Running on Vultr?

Vultr instance down or region misbehaving? We bring it back.

An instance stopped responding to pings last night, a fresh deploy keeps failing because the chosen region is out of the plan size the Terraform run asked for, and a Block Storage volume detached during a reboot and did not come back cleanly. We reach the box through the console or the rescue ISO, work out why the API call for the plan is being rejected, and get the service back without a Sunday evening rebuild.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

vultr-api.log

still broken

What it is

Vultr at a glance

Vultr is a US-based cloud provider founded by David Aninowsky in 2014, best known for cheap Linux and FreeBSD virtual private servers across a large footprint of global regions. The product line has grown beyond VPS to include Bare Metal, Block Storage, Object Storage, Reserved IPs, Load Balancers, Managed Databases, Vultr Kubernetes Engine and an increasingly serious set of GPU instances aimed at AI workloads.

Vultr competes directly with DigitalOcean and Linode on developer ergonomics and price, and with Hetzner on cost per core. The API is clean, Terraform support is first party, and the console is pleasant. The trouble in production is usually around region capacity for less common plans, Block Storage volumes that detach during resize or reboot, verification holds on fresh accounts, and billing holds that quietly power things off if a card expires over a weekend.

Type
Cloud VPS, Bare Metal, Kubernetes and GPU infrastructure
First released
2014
License
Proprietary service, open-source API clients
Maintained by
Vultr Holdings Corporation (The Constant Company)
Typically used for
Budget production hosting, staging and QA, globally distributed VPS, GPU inference and training, S3-compatible object storage

The honest version

Vultr pros and cons

No technology is all upside. This is where Vultr earns its place and where it tends to cause trouble.

Where Vultr is strong

  • Honest, predictable pricingHourly billing with a monthly cap, clear bandwidth allowances and no surprise service charges. The invoice tells you what was used in terms you can explain.
  • A very wide region footprintMore than thirty regions across North America, South America, Europe, Africa, Asia and Oceania means latency targets for most customer bases are within reach without a second provider.
  • GPU instances at reasonable ratesNVIDIA A16, L40S and H100 instances come with per-hour pricing that suits inference workloads and smaller training runs without the paperwork of a hyperscaler contract.
  • Clean API and solid Terraform providerThe Vultr Terraform provider covers most of the product line, so you can describe an estate in HCL rather than clicking through the console at 2am.
  • S3-compatible Object StorageVultr Object Storage speaks the S3 API, which makes it a drop-in backup target or static asset host without rewriting the client code.
  • DDoS Protection is availableEdge-level scrubbing is an option on supported plans, which gives a reasonable first line for the kind of application that occasionally attracts attention.

Where Vultr bites

  • Managed services are thinner than the hyperscalersManaged Databases and Kubernetes exist but do not cover the breadth of RDS, GKE or AKS. Teams that depend on deeply managed products still end up building pieces themselves.
  • Region capacity varies by planPopular plans in popular regions are usually fine; less common plans or GPU instances in specific regions regularly return capacity errors that cost time to route around.
  • Outbound SMTP port 25 is blocked on fresh accountsThis is a reasonable anti-abuse measure and routinely surprises teams who push a production deploy and find transactional mail silently stuck.
  • Billing holds move fastA card that fails will pause instances within a short window, so an expired card during a holiday weekend can translate into actual downtime if nobody notices the warning email.
  • Block Storage has quirks under resize and rebootVolumes occasionally detach on reboot and need a manual reattach, and resize operations are sometimes rejected even when the dashboard says they should be accepted.
  • Support is written-only unless you upgradeThe free support tier is competent and ticket-based, and a complex incident at 3am can take longer to resolve than a hyperscaler's enterprise plan would allow. Business-hours responsiveness varies by region.

Sound familiar?

Common Vultr problems we fix

The Vultr problems that bring people to us, and what you have once each one is fixed.

API error: Plan is not available in the selected region

Terraform runs break on region capacity

A pipeline targeting a specific plan in a region that no longer has it, or a Reserved Instance count that was recently raised. The run fails and partial resources need cleanup.

Plan and region choices aligned with real capacity

ssh: connect to host port 22: timed out

An instance has gone silent

Ping lost, console shows the OS running, but no network reaches the box. Usually a firewall rule applied in the wrong order, a kernel panic after unattended upgrades, or a filled disk.

