Hi Loris,
I assume you used "eb --rebuild --skip" here (as documented in
https://docs.easybuild.io/partial-installations/#partial_installation_skip).
The assumption here is that there is a module file already.
If not, then EasyBuild will assume there is no installation yet, and
start again from scratch.
For a partial installation, you can try to forcibly create a module file
using "eb --module-only --force" before proceeding with installing the
missing bits using "eb --rebuild --skip".
For installation bundles of R packages, I also recommend using the
--parallel-extensions-install option, which will result in launching the
installation of multiple R packages in parallel. That should save quite
a bit of time (but total installation will still take hours).
This only works for bundles of R packages, not for Python packages, Perl
modules, etc., mainly because we need a reliable way of determining
which packages have to be installed before the installation of a
particular package can be started. That's easy to do for R packages
thanks to good metadata, but it's very messy for Python packages,
unfortunately...
regards,
Kenneth
On 05/06/2025 10:12, Loris Bennett wrote:
Hi,
Building R-bundle-CRAN-2023.12-foss-2023a.eb fails after over 7 hours
due to an error (which I also encountered two years ago
https://www.mail-archive.com/[email protected]/msg06737.html>>
== installing extension GEOmap 2.5-5 (389/432)...
[snip (24 lines)]
== ... (took 7 hours 31 mins 4 secs)
Thus I have a solution and want to rebuild the module.
However, although the results of the build were not deleted after
failure, when I restart the build, all the packages are built again from
scratch.
Should this be happening? Can I prevent it?
Cheers,
Loris
