-
-
Notifications
You must be signed in to change notification settings - Fork 5
Restore Imager worker
We suppose you have a freshly installed Debian machine / VM
On Scaleway, machines are configured with 2 RAID1 volumes for /boot and /. With the current setup, we want the two 1TB SSD to be on RAID0 for cache+building dir.
# /!\ following is example only. assumes /dev/md1 is composed of /dev/sda3 and /dev/sdb3
# check status of md1
mdadm --detail /dev/md1
# we need to wait until it's clean (synced)
# change level to zero. will remove one member
mdadm /dev/md1 --grow --level=0
# check the name of the device that got removed
mdadm --detail /dev/md1
# and add it back
mdadm /dev/md1 --grow --raid-devices=2 --add /dev/sda3
# should answer that it's now RAID4.
# Wait for it to rebuild again and indicates raid0 with 2 members
mdadm --detail /dev/md1
mdadm --detail /dev/md1
# then resize the filesystem to grow up to the array size
resize2fs -f -p /dev/md1Follow the normal node setup procedure (up to what is specific to k8s): https://github.com/kiwix/k8s/wiki/Machine-and-k8s-node-Setup
Install Docker by following documentation at https://docs.docker.com/engine/install/debian/
Restore the imager configuration at /etc/imager.config ; this file contains secret values (to connect to S3) so it is kept by admins in their Key Vault
Sample file:
USERNAME="--------------"
PASSWORD="--------------"
CONTAINER="imager-worker"
IMAGE="ghcr.io/offspot/imager-worker:latest"
#IMAGE="ghcr.io/offspot/cardshop-worker:imager"
S3_ACCESS_KEY="-------------"
S3_SECRET_KEY="--------------------"Restore the startup script from https://github.com/offspot/imager-service/blob/main/contrib/imager-worker-restart.sh to /usr/local/bin/imager-worker-restart
curl -L -o /usr/local/bin/imager-worker-restart https://raw.githubusercontent.com/offspot/imager-service/main/contrib/imager-worker-restart.sh
chmod +x /usr/local/bin/imager-worker-restart
mkdir -p /data/imagerRestore the startup script from https://raw.githubusercontent.com/offspot/imager-service/main/contrib/cache-policy.yaml to /data/cache/policy.yaml
mkdir -p /data/cache
curl -L -o /data/cache/policy.yaml https://raw.githubusercontent.com/offspot/imager-service/main/contrib/cache-policy.yamlIf worker IP has changed, update the Wasabi policy containing IP whitelists (remove old worker IP / add new worker IP): arn:aws:iam::100000050990:policy/CardshopWorkers
imager-worker-restart