chore(electric): downsize Fly VM from performance-4x to shared-cpu-2x

[saddlepaddle]

Summary

• Downsizes superset-electric Fly VM from performance-4x (4 dedicated CPUs, 8GB RAM, ~$124/mo) to shared-cpu-2x (2 shared CPUs, 512MB, ~$4/mo)
• The machine handles ~200 req/min — massively over-provisioned at current size
• This was also causing manual fly machines update resizes to revert on every deploy

Test plan

• Merge and verify deploy sets the correct VM size
• Monitor Electric SQL health checks post-deploy

---

Summary by cubic

Downsizes the Fly VM for the superset-electric proxy from performance-4x (4 dedicated CPUs, 8GB) to shared-cpu-2x (2 shared CPUs, 512MB) to match ~200 req/min load and save about $120/month. Sets the size in fly.toml so deploys use the intended config and stop reverting manual resizes.

^{Written for commit f46e402. Summary will update on new commits.}

Summary by CodeRabbit

• Chores
  • Updated deployment infrastructure resource configuration.

[coderabbitai]

📝 Walkthrough

Walkthrough

The fly.toml configuration file is updated to reduce VM resource allocation: memory decreases from 8192mb to 512mb, CPU kind changes from performance to shared, and CPU count reduces from 4 to 2.

Changes

Cohort / File(s)	Summary
Fly VM Configuration fly.toml	Reduced VM resource allocation: memory from 8192mb to 512mb, CPU kind from performance to shared, CPUs from 4 to 2.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Poem

🐰 A rabbit hops through config land,
Where resources shrink by my small hand,
Four CPUs down to but two,
Eight gigs to five-twelve—a downsizing brew,
Performance traded for shared delight,
Fly's VM now lean and light! ✨

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately and concisely describes the main change: downsizing the Fly VM configuration from a high-performance to shared CPU instance.
Description check	✅ Passed	The description covers the key context (current over-provisioning, cost savings, deployment issue) but lacks explicit sections matching the template structure (Type of Change, Testing details).
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch saddlepaddle/downsize-electric-vm

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[cubic-dev-ai]

No issues found across 1 file

[coderabbitai]

🧹 Nitpick comments (1)

fly.toml (1)

8-10: Consider a more conservative configuration for the initial downsize.

Given the magnitude of the resource reduction and the single-machine setup (line 21), consider starting with more conservative settings to reduce the risk of service disruption:

• Memory: Start with 1024mb or 2048mb rather than 512mb
• Concurrent requests: Reduce ELECTRIC_MAX_CONCURRENT_REQUESTS proportionally (e.g., to {"initial": 1000, "existing": 3000})

You can further optimize after monitoring actual resource usage at the intermediate tier. This staged approach provides a safety buffer while still achieving significant cost savings (~$62/mo for 1GB shared-cpu-2x vs ~$124/mo for performance-4x).

Also applies to: 14-14

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@fly.toml` around lines 8 - 10, Update the fly.toml resource and environment
settings to use a more conservative initial downsize: change memory from "512mb"
to "1024mb" (or "2048mb") and keep cpu_kind "shared" with cpus 2, and reduce the
ELECTRIC_MAX_CONCURRENT_REQUESTS JSON to something like {"initial": 1000,
"existing": 3000} so the first-tier concurrency is lower; locate and edit the
memory/cpu_kind/cpus fields and the ELECTRIC_MAX_CONCURRENT_REQUESTS env var in
the same manifest to apply these safer defaults for the single-machine
deployment.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@fly.toml`:
- Around line 8-10: Update the fly.toml resource and environment settings to use
a more conservative initial downsize: change memory from "512mb" to "1024mb" (or
"2048mb") and keep cpu_kind "shared" with cpus 2, and reduce the
ELECTRIC_MAX_CONCURRENT_REQUESTS JSON to something like {"initial": 1000,
"existing": 3000} so the first-tier concurrency is lower; locate and edit the
memory/cpu_kind/cpus fields and the ELECTRIC_MAX_CONCURRENT_REQUESTS env var in
the same manifest to apply these safer defaults for the single-machine
deployment.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5deeff91-38fb-4ca0-bcba-5f545a871324

📥 Commits

Reviewing files that changed from the base of the PR and between 2153a95 and f46e402.

📒 Files selected for processing (1)

• fly.toml

[github-actions]

🧹 Preview Cleanup Complete

The following preview resources have been cleaned up:

• ✅ Neon database branch
• ✅ Electric Fly.io app

Thank you for your contribution! 🎉