- Prepare at least two node with an operation system (Centos7.8 minimal installation). The worker node should have GPU device.
- Configure static IP for all nodes
- Upload AIStation installation package to the master node on which you are running the deploy scripts.
- Perform MD5 verification to verify the integrity of the installation package, then unzip it and generate a deploy-script directory.
- Planning your cluster, modify the node information into file
inventory.ini
. You can reference the sample filesinventory.ini.single
(non-HA deployment example) andinventory.ini.muti
(HA deployment example). - Modify the file:
inspur.yml
. You can reference the sample filesinventory.yml.single
(non-HA deployment example) andinventory.yml.muti
(HA deployment example). - Execute the following commands
# user means linux username
# password means the linux password
# local_iP means the ip address of the node where you execute the command
# step_NO means the installation steps. You can choose one from below:
# - If you just want to get k8s clusters, set step_NO to 1 and 9 respectively and execute in order.
# - If you want to install the complete aistation software, you should choose which you need from your installation manual, then set step_NO respectively and execute in order.
$ bash -x install.sh [user] [password] [local_ip] [step_NO]
- After all process complete, all the nodes will reach "Ready" status and pods will reach "Running" status.
- Download a binary release of the sonobuoy CLI from https://github.com/heptio/sonobuoy/releases:
$ sonobuoy run --mode=certified-conformance
- Get the results:
$ outfile=$(sonobuoy retrieve)
$ mkdir ./results; tar xzf $outfile -C ./results
- Then grab
plugins/e2e/results/global/{e2e.log,junit_01.xml}
from results folder, add them to this PR.