Repository navigation
Release v4.5.0
Highlights
- Automated file fixes and cleaning now prevent account owners from changing files outside their account. Upgrade every host.
- With automatic blocking on, sources probing for the PHP2Shell flaw are now blocked at the firewall on their first blocked request.
- PHP hardening audits now flag unsafe fatal error backtrace settings on PHP 8.5 and later.
- Central intelligence downloads now support node authentication with an existing reporting identity.
- The Email Security page now shows mail decision counts, monitoring gaps, recent rejects, rspamd status and groups needing attention.
- Validation and doctor now warn when the web UI accepts admin logins from any network without a CSM restriction.
- ModSecurity warnings alone no longer trigger firewall bans.
Security
- Automated file fixes and cleaning now prevent account owners from changing files outside their account.
- With automatic blocking on, sources probing for the PHP2Shell flaw are now blocked at the firewall on their first blocked request.
- Failed central intelligence downloads no longer expose download addresses or access tokens in logs.
- PHP hardening audits no longer report unsafe PHP-FPM settings as safe.
18 changes in this release. Full detail: CHANGELOG.md
Full changelog
Highlights
- Automated file fixes and cleaning now prevent account owners from changing files outside their account. Upgrade every host.
- With automatic blocking on, sources probing for the PHP2Shell flaw are now blocked at the firewall on their first blocked request.
- PHP hardening audits now flag unsafe fatal error backtrace settings on PHP 8.5 and later.
- Central intelligence downloads now support node authentication with an existing reporting identity.
- The Email Security page now shows mail decision counts, monitoring gaps, recent rejects, rspamd status and groups needing attention.
- Validation and doctor now warn when the web UI accepts admin logins from any network without a CSM restriction.
- ModSecurity warnings alone no longer trigger firewall bans.
Security
- Automated file fixes and cleaning now prevent account owners from changing files outside their account.
- With automatic blocking on, sources probing for the PHP2Shell flaw are now blocked at the firewall on their first blocked request.
- Failed central intelligence downloads no longer expose download addresses or access tokens in logs.
- PHP hardening audits no longer report unsafe PHP-FPM settings as safe.
Added
- Central intelligence downloads now support node authentication with an existing reporting identity. Select that identity in central settings and restart to enable it.
- Integrations can now read mail decision statistics, recent rejects, available email checks and rspamd health through the API.
Changed
- PHP hardening audits now flag unsafe fatal error backtrace settings on PHP 8.5 and later.
- Validation and doctor now warn when the web UI accepts admin logins from any network and neither the CSM firewall nor token address limits restrict it.
- The Email Security page now shows persistent mail counts, a chart, monitoring gaps, recent rejects, rspamd status and groups needing attention. Follow the antispam guide if rspamd counters are unavailable.
- Docs and command output now describe the integrity check as recorded hashes, not signatures, and note what it cannot stop.
Fixed
Firewall and blocking
- ModSecurity warnings alone no longer trigger firewall bans.
- Disabling firewall escalation for a ModSecurity rule now prevents further bans from its earlier matches.
- Firewall-only restores now import consistent firewall state and action history, refusing unfinished actions or undelivered audits without changing existing data.
Login and mail monitoring
- IPv6 loopback logins no longer raise alerts, and untrackable SSH hostnames produce a single diagnostic warning.
- Mail login tracking now skips private IPv6 clients as it already skipped private IPv4 ranges, so internal clients are not blocked.
Automatic response
- Automatic permission fixes no longer change a file that stopped being world- or group-writable after the scan.
- Findings resolved by the automatic permission fix now leave the Findings page instead of staying listed until the next scan.
Hardening audit
- The PHP end-of-life audit no longer says old branches get no security fixes; it points to vendor backports instead.