Hello,

I’m facing a persistent Docker issue on Unraid that I haven’t been able to resolve, even after extensive troubleshooting. I’d appreciate any insights or similar experiences.

---

Issue description

After a cold boot / power-on:

• The first start of the array always causes Docker to fail

• Error shown:

  Docker Service failed to start
  failed to start containerd: timeout waiting for containerd to start
  

• If I then stop the array and start it again, Docker starts normally

• Any subsequent Docker restart works fine

• The problem only occurs on the first array start after power-on

Important detail:

• This happens even with array auto-start disabled

• Boot Unraid → manually click Start Array

• On that first start, Docker fails

• Stop Array → Start Array → Docker works every time

---

What I’ve already tested

To rule out configuration and software issues, I’ve tested:

• Fresh Unraid installation on a new USB flash drive

• No plugins, no custom scripts

• Docker using:

  • docker.img

  • Docker directory mode

• Docker data located on:

  • NVMe

  • SATA SSD

• No containers set to autostart

• No containers at all

• Various delays (sleep, rootdelay)

• Manual Docker restart

• User Scripts to automatically restart Docker

• Docker and containers work perfectly after the first array restart

None of the above changed the behavior.

---

Observations

Based on all tests:

• This does not seem related to:

  • Docker configuration

  • Containers

  • docker.img corruption

  • Filesystem choice

  • Plugins or scripts

• It appears that the first array start after power-on leaves the system in a bad state

• A single array stop/start reliably fixes the issue

• After that, the system is completely stable

---

Additional context

• This setup used to work normally

• The issue appeared “out of nowhere”

• No hardware changes were made

• Once the array has been restarted once, everything works as expected

---

Question

Has anyone encountered a situation where:

• Docker fails only on the first array start after power-on

• containerd times out

• and a simple array stop/start fixes it every time?

Could this be related to:

• storage controller initialization timing?

• NVMe / SATA firmware behavior?

• kernel block device readiness?

• something that only fully initializes after the first array stop/start?

Any ideas on logs to check or known issues would be very helpful.

Thanks in advance.

unraid-diagnostics-20260109-1010.zip

Don't remember ever seeing that before.

System share exists on both pools; clean that up, and make sure it only exists on the pool you want, which I assume is "cache", then enable exclusive shares, or change the docker config to use the disk path (/mnt/cache) intead of /mnt/user and see if that makes any difference.

Same problem.

unraid-diagnostics-20260109-1447.zip

When you tested with a new install, was it also with all new data or the existing pools?

Existing pools.
But I tried to delete all docker folder and the problem continue.

I would recommend trying with a new install and new storage/pool, you can use an extra flash drive as a pool if you like to avoid reformatting the existing pools.