Revert: Keep Ubuntu version at 26.04 for Vulkan Docker build

[Copilot]

Make sure to read the contributing guidelines before submitting a PR

Description

This PR reverts the Ubuntu version change in .devops/vulkan.Dockerfile back to 26.04 as requested by the maintainer.

The original change attempted to fix the Vulkan Docker build failure by changing Ubuntu from 26.04 to 24.04, but per maintainer feedback, the fix should be implemented differently while staying on Ubuntu 26.04.

Related to the GitHub Actions docker workflow failure introduced in commit bc809e9.

• Fixes Misc. bug: Vulkan docker build fails on github actions ggml-org/llama.cpp#17522

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Misc. bug: Vulkan docker build fails on github actions</issue_title>
<issue_description>### Name and Version

Version 7134

Operating systems

Linux

Which llama.cpp modules do you know to be affected?

Other (Please specify in the next section)

Command line

Problem description & steps to reproduce

Hello again! Last week I opened an issue concerning an issue with the vulkan docker build. The issue was resolved with this commit, thank you btw :). The fix was to update the ubuntu version with which the dockerfile was built. Since that commit, docker builds of the vulkan image have been failing on github actions. A quick glance at the error logs from the first unsuccessful build shows that numpy fails to build, most likely due to a missing C compiler. Given that ubuntu 26.04 has not been released yet, there may be a conflict with the build essentials used in the dockerfile.

Paging @ericcurtin as it seems relevant.

First Bad Commit

The first bad commit is the one that introduced the fix to the issue I opened previously, commit bc809e9.

Relevant log output

ggml-org/llama.cpp#20 13.48   Downloading numpy-1.26.4.tar.gz (15.8 MB)
ggml-org/llama.cpp#20 13.61      ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 15.8/15.8 MB 135.0 MB/s eta 0:00:00
ggml-org/llama.cpp#20 15.54   Installing build dependencies: started
ggml-org/llama.cpp#20 17.26   Installing build dependencies: finished with status 'done'
ggml-org/llama.cpp#20 17.26   Getting requirements to build wheel: started
ggml-org/llama.cpp#20 17.39   Getting requirements to build wheel: finished with status 'done'
ggml-org/llama.cpp#20 17.39   Installing backend dependencies: started
ggml-org/llama.cpp#20 18.38   Installing backend dependencies: finished with status 'done'
ggml-org/llama.cpp#20 18.38   Preparing metadata (pyproject.toml): started
ggml-org/llama.cpp#20 19.21   Preparing metadata (pyproject.toml): finished with status 'error'
ggml-org/llama.cpp#20 19.22   error: subprocess-exited-with-error
ggml-org/llama.cpp#20 19.22   
ggml-org/llama.cpp#20 19.22   × Preparing metadata (pyproject.toml) did not run successfully.
ggml-org/llama.cpp#20 19.22   │ exit code: 1
ggml-org/llama.cpp#20 19.22   ╰─> [20 lines of output]
ggml-org/llama.cpp#20 19.22       + /usr/bin/python3 /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0/vendored-meson/meson/meson.py setup /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0 /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0/.mesonpy-19k9xnsx -Dbuildtype=release -Db_ndebug=if-release -Db_vscrt=md --native-file=/tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0/.mesonpy-19k9xnsx/meson-python-native-file.ini
ggml-org/llama.cpp#20 19.22       The Meson build system
ggml-org/llama.cpp#20 19.22       Version: 1.2.99
ggml-org/llama.cpp#20 19.22       Source dir: /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0
ggml-org/llama.cpp#20 19.22       Build dir: /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0/.mesonpy-19k9xnsx
ggml-org/llama.cpp#20 19.22       Build type: native build
ggml-org/llama.cpp#20 19.22       Project name: NumPy
ggml-org/llama.cpp#20 19.22       Project version: 1.26.4
ggml-org/llama.cpp#20 19.22       
ggml-org/llama.cpp#20 19.22       ../meson.build:1:0: ERROR: Unknown compiler(s): [['cc'], ['gcc'], ['clang'], ['nvc'], ['pgcc'], ['icc'], ['icx']]
ggml-org/llama.cpp#20 19.22       The following exception(s) were encountered:
ggml-org/llama.cpp#20 19.22       Running `cc --version` gave "[Errno 2] No such file or directory: 'cc'"
ggml-org/llama.cpp#20 19.22       Running `gcc --version` gave "[Errno 2] No such file or directory: 'gcc'"
ggml-org/llama.cpp#20 19.22       Running `clang --version` gave "[Errno 2] No such file or directory: 'clang'"
ggml-org/llama.cpp#20 19.22       Running `nvc --version` gave "[Errno 2] No such file or directory: 'nvc'"
ggml-org/llama.cpp#20 19.22       Running `pgcc --version` gave "[Errno 2] No such file or directory: 'pgcc'"
ggml-org/llama.cpp#20 19.22       Running `icc --version` gave "[Errno 2] No such file or directory: 'icc'"
ggml-org/llama.cpp#20 19.22       Running `icx --version` gave "[Errno 2] No such file or directory: 'icx'"
ggml-org/llama.cpp#20 19.22       
ggml-org/llama.cpp#20 19.22       A full log can be found at /tmp/pip-install-p4xgmo1y/numpy_ba63b7775fd142ee9f0f697015a2f2d0/.mesonpy-19k9xnsx/meson-logs/meson-log.txt
ggml-org/llama.cpp#20 19.22       [end of output]
ggml-org/llama.cpp#20 19.22   
ggml-org/llama.cpp#20 19.22   note: This error originates from a subprocess, and is lik...

</details>

- Fixes ggml-org/llama.cpp#17522

<!-- START COPILOT CODING AGENT TIPS -->
---

✨ Let Copilot coding agent [set things up for you](https://github.com/ericcurtin/llama.cpp/issues/new?title=✨+Set+up+Copilot+instructions&body=Configure%20instructions%20for%20this%20repository%20as%20documented%20in%20%5BBest%20practices%20for%20Copilot%20coding%20agent%20in%20your%20repository%5D%28https://gh.io/copilot-coding-agent-tips%29%2E%0A%0A%3COnboard%20this%20repo%3E&assignees=copilot) — coding agent works faster and does higher quality work when set up for your repo.

[ericcurtin]

@copilot Revert this change, we should fix things, but stay on 26.04

[Copilot]

@copilot Revert this change, we should fix things, but stay on 26.04

Reverted the change in commit 0c810a1. The Dockerfile now uses Ubuntu 26.04 again.