Sage Infrastructure Solutions Group IncSage Infrastructure Solutions Group IncLoading…

Sage Infrastructure Solutions Group Inc

Break-glass access

Emergency access is part of the design, not a gap in it. Keyreeve locks SSH down — hosts accept Keyreeve certificates only, with authorized_keys files, passwords and keyboard-interactive logins turned off — so you need a way in that doesn't depend on Keyreeve, for when Keyreeve or your identity provider is unavailable, or hosts and people can't reach it.

Break-glass policies are under Admin → Break-glass (on the SSH access rules page).

How it works

A break-glass policy names local accounts (never root) and the hosts or host groups it applies to. If several policies match a host, the highest priority wins. For each account you choose:

  • where from — the policy's source networks, networks of its own, or anywhere;
  • how — a key, a password, or both.

Credentials can be handled in two ways:

  • The host's own (the default): Keyreeve never creates the accounts or touches their passwords or authorized_keys; you manage them, for example in your password vault. Such a login switches off everything of Keyreeve's for that account, so it works with the Keyreeve server down and with the Keyreeve agent stopped.
  • Keyreeve manages them: passwords are sealed at rest, can be generated for you (shown once) and rotated, and are applied with the host's own chpasswd.

Either way, a break-glass login goes straight to the host's sshd and never contacts the Keyreeve server — which is the point.

[!NOTE] Licensing never blocks break-glass access, and when a host refuses logins because it can't record them, break-glass is still let in by default. Only the "refuse everyone" failure policy shuts it out too.

You will know it was used

  • Every break-glass login is reported by the host and written to the audit log at error severity.
  • Each policy can notify email addresses and a webhook whenever one of its accounts is used, so your team hears about it straight away. An account used by automation (an SSH tunnel, say) can be kept in the log without the notifications.
  • For a shared account, Keyreeve reports which key was used — from the comment on that key's line — so you can tell who it was.
  • The Home dashboard shows break-glass logins over the last seven days, and Audit → Logins lists every login hosts accepted.

Logins with the host's own credentials go around Keyreeve entirely, so they are not recorded; sshd's log and the login index still show every one. Keyreeve-managed accounts go through the recorder when it is available.

Logins that bypass Keyreeve

Audit → Logins flags any login that arrived without a Keyreeve certificate (shown as Other key or password), so access that went around the gateway is reported rather than invisible. On the host, the session guard ends any login whose processes escape the recorder, and with auditd the kernel's audit log lists the programs every session ran.

For Kubernetes the equivalent is covered in Kubernetes access.