| Installer refuses the chosen disk, or a destructive step is blocked |
The disk (or a partition on it) is currently mounted |
Re-check with lsblk -f and pick the correct spare device — the installer's mount check is deliberate and will not be bypassed |
| "Could not determine disk-by-id path" during the cluster role |
detect-disk-id.sh was not run on the peer node first, or the value was mistyped/reordered |
Run detect-disk-id.sh on BOTH nodes before starting either real install; keep disk order identical on both sides |
| Cluster installer refuses to run on an already-deployed cluster |
Intentional data-safety guard — the live configuration already contains a VIP resource |
To rebuild from scratch (destroys all data): pcs cluster stop --all; pcs cluster destroy --all, then re-run secondary, then primary |
| Cluster primary step fails with "DRBD ... never connected" / "not in a fresh state" |
Only the primary was (re-)run after a partial failure |
Re-run BOTH nodes, secondary first, then primary |
| SSH access refused after several failed login attempts |
fail2ban's sshd jail has banned the source IP |
Wait for the ban to expire (default 10 minutes), or from a still-reachable session: fail2ban-client set sshd unbanip <ip> |
| Cannot SSH in as root after install |
Remote root login is disabled by design at the end of installation |
Use the console, or the local "support" operator account, or the GUI |
Browser shows a certificate warning at https://<host>:8443/ |
The GUI uses a self-signed certificate until a real one is uploaded |
Expected — proceed past the warning, or upload a CA certificate under Configuration → TLS certificate |
| Standalone node's dashboard shows cluster/quorum/DRBD error panels |
The dashboard code has cluster-first defaults; those panels genuinely don't apply to a standalone node |
Cosmetic only, safe to ignore — confirmed non-fatal, every panel fails independently |
| Firewall page shows "No firewall apply has been run yet" |
The default ruleset has not been applied on this node yet |
Configuration → Firewall → Apply, once per node |
| Package install fails during an offline/unattended run ("no installation candidate") |
A stray active deb cdrom: line in /etc/apt/sources.list breaks apt-get update on that node |
Comment out the cdrom line in /etc/apt/sources.list and retry; current ISO builds no longer ship with this issue |
| "sudo: unable to resolve host \<name>" warnings during install |
The node has no DNS configured yet at that point in the install |
Cosmetic only — does not affect the outcome, resolves itself once DNS is configured |
| Config backup or NTP/timezone change appears to have no effect |
Change was saved on one node only; these are per-node settings |
Confirm the same value on BOTH cluster nodes — Network/NTP/timezone settings are intentionally not auto-replicated |
| Need to redeploy just the GUI without a full reinstall |
— |
Run installer/install-gui.sh [--config FILE] on the node (needs /etc/nascore/cluster.conf, already written by the role) |