Fully managed infrastructure

From manual management to 220 nodes that heal themselves.

Downtime hits revenue immediately. We built a system that fixes outages itself, often before anyone notices.

ClientMyria Nodes Sectorblockchain infrastructure ServicesVPS management, monitoring, automation Read time3 min

Overview of 220 virtual servers with real-time status

Replaces manual management

Stack

Linux Python PHP SSH

Duration 3 years fully managed

The challenge

Keeping hundreds of nodes online at once turns every outage into a race against the clock. With blockchain nodes, every minute of downtime means missed earnings: a node's uptime directly determines what that node earns.

mockup

infra · nodes

Nodes220 onlineavg. uptime 99.98%1

node-01

99.99%uptime

node-02

100.0%uptime

node-03

99.98%uptime

node-04

99.97%uptime

node-05

99.99%uptime

node-06

100.0%uptime

node-07

99.41%uptime

node-08

99.99%uptime

Screen 1 Node overview. 220 nodes at a glance, with real-time status and uptime. 1 uptime directly determines earnings.

What we built

A fully automated management system on the Linux servers. If a node fails, it first tries an automatic restart. If that fails, the system fully rebuilds the node, without manual intervention. Only if automated recovery fails does an alert go to the admins.

mockup

auto-heal · node-147

wifi_off

Node offline detectednode-147 · no response

14:02

restart_alt

Automatic restart startedattempt 1

14:02 1

error

Restart failednode stays offline

14:03

build

Full rebuild startednode rebuilt from scratch

14:03 2

check_circle

Node online, uptime restoredno alert needed

14:07 3

Screen 2 Outage and auto-heal. 1 first an automatic restart, 2 if that fails a full rebuild, 3 an alert only reaches the admin if both fail.

What the system does

220 nodes continuously monitored with a real-time uptime overview

Automatic restart on failure, no manual intervention needed

If restart fails, the system fully rebuilds the node

Secure access via SSH keys, no passwords

Alert admins only when automated recovery fails

In practice

The result

We kept 220 nodes fully automated online, fully managed for three years. Outages were resolved on their own, often before anyone noticed. And because each node's uptime directly determined earnings, every second we kept online protected revenue.

← Previous case From Zendesk to custom Next case → GemCards