Bluebuild running out of memory

General
1

How much space should i allocate to /tmp to avoid getting out of memory when building a custom image? https://github.com/thoady/aurora-nvidia-nordvpn/tree/main

Unless it s better to build on disk, currently on RAM (I have 32GB ram)

tmpfs 16G 28M 16G 1% /tmp

Thanks

oci-archive:/tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz => [5/5] STEP 32/32: LABEL org.opencontainers.image.title="aurora-nvidia-nordvpn"
oci-archive:/tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz => [5/5] COMMIT oci-archive:/tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz
oci-archive:/tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz => Error: committing container for step {Env:[TARGETARCH=amd64 IMAGE_REGISTRY=localhost BB_BUILD_FEATURES= CONFIG_DIRECTORY=/tmp/files IMAGE_NAME=aurora-nvidia-nordvpn RECIPE=recipes/recipe.yml MODULE_DIRECTORY=/tmp/modules BASE_IMAGE=ghcr.io/ublue-os/aurora-dx-nvidia-open FORCE_COLOR=1 CLICOLOR_FORCE=1 RUST_LOG_STYLE=always PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin] Command:label Args:[org.opencontainers.image.title aurora-nvidia-nordvpn] Flags:[] Attrs:map[] Message:LABEL org.opencontainers.image.title "aurora-nvidia-nordvpn" Heredocs:[] Original:LABEL org.opencontainers.image.title="aurora-nvidia-nordvpn"}: copying layers and metadata for container "73cc958ea6cab9647f4f7c979e435f71ac227544c0ffa6b3a250e4a4cd909f5a": committing the finished image: write /tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz: no space left on device
ERROR => Failed:
  × Failed to build oci-archive:/tmp/.tmpXNe0PP/aurora-nvidia-nordvpn.gz
2

Bluebuild is a separate project. Ask them for support.

3

I fixed this on mine. I’ll try to dig up what I did and post. It’s pretty easy.

4

Add this to your build.yml file under .github/workflows (if not already there) It happens because some of the Ublue images are big. Do you have this in your build.yml? You could also free some space with an additional step in your build.yml before it writes to /tmp or use a /tmp space that is larger but I’ve never had to do any of that.

          maximize_build_space: true

Also, as a side note, Dependabot PRs that try to bump the blue-build/github-action version will often fail because those runs do not have access to the SIGNING_SECRET cosign key, which the action requires via the cosign_private_key input. Because of that, you generally need to update the BlueBuild action version in build.yml manually instead of relying on Dependabot to do it. Example below, you just change the version after the @ sign. You COULD I suppose give the action access to your cosign key, but that’s not something I’d do. :wink:

 uses: blue-build/github-action@v1.10.2
5

Thanks for your reply,

the maximize_build_space is already set to true and expended /tmp size with

sudo mount -o remount,size=28G /tmp

and it worked, bluebuild happened to use slightly more than 16GB

6

No prob! Interesting that you had to do that. You must be adding a bit more than I am. I didn’t have to increase tmp.

You should post a bug/enhancement to the bluebuild git to have them increase the defaults.

7

This was also an issue in using ublue-os image template, the build would fail due to running out of memory in Github. Someone already sent an issue regarding this and the solution there was to add to maximize_build_space = true in the build.yml file with:

with:
extra-squeeze: true

since Bluebuild pulls from ublue-os images and has the same error, it seems to be the same issue.

8

Thanks! I didn’t know about that option.

9

Tried that, I don’t think that’s a valid option:

Warning: Unexpected input(s) 'extra-squeeze', valid inputs are ['recipe', 'cosign_private_key', 'push', 'registry_token', 'registry_username', 'pr_event_number', 'maximize_build_space', 'use_unstable_cli', 'cli_version', 'registry', 'registry_namespace', 'build_chunked_oci', 'max_layers', 'rechunk', 'use_cache', 'retry_push_count', 'squash', 'build_opts', 'working_directory', 'skip_checkout', 'verify_install']```
10

that’s weird. This is the github issue with the suggestion that I followed: Custom Bazzite build fails due to insufficient space left on device while pulling base image · Issue #172 · ublue-os/image-template · GitHub

I did this using a completely new fork yesterday that would not build and doing this made it work.

11

Maybe I put it in the wrong section or something. edit: looks that way

What version of blue build are you using?

12

me, using

BlueBuild 0.9.27
branch:
commit_hash:
build_time:2025-12-05 10:07:42 +01:00
build_env:rustc 1.91.1 (ed61e7d7e 2025-11-07) (Fedora 1.91.1-1.fc43),

to be sure had further container problems because i was generating the iso locally, have been suggested to generate from a published image

13

I’m using the uBlue image-template here, not Bluebuild specifically. I had to add the extra squeeze to the build.yml file in the workflow. Would be interesting to know where to put it for Bluebuild since that is all yaml.

14

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.