2026.05.b
May 9, 2026
- NewRun-scan now waits up to 90s for the next push-agent snapshot when SSH is unavailable, so drift introduced seconds before the click is detected on the first scan instead of the next one.
- NewSnapshot-freshness pill on the dashboard shows how recent the latest fleet signal is at a glance, with green/amber/red thresholds.
- NewOnboarding wizard surfaces the live collector status ("Waiting for fresh agent snapshot…") so first-scan flows feel transparent rather than stalled.
- PerformanceDefault push-agent cadence reduced from 5 minutes to 60 seconds (systemd timer + cron fallback). Every scan now reflects state captured within the last minute.
- FixScan-job projection no longer fakes a "succeeded" status for real scans before the collector resolves them — the dashboard refresh now lines up with actual drift events.
- FixMulti-instance scan tracking: scan kind + progress now persist to Redis, so a poll routed to a different web container still sees the right state.