Your VPS bill jumped, or a price-adjustment page will not say if your box moved. A bigger RAM tier looks safer than measuring the stack you already run.
By Sudipta Dhara. Drafted 2026-10-19. AI disclosure: drafted with AI assistance.
Affiliate disclosure: Some links below are paid (affiliate) links. If you buy through them, ProDevs may earn a commission. If you become a DigitalOcean customer through our link, we may earn a commission. ProDevs does not run production on DigitalOcean. Details: /advertise/.
First check who the hike actually hits, then size from measured container RSS — not the order-form RAM label. This does not apply when you already have production docker stats and free -h under real peak traffic; use those figures instead.
Who this is for: developers and DevOps engineers on one Docker VPS who just saw a price email or a reseller comparison table · Intermediate · You’ll need: SSH, docker stats, and the provider’s public pricing or price-adjustment page open on the same day you decide.
When a VPS provider raises prices, confirm whether your instance is in the affected set, then pick the smallest plan above measured RSS plus headroom. Price tables change; your measured floor does not. We do not claim Contabo ran a 2026 VPS increase — we only quote their public USD list price from the day we read it.
Shareable fact
| Reading | Result | Caveat |
|---|---|---|
Idle postgres:16-alpine RSS |
17.34MiB inside a 512MiB cap | Published ProDevs lab; defaults image only (Postgres idle RAM) |
| Memory-cap planning rule | ~1.5× steady docker stats usage |
Same value for --memory and --memory-swap (memory limits) |
| Contabo Cloud VPS 4 USD list (schema) | USD 6.6 lowPrice · 8 GB RAM / 4 vCPU | Captured 19 Oct 2026 IST from contabo.com/en/vps; not a hike claim |
Sibling sizing notes: Docker or a VPS, VPS sizing notes, Docker disk space full.
Who is affected when a provider “raises prices”?
Providers do not all move every running box on the same day. Read the official note for new orders, rescales, and legacy servers before you migrate.
Hetzner Cloud (documented adjustment). Hetzner’s price-adjustment page states the change took effect for new orders and cloud instance rescales starting 15 June 2026, 8 AM CEST. Orders placed before that date but delivered after still use the previous prices. (Hetzner Price Adjustment 15 June 2026)
Their billing FAQ is sharper on legacy boxes: a price adjustment occurs for rescaling (up or down), restoring a deleted server, or transferring a server to a project with a different currency. It does not occur for transferring within the same currency or for rebuilding a server. (Hetzner Cloud billing FAQ)
Contabo (list price only — no hike claim). We did not find an official Contabo 2026 VPS increase notice to cite. On 19 Oct 2026 IST the public Contabo VPS page schema listed Cloud VPS 4 at USD lowPrice 6.6 (4 vCPU, 8 GB RAM) alongside a EUR offer of 5.5. Quote the currency you will pay; do not invent a year-over-year delta. (Contabo VPS)
If you are shopping Contabo’s public VPS catalog after you finish measuring, re-check the live USD number on Contabo’s site the day you buy, or use our tracked Contabo VPS page (paid link). Official catalog: contabo.com/en/vps.
DigitalOcean (dated list paste). Public Basic Droplet monthly strings we already captured for sizing context on 19 Oct 2026 IST include 1 GiB $6.00/mo and 2 GiB $12.00/mo. Those are list prices, not a ProDevs discount. Re-check the live list on our tracked DigitalOcean Droplet pricing page (paid link).
How do you size the next plan from measured Docker RAM?
Docker can limit container memory with resource constraints. We read that page on 19 Oct 2026. (Docker docs: resource constraints)
- Sample idle and light load:
docker stats --no-streamand hostfree -h. - Take the steady RSS (median if you sample), not a one-off spike.
