Skip to main content

Log reasons explained

Every row on Logs has a reason (and on Blocked IPs, a tier). This page lists every reason a sentry can record, what triggers it, and what you can do about it. The portal links each reason cell straight to its section here.

Rows on Blocked IPs are sampled: the same IP and reason is recorded at most once every 30 seconds, and packet-filter logging is rate limited. They show the pattern of what was blocked, not every dropped packet or request.

What visitors see​

When stillorigin answers a blocked request in the web proxy (HTTP, or HTTPS on a domain with Website protection that is being inspected), the visitor gets a page titled Request not accepted · stillorigin and headed This request was not accepted. It has a short reason when one is known, then If you think this is a mistake, contact the website's owner or their host, and tell them when you saw this page., and a small Protected by stillorigin footer. The HTTP status is 403 or, for a rate limit, 429. The page doesn't name the service, the rule or the host.

The reason line is one of these (alphabetical), or is left out when the reason isn't one of them:

  • Reason: automated visitors of this kind are not allowed.: bot blocking (bot-ai, bot-seo, bot-scripts).
  • Reason: this automated service is not allowed.: a blocked identified service (blocked:, blocked by owner:, blocked by default:) or a fake crawler.
  • Reason: this page is limited to allowed addresses.: the WordPress admin login lock.
  • Reason: this request looked like an attack.: built-in WordPress rules, private file and folder escape rules, and WAF rules.
  • Reason: too many requests. Please try again shortly.: a rate limit (429).
  • Reason: visitors from this country are not allowed.: a country rule applied in the web proxy.

A rate-limit 429 shows that reason on the same page, unless Captcha Protection turns it into a browser check.

Blocks enforced by the packet filter (for example the deny list, the naughty list when it is on the deny list, a temporary ban, connection or rate limits at the network, a network-wide identified-traffic block, or a country rule on HTTPS) drop the connection. The visitor doesn't get a page: the site just doesn't load, or times out. The same happens for a blocked HTTPS request on a domain whose HTTPS is passed through.

An error the site itself returns, such as a 404 or the origin's own 403, is passed through unchanged.

Quick reference​

ReasonTierWhere it is enforcedUsual cause
blocked by default: serviceHTTPSentry web proxyYou block that identified service on your clients' domains with Website protection by default, and the domain owner hasn't allowed it
blocked by owner: serviceHTTPSentry web proxyThe domain owner blocks that identified service on their Website protection domain
blocked: serviceIdentified trafficPacket filterYou blocked that identified service (for example GPTBot) on your network or cluster
bot-ai, bot-seo, bot-scriptsHTTPSentry web proxyThe domain owner chose to block that kind of bot
Browser check failedHTTPCaptcha ProtectionThe browser check wasn't completed correctly
Browser check passedAllowedCaptcha ProtectionThe visitor completed the browser check
Browser check: Under Attack (or suspected bot, naughty list, too many requests, chosen page, site under load, country rule)HTTPCaptcha ProtectionThe visitor was shown the browser check
connection limitConnectionPacket filterOne IP held more connections open than the cluster allows
country: countryCountry or HTTPPacket filter (HTTPS) or sentry web proxy (HTTP)A country rule you or the domain owner set: blocked, or only logged while the rule is in log mode
crawler limitCrawlerPacket filterA verified crawler went over its connection limits
deny-listDeny listPacket filterTraffic from an IP on the cluster deny list
excessive trafficRatePacket filterOne IP opened new connections faster than the cluster limit and burst allow
fake-crawlerHTTPSentry web proxyThe request claimed to be a known crawler but didn't come from its official addresses
http-429HTTPSentry web proxyToo many HTTP requests from one IP
http-NNN (any other number)HTTPSentry web proxyThe sentry answered with that HTTP error status
origin limitOriginPacket filterA configured origin, sentry or cluster VIP opened new connections faster than ten times the visitor limit
path-traversalHTTPSentry web proxyURL tries to climb out of the site folder (../)
sensitive-pathHTTPSentry web proxyRequest for a secrets or backup file (.env, .git, .sql, …)
temporary banTemporaryPacket filterTraffic from an IP that is already temporarily banned
unknown-hostHTTPSentry web proxyRequest for a hostname that is not one of your domains with Website protection
uploads-execHTTPWordPress rulesRequest to run a script inside wp-content/uploads
user-enumHTTPWordPress rulesRequest to list WordPress users through the REST API
WAF rules, for example Database attack (rule 942100)HTTPWAF (OWASP Core Rule Set)The request looked like a common attack, such as SQL injection or cross-site scripting
wp-admin-aclHTTPWordPress rulesOlder sentries only: /wp-admin/ from an address not on the cluster's old admin IP list
wp-admin-lockHTTPWordPress ruleswp-login.php or /wp-admin/ from an address that isn't on the domain's admin login lock
wp-includes-phpHTTPWordPress rulesDirect request for a PHP file inside wp-includes
wp-installHTTPWordPress rulesRequest for wp-config.php or the install / upgrade scripts
wp-login-aclHTTPWordPress rulesOlder sentries only: wp-login.php from an address not on the cluster's old admin IP list
xmlrpcHTTPWordPress rulesRequest for /xmlrpc.php

