Open Trzs opened 6 months ago
$ time git clone git@github.com:cctbx/cctbx_project.git
real 0m44.325s
$ du -hs cctbx_project/
255M cctbx_project/
$ time git clone --depth=1 git@github.com:cctbx/cctbx_project.git
real 0m12.546s
$ du -hs cctbx_project/
157M cctbx_project/
XFEL CI on Azure is so tight on space this kinda thing can help (plus it's good dinosaur management). But how does this affect developers? With depth=1, the old commit history is literally not there. Git log shows only the latest commit. So I think we'd only want this for Azure, right? Maybe a flag to bootstrap update that we can set in Azure?
How much more disk space is needed? There is a disk clean up step in
You essentially have root access to the virtual machine on Azure so you can delete whatever you want. There is sort of a limit in that in this pipeline, you are already inside the Docker image, but you can always make other host directories writeable here
For example, in some pipelines, I clean up more stuff and make a swap file
https://github.com/phenix-project/phenix-installer/blob/main/scripts/clean_linux.sh
But that is done in the normal Azure image, not the Docker image. I get about 47 GB free in / after this step. Also, you do not need to use the 14 GB partition, you can use any partition on the image.
I can probably help more next week once I get back.
@phyy-nx to get the full git history, you can run either git fetch --unshallow
or git pull --unshallow
--depth=1
to git clone commands in bootstrap.py, so that only the necessary data is downloaded in the first placegit fetch --unshallow
Note to @nksauter, @bkpoon