See current issues with the cluster.
Whenever possible, changes should be made by editing update.sh
and then running this on all machines in the cluster. Todo this, proceed as
follows:
- Make the change (ideally as a PR, continue here when merged)
- Do
git pull
in/shared/config
(on any of the machines, this is shared via NFS) - Open a
tmux
- Run
shoprun /shared/config/update.sh
. This will spawn one tmux window per machine, each running the script. On some machines, you will need to enter a password for sudo. (If you don't have direct-to-root ssh, this may require changes toshoprun
). Seeshoprun
. - Babysit all the shell sessions, hitting Enter and confirming as necessary.
Make sure every session says
COMPLETED SUCCESSFULLY
at the end. If it didn't, troubleshoot.
Note that update.sh
is written to be idempotent, i.e. it should be possible to
rerun it as often as needed, and no changes should be made on the second run.