Skip to content

tf: smaller supernode droplets (s-8vcpu-16gb-480gb-intel) - #2

Merged
qu0b merged 1 commit into
masterfrom
qu0b/smaller-supernodes
Sep 23, 2026
Merged

qu0b merged 1 commit into
masterfrom
qu0b/smaller-supernodes

Conversation

@qu0b

@qu0b qu0b commented Sep 23, 2026

Copy link
Copy Markdown
Member

Every frames-devnet-0 droplet is a supernode: the 6 client nodes set supernode = true, and the bootnode matches the bootnode|mev rule. So digitalocean_supernode_size sizes the whole fleet.

before after
size s-8vcpu-32gb-640gb-intel s-8vcpu-16gb-480gb-intel
vCPU / RAM / disk 8 / 32 GB / 640 GB 8 / 16 GB / 480 GB
fleet (7 droplets) $1,344/mo $896/mo
  • Why 8 vCPU: a supernode custodies every data column, while the EL next to it runs 200M-gas blocks. Only the memory is halved.
  • Availability: the new size exists in all 10 digitalocean_regions (checked via the DO API). The planned placement is ams3, tor1, sgp1, ams3, nyc3, fra1 and sgp1. Plain s-8vcpu-16gb isn't offered in fra1, where lighthouse-reth-1 lands, which rules it out.
  • Checks: terraform fmt and terraform validate pass.

https://claude.ai/code/session_017oM6BuiHivCb3CXVx2kboG

Keeps 8 vCPU for full column custody with 200M-gas EL blocks, halves memory;
available in every region in digitalocean_regions. 7 x $128 instead of 7 x $192/mo.

Claude-Session: https://claude.ai/code/session_017oM6BuiHivCb3CXVx2kboG
@qu0b
qu0b merged commit db6be47 into master Sep 23, 2026
1 check passed

@redpandabot redpandabot Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Summary

Single-line change lowering the supernode droplet default from s-8vcpu-32gb-640gb-intel to s-8vcpu-16gb-480gb-intel. The fleet math in the PR matches the code: all 7 DigitalOcean droplets resolve through digitalocean_supernode_size, since no node entry sets a size override. Only open question is the apply path against an already-running devnet, since the droplet resource pins resize_disk = true and DO cannot shrink disks.

Issues

  • 🟡 terraform/devnet-0/digitalocean.tf:16 — in-place resize of existing 640 GB droplets would fail — DO cannot shrink disk — see the thread on that line

Reviewed @ b23bd977
The word "panda" likely comes from the Nepali "ponya" — and it referred to the red panda first.

variable "digitalocean_supernode_size" {
type = string
default = "s-8vcpu-32gb-640gb-intel"
default = "s-8vcpu-16gb-480gb-intel"

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 in-place resize of existing 640 GB droplets would fail — DO cannot shrink disk

digitalocean_droplet.main sets resize_disk = true, so if this is applied to an existing devnet-0 fleet still running the 640 GB slug, terraform will attempt an in-place resize with a disk decrease, which the DigitalOcean API rejects. Fine for a fresh apply/destroy-recreate cycle; worth confirming the devnet is rebuilt rather than resized in place.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant