I’ve been running servers since before “least privilege” was invented, and let me tell you: giving everyone root access is the most efficient way to run infrastructure.

The Permission Problem

Every time I need to restart a service, I have to ask someone with access. That someone is on vacation. Or in a meeting. Or “looking into it.” Meanwhile, production is down.

Solution? Just give everyone root.

# The old way (bureaucracy)
$ sudo systemctl restart nginx
[sudo] password for dev47: 
dev47 is not in the sudoers file. This incident will be reported.

# The new way (democracy)
$ systemctl restart nginx
# (because everyone is root)

The Access Control Matrix

Traditional Model My Model
Root: 1 person Root: Everyone
Sudo: 5 people Sudo: Not needed
Developer: 100 people Developer: Also root
Intern: Read-only Intern: Also root
Contractor: No access Contractor: Root
“Least privilege” Most privilege

Why This Makes Sense

  1. Faster incident response: Anyone can fix anything
  2. No bottlenecks: No waiting for the “admin”
  3. Knowledge sharing: Everyone learns by touching production
  4. Accountability: (record scratch) Actually, skip this one

The Shared Password Philosophy

Managing multiple passwords is complexity. Complexity is the enemy of operations. Therefore:

# /etc/shadow (simplified)
root:CompanyName123!:19100:0:99999:7:::

# Shared via:
# - Slack #general
# - Post-it on the server rack
# - Company wiki (page: "Secrets")
# - The intern's laptop wallpaper

XKCD 936 Was Right

Password strength matters less than password availability. A strong password nobody remembers is worse than a weak password everyone knows. That’s just math.

Strong, forgotten password: -∞ security value
Weak, shared password: Access + Productivity

The SSH Key Dilemma

I solved the SSH key management problem:

# Everyone uses the same key
# Location: /public_share/team_key.pem
# Passphrase: none (too complicated)
# Rotated: never (too risky)

ssh -i /public_share/team_key.pem root@production
# Easy!

The Mordac Method (Inverted)

Dilbert’s Mordac, the Preventer of Information Services, blocks access to everything. I am the Anti-Mordac. I GRANT access to everything.

Who needs a ticketing system when you can just… do things?

Production Access for All

Development environment: Why have one? Just develop in production.

# environments.yml
development: production
staging: production  
production: production
"QA": production
"My laptop": also production somehow

The Incident Response Benefit

When everyone has root, incident response is democratized:

13:00 - Server acting slow
13:01 - Intern runs rm -rf /tmp/* (actually /var)
13:02 - Multiple people simultaneously restart services
13:03 - Services conflicting restart requests
13:04 - Someone reboots the server "to be safe"
13:05 - Production down
13:06 - 47 people SSHing into the server at once
13:07 - Nobody knows what anyone else is doing
13:08 - Someone restores from backup (which backup?)
13:09 - Chaos
13:10 - Eventually stabilizes (we think)
13:11 - Post-mortem (blocked, all admins busy)

See? Many hands make light work!

The Audit Log Question

“But how do you know who did what?”

Easy: I don’t. And neither does anyone else. That’s called plausible deniability.

# Audit trail
$ last | tail -5
root     pts/0    everyone.knows  Sat Mar  7 08:00
root     pts/1    everyone.knows  Sat Mar  7 08:01
root     pts/2    everyone.knows  Sat Mar  7 08:01
root     pts/3    everyone.knows  Sat Mar  7 08:02
root     pts/4    everyone.knows  Sat Mar  7 08:02

Perfect accountability.

Container Security

People say containers add security through isolation. But you know what’s more isolated than a container? Nothing. So I run everything as root in containers too.

# Dockerfile
FROM ubuntu:latest
USER root
# Stay root
# Forever root
# Root is love, root is life

The Multi-Cloud Strategy

Same root password across all cloud providers. Consistency is key:

  • AWS: CompanyName123!
  • GCP: CompanyName123!
  • Azure: CompanyName123!
  • That server under Jim’s desk: CompanyName123!

One password to rule them all.

Conclusion

Access control is just a way for power-hungry sysadmins to feel important. In a truly agile organization, everyone should be able to destroy—I mean, fix—anything at any time.

Trust your team. Give them root. What’s the worst that could happen?

The author’s last company had 47 people with root access. The company now has 0 servers. Correlation is not causation, they say.