- Move positions.yaml from /tmp (wiped on reboot) to /var/lib/promtail so
promtail resumes where it left off after a reboot instead of either
re-ingesting (duplicate lines in Loki) or skipping a chunk when a log
was rotated during downtime.
- Create /var/lib/promtail with promtail:nogroup ownership (the
apt-installed promtail user's primary group is nogroup).
- systemctl enable promtail so the service auto-starts on boot; previously
only restart was called, so promtail did not survive a VM reboot.
- set -euo pipefail + DEBIAN_FRONTEND=noninteractive so the script aborts
on first failure and runs cleanly under automation, without debconf
prompts on minimal hosts.
- Fix shebang typo (#/bin/bash -> #!/bin/bash).
when defining Zero-OS as first boot, we can not get the node back
into rescue mode as that is the way Hetzner functions:
tne netboot get an ipxe that doe a local reload or, in case of rescue
starts a full netboot
removed zos-on-hetzner as it's more generic
added zosinstalluefipxeondisk.sh as generic isntall script
to install the bootblock on a node's disk
select zos4 as main driver for zos