Auth failures have their own tab; see Auth fails.

Packet-filter reasons​

These come from the sentry's packet filter, before a request reaches the web proxy. Limits are per source IP and are set per cluster on Clusters › Edit: New connections per second per IP, Burst (new connections per IP), Open connections per IP, Crawler new connections per second per IP, Crawler burst (new connections per IP) and Temp ban (minutes). The path column shows the destination port, for example tcp/443.

excessive traffic​

Tier: Rate

What triggers it: an IP opened new web connections (ports 80/443) faster than the cluster New connections per second per IP plus Burst. Traffic on connections already open is not counted. The first excess connections are dropped and logged; if the IP keeps going, it is added to the sentry's temporary ban list for Temp ban (minutes). Every excessive traffic event a sentry reports also adds the IP to the cluster Deny list as a timed entry, using your Settings → Deny expiry hours (24 hours when that is not set or is 0). Addresses of your origin servers and sentries are not added: they are limited separately (see origin limit) and an alert is shown on the dashboard instead.

On the Temporary bans tab, excessive traffic is the reason shown for any temporarily banned IP that is not a verified crawler.

What you can do:

  • If the IP is genuinely abusive, nothing: the ban expires on its own. To keep it out longer, add it to Deny list as permanent, or use Promote (permanent) on Temporary bans.
  • If it is a real customer, office or monitoring service (many users behind one NAT address, an uptime checker, an API client), raise New connections per second per IP / Burst on the cluster, then remove the IP from Deny list and use Unban on Temporary bans.
  • If many legitimate visitors are hit at once, the limits are too low for your site; raise them gradually.

connection limit​

Tier: Connection

What triggers it: an IP tried to open a new connection to ports 80/443 while it already had more open connections than the cluster Open connections per IP. The new connection is dropped; existing ones carry on. This does not ban the IP.

What you can do: usually nothing; this stops one address tying up the origin. If a shared address (office, school, mobile carrier NAT, proxy) legitimately needs more parallel connections, raise Open connections per IP on the cluster.

crawler limit​

Tier: Crawler

What triggers it: the IP belongs to a range in an Identified traffic group that has crawler limits on (by default Search engines and AI crawlers; see Verified crawlers) and it went over the crawler limits: more open connections than Open connections per IP, or new connections faster than Crawler new connections per second per IP plus Crawler burst. Excess connections are dropped and logged; a crawler that keeps going is temporarily banned for Temp ban (minutes), and Temporary bans shows the reason crawler limit.

Crawler limits do not add the IP to the deny list.

What you can do: if you want a search engine to crawl faster, raise the crawler limit / Crawler burst; if a crawler is costing too much, lower them. See Scraper throttling. Use Unban on Temporary bans to let a crawler back in immediately.

temporary ban​

Tier: Temporary

What triggers it: traffic (any port) from an IP that is currently on the sentry's temporary ban list, which is filled by the excessive traffic and crawler limit rules. The ban lasts Temp ban (minutes); you can see what is banned right now on Temporary bans.

