Automatic vs. Manual Partitioning in Linux: Which Should You Choose?

A couple months back, I was spinning up a new EKS cluster for a mid-sized client here in the Netherlands. I hit a snag with my RHEL 10 test VM. Automatic partitioning had eaten the whole disk for root and swap. No room left when Docker 24.0 containers started ballooning during Jenkins builds.

I wasn’t sure if manual partitioning would save me. Or if it’d just brick the thing, like that time years ago when I accidentally rm’d the kernel on a production RHEL box. I rescued it with a live USB, but barely. This kind of disaster actually relates to something I covered in 5 Common Linux Installation Mistakes and How to Avoid Them—partitioning mishaps are definitely up there.

With 14 years as a DevOps engineer bouncing between startups and enterprises—RHCE, CKA, AWS Solutions Architect under my belt—I figured it was time to really compare the two.

Automatic partitioning is dead simple, especially in RHEL/Anaconda installers. You just click “Done” on the storage screen. It spits out /boot (usually 1GB ext4), a root / (most of the disk), and swap (RAM-sized).

Great for quick dev setups. Or when you’re not sweating disk quotas—like my EKS nodes where pods manage their own mess. That “Encrypt my data” checkbox down there full-disk encrypts everything if you set a passphrase. Bootloader goes on the main disk automatically.

Last month, I let it auto-partition a Jenkins agent VM. It just worked. No fuss, full disk for / and swap.

Manual partitioning? You own it completely. Select your /dev/sda (or whatever the VM disk shows as) and hit “Manual partitioning.” Then craft away: 512MB /boot ext4, 15-20GB root, 4GB swap (twice my 2GB test RAM), rest to /home or LVM for flexibility.

I did this for a SRE gig at a large enterprise. Separated /var/lib/docker so container sprawl wouldn’t trash root. Honest mistake early on: I equated Flux CD’s GitOps to Argo CD—turns out partitioning nuance is similar. Defaults bite you later.

Looking back, I probably should’ve sized /boot bigger initially. If you’re looking to practice these concepts without spending money, I’ve written about How to Build a Zero-Cost RHEL 10 Practice Lab Using VirtualBox separately—perfect for testing different partitioning schemes safely.

Dual-boots need manual, since bootloaders fight otherwise. Or encryption per-partition if you’re paranoid.[4] Single partition simplifies—no resize drama if /home explodes. LVM on manual is underrated for resizing without downtime, though. Tangent: I know this is slightly off-topic. On btrfs in newer setups, swap files complicate things; stick to partitions there.

Still not 100% sure on zram vs traditional swap for EKS—anyone?

Auto for speed, manual for control. Depends on your stack. What’s your go-to?

💡 Take the next step

If this resonated, RHCSA Bootcamp (RHEL 10)—Arabic is where I teach this systematically—no fluff, just the skills that matter in the real world.

👉 Enroll today and start learning →