DigitalOcean home
  • Droplets
  • Spaces
  • Kubernetes
  • Tools & Integrations
  • One-click Apps
  • API Documentation
  • Community
  • Tutorials
  • Q&A
  • Projects
  • Meetups
  • Customers
  • Pricing
  • Docs
  • Support
  • DigitalOcean home
  • Products
    • Droplets

      Scalable compute services.

    • Spaces

      Simple object storage.

    • Kubernetes

      Run managed Kubernetes clusters.

    • Tools & Integrations

      Automate your infrastructure.

    • One-click Apps

      Deploy pre-built applications.

    • API Documentation
  • Customers
  • Community
    • Community Overview

      Connect, share and learn

    • Tutorials

      DevOps and development guides

    • Questions & Answers

      Development and systems Q&A

    • Projects

      Community-built integrations

    Get Involved
    Write for DOnations
    Join us at a Meetup
    Featured Post
    An Introduction to Kubernetes

    by Justin Ellingwood

  • Pricing
  • Docs
  • Support
    • Documentation

    • Contact Support

    • Network Status

  • Home /
  • DO-I-336 /
  • New idea
2 Vote

Memory Used graph way too high

Last week I created a new droplet with the default Ubuntu 16.04.3 x64 installation. Since then, the Memory Used graph shown on the web site has been showing almost twice the amount of used memory compared to what I see on the server in ssh when running "free -h" or "htop".

Right now the graph says 59% used, while running "free -h" shows 144M used of 488M, i.e. 30%. Running "htop" shows 170M/488M used, i.e. 35%, which is still far from 59%.

I only have nginx and two Nodejs instances installed and running, so the server is pretty barebone. I don't expect perfect consistency, but would expect something in the range of 30-40%, not 60%.

It seems like the graph on your site might be classifying the memory buffer cache as used memory. To me, this is incorrect, as cached memory is freed up automatically when used by applications, so it's not a sign that you're nearing your memory limit, which the graph currently suggests.

Because memory management is the most important part of a low-memory server I would expect the graph to report actual memory used by my apps and ignore cache, to give me a much more useful metric.

Here's the complete output of "free -h":

total used free shared buff/cache available
Mem: 488M 144M 15M 5.6M 328M 304M
Swap: 0B 0B 0B

  • Hannes
  • Sep 11 2018
  • Already exists
Insights
  • Comments (1)
  • Votes (2)
  • Attach files
  • Michelle Peot commented
    19 Nov, 2019 11:14pm

    Cache is currently not included in used.

    ×

    Attachments Open full size

Log in / Sign up

Identify yourself with your email address

Subscribe

You won't be notified about changes to this idea.

Related ideas

DigitalOcean home

© 2018 DigitalOcean, LLC. All rights reserved.
Proudly made in NY

  • Twitter
  • Facebook
  • Instagram
  • YouTube
  • LinkedIn
  • Glassdoor
Company
About
Leadership
Blog
Careers
Partner Network
Referral Program
Events
Press
Legal & Security
Products
Droplets
Spaces
Kubernetes
Tools & Integrations
One-click Apps
API
Pricing
Documentation
Release Notes
Community
Tutorials
Meetups
Q&A
Write for DOnations
Droplets for Demos
Hatch
Shop Swag
Research Program
Currents Research
Open Source
Support
Contact Support
FAQ
Network Status