- Cap each service near 1.5× that steady reading with
--memoryand--memory-swapequal, then re-run the job and checkOOMKilled. (Docker memory limits) - Map RSS sum + OS/cache headroom to the smallest public tier above that floor. Idle Postgres alone is not host size: our published defaults reading was 17.34MiB RSS under a 512MiB cap while
SHOW shared_buffersprinted 128MB. (Postgres idle RAM)
On a disposable lab stack (19 Oct 2026 IST, Docker Engine 29.8.2, Linux aarch64, 7.8 GiB host) a pinned nginx:1.27-alpine + postgres:16-alpine + tiny Python HTTP stack showed load median RSS 40.11 MiB (idle sum 36.94 MiB). That floor is a method check, not a shopping recommendation by itself — leave headroom for OS, page cache, and growth (≥1 GiB public tier as a planning start for that demo; prefer more if Postgres holds real data).
docker stats --no-stream
free -h
docker run --rm -d --name web \
--memory=64m --memory-swap=64m \
nginx:1.27-alpine
Who it’s for
Operators who run Compose or a few containers on one VPS, just got a price email or opened a comparison tab, and need a keep / right-size / move decision the same day. You can SSH and read docker stats.
Not for
Not for multi-node Kubernetes capacity planning, GPU instances, Windows VPS, or anyone hunting an unsourced “Contabo hiked in 2026” story. Not a promise that Hetzner, Contabo, or DigitalOcean will keep today’s list prices next month.
Alternatives
- Stay on the current plan: prune images/logs first (disk tutorial), add memory limits, re-measure.
- Right-size in place if the provider lets you rescale — but on Hetzner, rescaling a legacy-priced cloud server is an action that moves you to current pricing. (billing FAQ)
- Move providers only after a same-day list-price paste beats the hiked bill with headroom intact and a restore-tested backup.
Cons / drawbacks
- Public list prices go stale within days; a draft table is not a live cart.
- A paid Contabo link still needs a visible disclosure; list prices on Contabo’s site can differ from the schema snapshot in this draft.
- Lab RSS floors understate production Postgres with real data and connection counts.
- Hetzner’s “who is affected” rules are easy to misread — rescaling to “save money” can end the legacy rate.
What broke when we sized from the marketing label first?
We almost treated a ~40 MiB container RSS sum as permission to buy the cheapest tiny Droplet on a comparison page. That ignored OS, cache, and Postgres growth, and it skipped the provider question of whether a rescale would change the rate. Reading Hetzner’s rescale rule, quoting Contabo’s USD schema price without inventing a hike, and anchoring size to published idle RSS + the 1.5× cap rule fixed the decision order: affected set → measure → paste today’s list price → keep / right-size / move.
Raw readings
Captured or published figures that do not change the recommendation above.
- Contabo Cloud VPS 4 EUR schema
lowPrice5.5 on the same 19 Oct 2026 IST page read (USD 6.6 is the figure this draft shops against). - Hetzner price-adjustment tables also list many CX/CAX/CCX cells; use the live doc the day you buy, not a mirrored sheet.
- DigitalOcean Basic 4 GiB
$24.00/mo and 8 GiB$48.00/mo from the same 19 Oct paste window. - Disposable-lab idle per-container medians (MiB): nginx 6.76, postgres 18.59, app 11.59 — method only.
FAQ
Did Contabo publish a 2026 VPS price increase?
We found no official Contabo 2026 VPS increase notice to cite. This page quotes Contabo’s public Cloud VPS 4 USD list price from contabo.com on 19 Oct 2026 only. Do not treat third-party hike claims as fact.
Who is affected by Hetzner’s 15 June 2026 price adjustment?
Hetzner’s docs say the adjustment applies to new orders and cloud instance rescales from 15 June 2026, 8 AM CEST. Rescaling up or down moves a legacy-priced server to current pricing. Same-currency project transfers and rebuilds do not.
How much RAM do I need for Docker Postgres on a small VPS?
On our published idle reading, postgres:16-alpine at defaults used 17.34MiB inside a 512MiB Docker cap while SHOW shared_buffers printed 128MB. Treat shared_buffers as PostgreSQL’s setting, not idle RSS or host size.
How should I set container memory limits after a price hike?
Set –memory and –memory-swap to the same value at about 1.5× the steady docker stats usage from ordinary work, then prove the job under that cap. Raise both together if a real job needs more.
Open your provider’s price or adjustment page and your docker stats output side by side. Write down whether a rescale would change your rate, then pick the smallest tier above measured RSS plus headroom.
Need a fixed-scope pass to right-size or harden the host? ProDevs DevOps and CI setup. Prefer a command explainer while you dig through docker stats? Talk2Term.