Instance reachable, cause understood

Block Storage: detached unexpectedly

A Block Storage volume dropped off after reboot

A volume that was mounted via /etc/fstab without nofail, so the instance panics at boot when the volume takes a moment to reattach, or a resize that completed but did not resync the attachment.

Volumes attach cleanly and survive reboots

outbound SMTP port 25 blocked

Transactional mail stops leaving the server

A code path that sent mail fine from the developer laptop and silently stacks up on the Vultr instance, because port 25 is closed by default on fresh accounts until support lifts the block.

Mail delivery routed through a relay

billing hold: payment method declined

Billing has quietly paused the account

An expired card over a holiday weekend, a bank that blocked a foreign charge, or a billing address mismatch. Instances are flagged for power-off before anyone opens the email.

Payment restored, account healthy

Kubernetes Engine: node pool stuck in provisioning

VKE cluster operations do not complete

A node pool upgrade that stalled, a region that no longer has the plan size the pool was built on, or a quota cap that was raised in the dashboard but not yet applied.

VKE operations finish cleanly

snapshot limit reached for account

Snapshots can no longer be created

An account that collected a snapshot per nightly backup for every instance without cleanup. Future backups fail silently from the perspective of the calling script.

Snapshot rotation on a plan

Terraform: 429 Too Many Requests

The provider is being rate-limited

A large plan or apply that calls the Vultr API hundreds of times in a short window, hitting account rate limits and failing mid-run.

Terraform parallelism and backoff tuned

What we do

Vultr help, from rescue to day-to-day

Call us for one dark instance, or let us run the whole Vultr estate.

Rescue
Instances that stopped responding, Block Storage that will not reattach, or regions that keep rejecting deploys. We get the service back up, then write up how it got there.
Build-out
A new Vultr estate laid out in Terraform with firewalls, VPCs, Block Storage, Object Storage and Load Balancers, so nothing is clicked into existence without record.
Migrations onto and off Vultr
Moving workloads from AWS, GCP, DigitalOcean or Linode onto Vultr and back the other way, with DNS cutover windows planned rather than hoped for.
VKE and container platforms
Vultr Kubernetes Engine clusters sized, upgraded and wired to Object Storage and Managed Databases with the add-ons you actually need.
Account recovery
Taking over Vultr accounts stuck under a former contractor, rotating API keys, moving billing and inventorying what is actually running.
Managed Tech
Patching, snapshot rotation, cost reviews and someone to page when a region is quietly dropping packets at an odd hour.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Vultr alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Vultr questions

Something else? Call+1 737 727 7327

Is Vultr reliable enough for production?

For most workloads yes, and for a long list of SaaS companies it already is. The network is competent, the regions are stable, and the control plane rarely surprises people. Where it is weaker than a hyperscaler is in the depth of managed services and in the response time for very complex support cases at odd hours, which is a trade some teams make deliberately for the price.

Why can we not launch a specific plan in a specific region on Vultr?

The honest answer is capacity. Popular plans in popular regions almost always work; less common plans, large memory tiers, and GPU instances in specific regions regularly return a Plan is not available response. The Terraform provider surfaces that cleanly, and the usual fix is either waiting, picking a neighboring region, or having Vultr move an allocation through a support ticket.

Why has Vultr stopped sending our transactional mail?

Outbound SMTP on port 25 is blocked on new accounts to protect the shared network reputation. You either open a ticket asking for the block to be lifted after showing the use case, or you send transactional mail through a relay such as Postmark, SendGrid or Mailgun from the host instead of directly. The second option is usually the better one anyway.

Can you move us from DigitalOcean or Linode to Vultr?

Yes, and the request usually comes up because the bill or a region footprint drove it. The path we follow is a Terraform build of the Vultr side, a database replica streaming across in advance, a DNS cutover during a quiet window, and the old provider kept running in standby until the new estate has proven itself for a week or two.

Can you take over a Vultr account from a previous developer?

Yes. We work through the account transfer, rotate every API key and SSH key, move billing onto your card, and write down what is actually running across instances, Block Storage, Object Storage and VKE so nothing goes dark during the handover.

How much does Vultr help cost?

Estimates come free. Tell us which regions you are running in and what has gone wrong, and we return a fixed price before touching anything.

Free estimate

Vultr giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.