMb } at runtime
• UploadPage — fetches /api/system/limits on mount; upload zone displays the live limit instead of a build-time baked value

Test plan

• Set APP_MEMORY_MB=4096 in .env, run docker compose build && docker compose up -d
• Backend logs show Max upload size = 1024 MB, JVM heap = 3072 MB
• Nginx logs show Nginx body limit = 1074M, Proxy timeout = 900 s
• Upload page shows "max 1024MB" in the drop zone
• Default (APP_MEMORY_MB=2048) still shows "max 512MB"

🤖 Generated with Claude Code

[gemini-code-assist]

Code Review

This pull request centralizes system configuration by deriving JVM heap size, upload limits, and timeouts from a single APP_MEMORY_MB environment variable. It introduces entrypoint scripts for the backend and Nginx to perform these calculations and adds a new API endpoint for the frontend to retrieve upload limits dynamically. Review feedback suggests setting the initial JVM heap size equal to the maximum for performance, adding error logging to the frontend's limit-fetching logic, and deduplicating the timeout calculation logic shared between the backend and Nginx entrypoint scripts.

[NotYuSheng]

/gemini review

[gemini-code-assist]

Code Review

This pull request introduces a centralized configuration model where system limits, such as JVM heap size, max upload size, and timeouts, are dynamically derived from a single APP_MEMORY_MB environment variable. Key improvements include enhanced memory management during PCAP analysis through JPA session flushing and batching, the addition of a system limits API to inform the frontend, and a more informative analysis loading UI with progress estimation. A suggestion was made to refactor the packet counting logic in PcapParserService to improve readability and maintainability.

[NotYuSheng]

Addressed all 4 review comments: fixed -Xms/-Xmx parity and added console.error for the limits fetch error. Declined the DRY entrypoint suggestion (separate containers, no shared filesystem) and the packetNumber refactor (pre-increment + decrement on EOF is less readable than the current two-path approach).