[bug]: Failed to get pins for version v6.10.0: Failed to fetch pins

[roland3564]

Is there an existing issue for this problem?

• I have searched the existing issues

Install method

Invoke's Launcher

Operating system

Linux

GPU vendor

Nvidia (CUDA)

GPU model

RTX 4060

GPU VRAM

8GB

Version number

v6.10.0

Browser

No response

System Information

Operating System: KDE neon User Edition
KDE Plasma Version: 6.5.4
KDE Frameworks Version: 6.21.0
Qt Version: 6.10.1
Kernel Version: 6.14.0-37-generic (64-bit)
Graphics Platform: X11
Processors: 20 × 13th Gen Intel® Core™ i7-13700H
Memory: 16 GiB of RAM (15.3 GiB usable)
Graphics Processor 1: NVIDIA GeForce RTX 4060 Laptop GPU
Graphics Processor 2: Intel® Iris® Xe Graphics
Manufacturer: HP
Product Name: Victus by HP Gaming Laptop 15-fa1xxx

What happened

I tried to install Invoke with the Launcher, went through the folder location, version, driver and it gave me "Failed to fetch pins" when it started.

What you expected to happen

I expect the installation to succeed

How to reproduce the problem

1. Grab the appimage from the website
2. Go through the installation

Additional context

I have to disable appimage's sandbox with "--no-sandbox" on each run because it does not launch otherwise.

Discord username

No response

[roland3564]

Ok, I disabled the VPN and now it reached for the pins, I have another problem now:

Installation parameters:
- Invoke version: v6.10.0
- Install location: /home/n1/Documenti/App/Invoke
- Python version: 3.12
- GPU type: nvidia>=30xx
- Torch platform: cuda
- Using torch index: https://download.pytorch.org/whl/cu128
Installing Python 3.12...
> /tmp/.mount_InvokeJC0jIt/resources/bin/uv python install 3.12 --python-preference only-managed --reinstall
Installed Python 3.12.9 in 9.66s
 ~ cpython-3.12.9-linux-x86_64-gnu
Deleting existing virtual environment...
Creating virtual environment...
> /tmp/.mount_InvokeJC0jIt/resources/bin/uv venv --relocatable --prompt invoke --python 3.12 --python-preference only-managed /home/n1/Documenti/App/Invoke/.venv
Using CPython 3.12.9
Creating virtual environment at: /home/n1/Documenti/App/Invoke/.venv
Activate with: source /home/n1/Documenti/App/Invoke/.venv/bin/activate
Installing invokeai package...
> /tmp/.mount_InvokeJC0jIt/resources/bin/uv pip install --python 3.12 --python-preference only-managed invokeai==v6.10.0 --force-reinstall --compile-bytecode --index=https://download.pytorch.org/whl/cu128
Resolved 118 packages in 9.30s
  × Failed to download `nvidia-cuda-runtime-cu12==12.8.57`
  ├─▶ Failed to extract archive:
  │   nvidia_cuda_runtime_cu12-12.8.57-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
  ╰─▶ Failed to download distribution due to network timeout. Try increasing UV_HTTP_TIMEOUT (current
      value: 30s).
  help: `nvidia-cuda-runtime-cu12` (v12.8.57) was included because `invokeai` (v6.10.0) depends on
        `torch` (v2.7.1+cu128) which depends on `nvidia-cuda-runtime-cu12`
Failed to install invokeai python package: Process exited with code 1
Try installing again with Repair mode enabled to fix this.
Ask for help on Discord or GitHub if you continue to have issues.

[BieHDC]

I had the same error, installing the update in repair mode fixed it.

[roland3564]

I had the same error, installing the update in repair mode fixed it.

I have Linux so I don't know how to do it there.

[BieHDC]

I had the same error, installing the update in repair mode fixed it.

I have Linux so I don't know how to do it there.

If you are using the launcher there is a "repair mode" checkbox at the bottom when you go through the install update process. For non-launchers i dont know how it works, probably wiping the venv manually so it sets it up again.

[roland3564]

Unfortunately it's still giving me issues after selecting repair mode.

  × Failed to download `nvidia-cusparselt-cu12==0.6.3`
  ├─▶ Failed to write to the distribution cache
  ├─▶ error decoding response body
  ├─▶ request or response body error
  ├─▶ error reading a body from connection
  ╰─▶ peer closed connection without sending TLS close_notify:
      https://docs.rs/rustls/latest/rustls/manual/_03_howto/index.html#unexpected-eof
  help: `nvidia-cusparselt-cu12` (v0.6.3) was included because `invokeai` (v6.11.0rc1) depends on
        `torch` (v2.7.1+cu128) which depends on `nvidia-cusparselt-cu12`
Failed to install invokeai python package: Process exited with code 1
Try installing again with Repair mode enabled to fix this.
Ask for help on Discord or GitHub if you continue to have issues.

[lstein]

I think this sort of thing can happen when an install is interrupted by a network glitch.

If the repair checkbox isn't working. Then you can try removing or renaming the .venv directory inside the InvokeAI directory, and relaunching the launcher. It will tell you that the installation is missing and offer to reinstall. You can go ahead with this -- your database, images and other settings will be preserved.

If you are still having the same error, then you could try clearing out your package cache. To do this, click on the ">" icon in the bottom left corner of the launcher, and then click "Start Dev Console". When the console appears, type uv cache clean. Close the console window and try to reinstall.

Try this and let me know how it works