Repository navigation
Troubleshooting
Start with cvescope status (feed health) and cvescope inventory -v (each collector, with a
traceback when one fails). cvescope sync -v does the same for the feeds.
The collector talks HTTPS to a device and you haven't said how to trust its certificate. Run
cvescope pin host:port, compare the fingerprint with the one in the device's UI, and add
pin_sha256. See Configuration.
The device's certificate changed, usually a renewal. Re-run cvescope pin, check the new
fingerprint against the device, and update the config. If you didn't renew anything, find out
why before you trust the new one.
Secrets can't live in the config file. Move the value to a file (one line) or an environment variable and point at it.
The device was unreachable or refused the credentials, so cvescope kept the versions from the
last good run. cvescope inventory -v shows the error.
-
Pi-hole:
Pi-hole login failedmeans the password file is wrong. An app password works. - Home Assistant: the OS and add-on versions need an admin token; Core works with any long-lived token. On Proxmox, the guest agent reports Home Assistant OS anyway.
-
Proxmox guests without an OS: the QEMU guest agent isn't installed or running in that VM.
Containers never report a version; add a
dpkgcollector for them.
Install osquery, or set osqueryi = "/path/to/osqueryi" in the block. cvescope looks on PATH
and in C:\Program Files\osquery, /usr/local/bin and /opt/osquery/bin.
The dpkg command needs the os-release part, so cvescope knows the release and can ask the
distro's tracker. Use the full command from Collectors, then run cvescope sync (the tracker
for that release downloads on the next sync).
Its tag floats (latest, stable) and the image has no version label. Pin a version tag.
The CVE List hasn't been imported. Run cvescope sync and let the first download finish.
cvescope was off long enough that the daily CVE List deltas it needs are gone. --rebaseline
downloads the full snapshot again.
The last full sync is older than serve.stale_after_hours, or none has run since you upgraded.
Check the scheduled task (Get-ScheduledTaskInfo cvescope-sync on Windows, the cron log
elsewhere) and the end of the sync log. An amber banner means the last sync ran but a feed or
collector failed; it names which one, and cvescope inventory -v or cvescope sync -v shows why.
You opened it under a name it doesn't serve: it only answers on 127.0.0.1 and localhost by
default. If you deliberately bound it wider, add the name to serve.allowed_hosts, and put
authentication in front of it.
That's by design: every feed's first load counts as history. Alerts start with the next change.
Run cvescope show CVE-YYYY-NNNNN to see the record's affected entries and the reason cvescope
gave, then open a wrong-match issue.
Product, version and CVE ID are enough: leave out hostnames and IPs.