mirror of
https://github.com/clearlinux/koji-setup-scripts.git
synced 2026-04-28 11:03:50 +00:00
f35848b8e0
Because the mash script will always need to run on the kojimaster server, it has local access to RPMs. Rather than copying the data, create a symlink to the data. This reduces disk usage and allows the mash hosting location to be flexible. Because the mash is re-generated every time kojira runs and creates a new koji repository, by nature, the mash is an iterative process. Existing repository metadata is always available (excpet for the first run). Pass this existing metadata to createrepo_c so that it may more efficiently diff and recalculate the updated metadata in order to reduce CPU and disk usage. By reducing disk and CPU usage, this reduces the impact of running this process on a kojimaster which is a shared system responsible for hosting many resources. Signed-off-by: George T Kramer <george.t.kramer@intel.com>
Assumes:
* All scripts are run as the root user
* Basic configurations (e.g. network, time, etc.) have been applied to the host to allow applications to run well
How these scripts work:
#. Edit parameters.sh as needed
#. Run (playbook does only the following)
#. deploy-koji.sh
#. bootstrap-build.sh
#. Optionally, for supporting a full DevOps workflow, also run
#. deploy-mash.sh
#. deploy-git.sh
#. deploy-upstreams.sh
If koji builder machine is not the same as koji master machine:
#. On the koji master machine, run
#. deploy-koji-nfs-server.sh
#. On the koji builder machine, run
#. deploy-koji-nfs-client.sh
#. deploy-koji-builder.sh