perf : optimize gms startup

[Dutt23]

Adds WAR extraction to tmpfs (RAM disk) for 15-30% faster startup. Bypasses OverlayFS overhead by extracting to RAM before class loading. Uses Spring Boot's classpath.idx for deterministic,
ordered class loading across boots.

Changes

Docker (docker/datahub-gms/start.sh)

• Added production-ready logging with WAR size and available RAM checks
• Warns if WAR > 1Gi (tmpfs limit) or RAM < 500MB
• Extracts WAR to /tmp/gms/extaction when EXTRACT_JAR_ENABLED=true
• Uses Spring's classpath.idx to build deterministic, ordered classpath

Kubernetes Helm Chart

• Added extractJarEnabled: false flag to values.yaml (disabled by default)
• Added conditional tmpfs emptyDir volume (1Gi, Memory-backed) in deployment.yaml
• Added EXTRACT_JAR_ENABLED environment variable to container spec
• Added conditional volumeMount at /tmp/gms/extaction

Why This Matters

OverlayFS (Docker/K8s default) adds copy-on-write overhead to every JAR class load. Extracting to tmpfs bypasses this layer traversal entirely:

• Traditional: Read WAR from OverlayFS → decompress nested JARs → extract classes (slow)
• Optimized: Extract to RAM once → load classes directly (fast)

Performance

• Startup time local : 15-30% faster (30-35s → 16-18s on local)
• Startup time deployment : (34-36 -> 28-30s) In deployments we have a bunch of tasks running on bootstrap loader causes interference.
• Class loading: 2-3× faster via RAM + deterministic classpath ordering
• Consistency: Eliminates OverlayFS cache variance for predictable SLAs
• Memory: Temporary +150-300MB (freed after startup)

Requirements

• Spring Boot 3.2+ (for classpath.idx support)
• 8GB+ available cluster RAM
• WAR file < 500MB (default 1Gi tmpfs limit)

Startup Output

[STARTUP] JAR extraction enabled. WAR size: 561MB, Available RAM: 3578MB
[STARTUP] Extracting WAR to tmpfs: /tmp/gms/extraction
[STARTUP] WAR extracted in 768ms
[STARTUP] Generating deterministic classpath from BOOT-INF/classpath.idx
[STARTUP] Deterministic classpath: 660 entries (from classpath.idx)

See docs/jar-extraction-tmpfs-optimization.md for detailed documentation.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!

[alwaysmeticulous]

✅ Meticulous spotted 0 visual differences across 1008 screens tested: view results.

Meticulous evaluated ~8 hours of user flows against your PR.

_{Expected differences? Click here. Last updated for commit 39fcac7. This comment will update as new commits are pushed.}

[codecov]

Bundle Report

Bundle size has no change ✅

[codecov]

Bundle Report

Bundle size has no change ✅