Scanning
Troubleshooting Scans
A scan that failed, timed out, was refused, or came back with nothing. What each outcome actually means, which of them are the target's behaviour rather than a bug, and what to do next.
- Scan statuses
- 4
- Default scan ceiling
- 300s
- Redirect hops followed
- 5
What each scan status means
Every scan is a row with one of four statuses. Two are in progress and two are final, and a scan that has reached a final status never moves again: if a scan is marked failed by the timeout watchdog and the work later finishes anyway, it stays failed.
| Status | Meaning |
|---|---|
| pending | Accepted and queued, no work started yet. A bulk batch reserves a row for every URL up front, so the last URL in a large batch can sit here for a while. |
| running | Checks are executing. Progress counts up as categories finish. |
| completed | The scan finished. This does not mean findings were found, and it does not mean every check ran: see partial results below. |
| failed | The scan stopped without a result. The reason is stored on the scan and shown on the scanning screen. |
Failure messages, and what to do
Failure reasons are deliberately normalised before you see them: the raw error can contain internal detail, so it is classified into one of a small set of messages. Here is what each one means in practice.
| What you see | What it usually is |
|---|---|
| The target's hostname could not be resolved. | DNS returned nothing for the hostname. Check for a typo, a domain that has not propagated yet, or a host that only resolves inside a private network. |
| The target did not respond in time. | The connection opened but the response never arrived inside the timeout. Common on very slow origins and on hosts that silently drop non-browser traffic. |
| The target refused the connection or closed it early. | Connection refused or reset. Nothing is listening on that port, or a firewall dropped us. Confirm the URL scheme and port are the ones the site actually serves. |
| The target's TLS certificate could not be validated. | Self-signed, expired, wrong hostname, or an incomplete chain. This is a real finding about the site, not a scanner fault: check it in a browser. |
| Cancelled | You stopped the scan. Nothing was recorded. |
| Scan exceeded the 300s time limit. | The watchdog fired. A crawl reports the same thing as 'Crawl scan exceeded the ...s time limit.' against its own, larger budget. See timeouts below. |
| The scan could not be completed because of an internal error. Please try again. | The catch-all. Retry once; if it repeats on the same URL, open a support ticket with the URL and roughly when you ran it, and the error log will have the detail. |
Timeouts
A watchdog gives every scan a hard ceiling, so a target that stalls cannot hold a slot open indefinitely. When it fires, in-flight requests are aborted and the scan is marked failed. The ceilings below are shipped defaults; on a self-hosted instance they are admin settings.
| Scan type | Default ceiling |
|---|---|
| Single URL | 300 seconds |
| Crawl | 900 seconds |
| Bulk batch | 1800 seconds for the whole batch |
Separately, each individual request has its own much shorter timeout: 15 seconds for a page fetch and 8 seconds for a crawled page. A request hitting that limit does not fail the scan: the affected group of checks is reported as incomplete and the rest of the scan continues.
Targets we refuse to scan
Some URLs are rejected before a scan starts. These come back immediately rather than after a wait, and they cost no daily quota.
Internal and private addresses
VulnRadar will not scan anything that resolves inside a private network. That covers literal private and loopback addresses, localhost and anything under .local, .internal or .lan, cloud metadata ranges, and any public hostname whose DNS answer points at a private address. The check is re-run on every redirect hop, so a public URL cannot redirect the scanner onto an internal one.
This is not configurable, and it is the correct behaviour for a hosted scanner: without it, anyone could use the service to probe our own infrastructure. To scan an internal application, run a self-hosted instance inside that network.
Blocked by an operator
An operator can blocklist specific hosts, domains or IP ranges. The response is This target cannot be scanned, with a note that the domain has been restricted for security, privacy or compliance reasons. If you own the domain and believe it was blocked in error, open a support ticket; the block is a deliberate operator decision, so retrying will not clear it.
Active probing on an unverified domain
Active probes send real payloads at a target, so they only run against a domain you have proved you own. Asking for them on an unverified domain is refused outright rather than silently downgraded to a passive scan. Verify the domain first, or run the scan without active probes.
Too many at once
Each plan caps how many scans you may have running simultaneously, separately from the daily total. Hitting it returns a message naming how many you already have running. Wait for one to finish. See Rate Limits.
A scan with no findings
Zero findings is a valid result, and the report says so explicitly rather than looking broken. It usually means exactly what it says. There are a few cases where it means fewer checks ran than you might expect, and all of them are visible on the report:
- You scanned a raw IP. The page-content engine is skipped entirely for a bare IPv4 target. Only DNS, TLS, reputation and the live fetch run. Scan the hostname instead if there is one.
- You scanned over plain HTTP. The whole TLS and certificate family is not applicable and is not planned at all.
- The response was not HTML. Checks that need a document to inspect (scripts, forms, cookies, CSP) declare that requirement and are counted as skipped rather than passed. A JSON API endpoint legitimately produces a short report.
- You scanned a non-HTTP protocol. For a target like an FTP or database URL, only the protocol-specific and banner checks apply.
- You narrowed the check families. Selecting a subset on the scan form does exactly that. The report shows how many checks ran, so compare that number against a full scan.
- Active probes were not requested. They never run unless you ask for them, so an injection-class finding will not appear in a default scan.
Partial results
When one group of checks cannot complete, the scan is not failed and the rest is not thrown away. The report names which group was incomplete: DNS records, TLS and certificate checks, or Live page fetch. The engine confidence figure on the report drops accordingly, which is the honest way to say the result is thinner than usual rather than pretending the missing checks passed.
A repeated incomplete group on the same target is usually a slow or rate-limiting authoritative DNS server, or a TLS handshake the target drops for non-browser clients. Rerunning often clears a one-off.
robots.txt and crawling
A single-URL scan does not ask robots.txt for permission: you supplied the URL. Crawl discovery does read it, and honours Disallow rules from a group naming VulnRadar as the user agent. A blanket User-agent: * group is deliberately ignored, because a generic bot policy should not stop you from scanning your own site.
So if a crawl came back with fewer pages than you expected, check for a User-agent: VulnRadar group in the target’s robots.txt. Disallowed paths are skipped silently: they are not listed and not fetched. Selecting pages by hand bypasses discovery, so a path excluded that way can still be scanned deliberately.
Does a failed scan cost quota
It depends on where it stopped, and the rule is deliberate: your daily count is charged after validation, not before.
| Outcome | Quota |
|---|---|
| Refused as an invalid or blocked URL | Not charged |
| Refused for an unverified domain | Not charged |
| Refused by the concurrency or burst limit | Not charged |
| Started, then failed or timed out | Charged |
| Cancelled by you after it started | Charged |
| Completed | Charged |
A crawl charges per page as it goes rather than up front, so a crawl that stops early only costs the pages it actually scanned. If your remaining quota is smaller than the page count you asked for, the crawl scans what it can and reports the rest as skipped instead of failing.
Still stuck
The scanning screen shows the sanitized failure reason with a copy button under “Scanner output”. Include that string, the URL, and roughly when you ran it in a support ticket and an operator can match it against the server-side error log, which carries the detail the public message deliberately omits.
Running your own instance? The same failures are visible under Admin → Operations → Error Logs with the underlying exception, and Scanner Queue shows anything stuck in progress. Both are covered on the Administration page.