Baike.dev
Log in
Back to tool/Back to issues
#4156·mamba

BUG: mamba does not apply repodata patches for already-downloaded packages

Author: h-vetinariCreated Feb 4, 2026Updated Aug 4, 2026

Troubleshooting docs

  • My problem is not solved in the Troubleshooting docs

Anaconda default channels

  • I do NOT use the Anaconda default channels (pkgs/* etc.)

How did you install Mamba?

Mambaforge or latest Miniforge

Search tried in issue tracker

done

Latest version of Mamba

  • My problem is not solved with the latest version

Tried in Conda?

I do not have this problem with Conda, just with Mamba

Describe your issue

I updated my build environment as usual after some relevant new release using mamba update --all. The update in question was conda-smithy 3.54.1 -> conda-smithy 3.54.2, but as it turns out, py-rattler had a release more or less at the same time, which ended up breaking smithy: https://github.com/conda/rattler/issues/2023

So I downgraded to py-rattler <0.22 locally in order to be able to keep working, while also pushing various fixes

Now that conda{,-build} 26.1 came out, I did mamba update --all again, and ran into the same breakage (because I hadn't noticed that mamba had wrongly updated to py-rattler v0.22 again). Only this time, the repodata patch should have forbidden mamba updating py-rattler even for the _0 build of conda smithy 3.54.2 I had already installed.

This is certainly not ideal, but could perhaps be written off as transitory. However, if I now try mamba update --all again (after once more forcing a downgrade to py-rattler <0.22 and doing mamba update conda-smithy separately), mamba not only wants to install the broken combination again, but wants it to the point of downgrading smithy

>mamba update --all
conda-forge/noarch                                          Using cache
conda-forge/win-64                                          Using cache

Pinned packages:

  - python=3.12

Transaction

  Prefix: E:\miniforge\envs\builder


  No specs added or removed.

  Package         Version  Build             Channel           Size
---------------------------------------------------------------------
  Upgrade:
---------------------------------------------------------------------

  - py-rattler     0.21.0  py310hb39080a_0   conda-forge     Cached
  + py-rattler     0.22.0  py310hb39080a_0   conda-forge     Cached

  Downgrade:
---------------------------------------------------------------------

  - conda-smithy   3.54.2  win_pyhf1ceabd_1  conda-forge     Cached
  + conda-smithy   3.54.1  win_pyhf1ceabd_0  conda-forge     Cached

  Summary:

  Upgrade: 1 packages
  Downgrade: 1 packages

  Total download: 0 B

---------------------------------------------------------------------

In other words (AFAICT), the unpatched repodata of conda-smithy-3.54.1-win_pyhf1ceabd_0 has effectively poisoned my local package metadata. The behaviour persists even after conda clean --all and mamba clean --all.

Since the issue-template asked me to, I ran the same test for conda, i.e. conda update --all, and it considers the environment up to date.

mamba info / micromamba info

bash
libmamba version : 2.5.0
          mamba version : 2.5.0
           curl version : libcurl/8.18.0 Schannel zlib/1.3.1 libssh2/1.11.1 WinLDAP
     libarchive version : libarchive 3.8.5 zlib/1.3.1 liblzma/5.8.2 bz2lib/1.0.8 liblz4/1.10.0 libzstd/1.5.7 libxml2/2.d
       envs directories : E:\miniforge\envs
          package cache : E:\miniforge\pkgs
                          C:\Users\[...]\.mamba\pkgs
                          C:\Users\[...]\AppData\Roaming\.mamba\pkgs
            environment : builder (active)
           env location : E:\miniforge\envs\builder
      user config files : C:\Users\[...]\.mambarc
 populated config files : E:\miniforge\envs\builder\condarc.d\anaconda-auth.yml
                          C:\Users\[...]\.condarc
                          E:\miniforge\.condarc
       virtual packages : __win=10.0.26100=0
                          __archspec=1=x86_64_v3
               channels : https://conda.anaconda.org/conda-forge/noarch
                          https://conda.anaconda.org/conda-forge/win-64
       base environment : E:\miniforge
               platform : win-64

Logs

bash

environment.yml

yaml

~/.condarc

yaml

Source: mamba-org/mamba

View original on GitHubView discussion on GitHub