Your message dated Sat, 25 Oct 2025 15:47:24 -0400
with message-id <[email protected]>
and subject line Re: Bug#934686: dockerd[3598]: Error starting daemon: error 
initializing graphdriver: driver not supported
has caused the Debian Bug report #934686,
regarding dockerd[3598]: Error starting daemon: error initializing graphdriver: 
driver not supported
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)


-- 
934686: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=934686
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: docker.io
Version: 18.09.5+dfsg1-1
Severity: normal

Dear Maintainer,

The unit docker.socket has entered the 'failed' state with result 
'service-start-limit-hit'.

It works with kernel 4, but fails with 5

*** Reporter, please consider answering these questions, where appropriate ***

   * What led up to the situation?
   * What exactly did you do (or not do) that was effective (or
     ineffective)?
   * What was the outcome of this action?
   * What outcome did you expect instead?

*** End of the template - remove these template lines ***


-- System Information:
Debian Release: bullseye/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (500, 'testing'), (500, 'stable'), (500, 
'oldstable'), (1, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 5.2.0-2-amd64 (SMP w/4 CPU cores)
Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE, 
TAINT_UNSIGNED_MODULE
Locale: LANG=en_US.utf8, LC_CTYPE=ru_RU.UTF-8 (charmap=UTF-8), 
LANGUAGE=en_US.utf8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled

Versions of packages docker.io depends on:
ii  adduser             3.118
ii  iptables            1.8.3-2
ii  libc6               2.28-10
ii  libdevmapper1.02.1  2:1.02.155-3
ii  libltdl7            2.4.6-10
ii  libnspr4            2:4.21-1
ii  libnss3             2:3.45-1
ii  libseccomp2         2.4.1-2
ii  libsystemd0         241-7
ii  lsb-base            10.2019051400
ii  runc                1.0.0~rc6+dfsg1-3
ii  tini                0.18.0-1+b1

Versions of packages docker.io recommends:
ii  ca-certificates  20190110
ii  cgroupfs-mount   1.4
ii  git              1:2.23.0~rc1-1
ii  needrestart      3.4-5
ii  xz-utils         5.2.4-1

Versions of packages docker.io suggests:
ii  aufs-tools           1:4.14+20190211-1
ii  btrfs-progs          5.2.1-1
ii  debootstrap          1.0.115
pn  docker-doc           <none>
ii  e2fsprogs            1.45.3-4
pn  rinse                <none>
ii  xfsprogs             5.0.0-1
pn  zfs-fuse | zfsutils  <none>

-- no debconf information

--- End Message ---
--- Begin Message ---
Vladmimir Stavrinov <[email protected]> writes:


> The unit docker.socket has entered the 'failed' state with result 
> 'service-start-limit-hit'.
>
> It works with kernel 4, but fails with 5
>


Hi Vladimir,

First -- thank you for the clear, detailed report and for including the
full system information. That made it very easy to see what’s
happening. I completely understand the frustration: Docker has been
working perfectly on your 4.x kernels, and suddenly with 5.2 it refuses
to start. That kind of regression feels like a step backward, even when
it’s not really a bug in the package itself.  Here is what's going on:

Docker needs a storage driver (sometimes called a graphdriver) to manage
image layers and container filesystems. In older Debian
setups—especially on kernels in the 4.x series—the default driver was
often aufs, because it was stable, well-understood, and widely used in
early Docker deployments.

However, aufs was never merged into the upstream Linux kernel. It has
always lived as an out-of-tree module. Starting around kernel 5.0, and
especially by 5.2, most distributions (including Debian unstable/sid)
dropped built-in support for aufs because:

- It’s effectively unmaintained upstream.
- The official replacement, overlay2, is faster, more secure, and fully
  integrated into the mainline kernel.
- Carrying aufs as a DKMS module became increasingly fragile with each
  new kernel.

Your system is trying to start dockerd using aufs (likely the default
from your older config or inherited behavior), but kernel 5.2.0-2-amd64
simply doesn’t have aufs available—no module, no support. Hence:

  texterror initializing graphdriver: driver not supported

And because the daemon fails immediately, systemd keeps retrying until it hits 
the start limit:

  textdocker.socket: Failed with result 'service-start-limit-hit'

This isn’t a crash in Docker itself -- it’s Docker doing exactly what it
should: refusing to start with an unsupported backend.

Linux 5 fully supports overlay2, and it’s the recommended driver for
modern Debian systems. All you need is a one-line config change.

  $ sudo mkdir -p /etc/docker
  $ sudo tee /etc/docker/daemon.json > /dev/null <<EOF
  {
    "storage-driver": "overlay2"
  }
  EOF

Then restart the service:

  $ sudo systemctl restart docker

You should now see:

  $ docker info --format '{{.Driver}}'
  overlay2

And everything should work exactly as before—often faster and with
better snapshot performance.

Why This Wasn’t Caught Earlier
------------------------------

Docker doesn’t auto-migrate storage drivers (for good reason—it would
risk data corruption). So even though overlay2 has been the upstream
default for years, systems that were set up years ago on aufs keep
trying to use it unless explicitly told otherwise.

Ideally docker would provide an error such as:

   Storage driver 'aufs' is not supported on this kernel. Consider using 
'overlay2'.

Not sure if it makes sense these days though, since the issue is kindof old and 
fresh
installations are not affected by this.

All the best,
-rt

--- End Message ---

Reply via email to