Closed victorlin closed 1 month ago
I'm not sure why this is happened, but it was a bit hard to tell this was the case. I first noticed this when a monkeypox CI run which relied on augur=22.2.0
failed with 20230731T202813Z
. When I debugged on my Macbook using the same build, it pulled augur=22.2.0
. Without realizing I was using a different subdir build, this was confusing. Only after looking into how the Conda runtime's environment is built did I realize what was going on.
Interesting. The augur Conda package is noarch, so the Bioconda channel index shouldn't have different versions between osx-64 and linux-64. The 22.2.0 package was uploaded at 2023-07-31 20:08:01.949000+00:00
. So I guess our builds maybe fell on either side of the index update that included the new version?
I triggered the builds shortly after seeing that Bioconda was updated, so that's very possible. Maybe we just have to wait a bit longer, to reflect what's built into the automatically triggered docker-base rebuild.
No mentions of this since it was reported 9 months ago. We haven't taken any action so it's still possible, but I'm closing for now.
Today, I observed the conda-base equivalent of https://github.com/nextstrain/docker-base/issues/128. In the build for 20230731T202813Z,
augur
was bumped to 22.2.0 forosx-64
but notlinux-64
. Only on the subsequent build for 20230731T212806Z didaugur
get bumped to 22.2.0 forlinux-64
.Possible solutions
23