What you can do: wait for the ban to expire, Unban the IP on Temporary bans (every sentry in the cluster applies it on its next config pull, usually within a minute), or Promote it onto the deny list to keep it out.

deny-list​

Tier: Deny list

What triggers it: traffic (any port) from an IP or range on the cluster's deny list. Entries get there when:

  • you add them on Deny list, or promote a temporary ban;
  • a sentry reports excessive traffic for the IP (timed entry, see above);
  • your origin servers report failed logins from the IP (timed entry, see Auth fails);
  • it is on stillorigin's shared list of known bad addresses.

Timed entries expire after your deny expiry hours. When deny expiry is 0, entries you add by hand stay until you remove them, and automatic entries last 24 hours.

What you can do: if the address is legitimate, remove it on Deny list (for entries you did not add, check the reason column there). If it reappears, find out why on this page (for example repeated failed logins). Contact support if an address you need is blocked and is not on your own deny list.

origin limit​

Tier: Origin

What triggers it: a configured origin server, or a sentry or cluster VIP, opened new web connections (ports 80/443) faster than ten times the cluster New connections per second per IP plus Burst, or held more open connections than ten times Open connections per IP. Excess connections are dropped and logged. The address is not added to the temporary ban list or the deny list, because blocking it would drop return traffic from the origin and take every site on that origin offline.

Replies from the origin on connections the sentry already opened are always accepted.

What you can do: check the origin for malware, a compromised site, or a loop (the server requesting its own sites through the sentry). The dashboard shows an alert when this happens. You cannot add this address to Deny list.

blocked: service​

Tier: Identified traffic

What triggers it: you blocked an identified service, or its whole group (for example AI crawlers), on Settings › Identified traffic, for all clusters or for one cluster. See Block identified traffic. The sentry drops new web connections (ports 80/443) from that service's official IP addresses before any other limit, and logs the service's name, for example blocked: GPTBot. On the Firewall blocks chart these appear as Identified traffic block. Blocked connections are also counted in the Blocked column of Identified traffic. Your origin servers, sentries and cluster VIP are never blocked this way. This needs the latest sentry release: older sentries keep letting the traffic through.

A domain owner can also block a service on their own Website protection domain, if you allow it under Owner access. Those requests are answered with 403 by the sentry's web proxy and logged as blocked by owner: followed by the service's name. They always block, even while Website protection rules are in log mode. Owners can only add blocks: your blocks show as locked in their area. Domain owners see the reason as the service's name, with the note Blocked because this service is blocked on this domain. For a block you set, they see Blocked because your host blocks this service on its network.

If you set a default block on your clients' domains with Website protection (see Default blocks for owners), those requests are answered with 403 and logged as blocked by default: followed by the service's name, until the domain owner allows the service. They always block, even while Website protection rules are in log mode. Domain owners see the note Blocked because your host blocks this service on this domain by default.

What you can do: to lift a network block, untick the service on Settings › Identified traffic (on the tab where it was blocked). To lift a default block, untick it under Default blocks for your clients' domains, or the domain owner can allow the service on their own domain. Blocking a search engine, such as Googlebot or Bingbot, stops your sites appearing in its search results, so the page asks you to confirm first with Block anyway.

country: country​

Tier: Country for HTTPS dropped at the firewall; HTTP for requests the sentry's web proxy answers with 403.

What triggers it: a country rule. You set countries on Settings › Country blocking, for all clusters or for one cluster (a cluster can only add countries). This covers every domain, including domains without Website protection and domains whose HTTPS is passed through.

For HTTPS on port 443, the sentry drops new connections from a blocked country at its firewall, before any other limit. The log shows the country, for example country: CN, and the Firewall blocks chart shows these as Country block.

Plain HTTP on port 80 is stopped by the sentry's web server instead, with a 403, and logged the same way as country: CN. Those requests have tier HTTP (filter HTTP denial) and do not appear on the Firewall blocks chart. Certificate checks on port 80 always get through. See Never blocked by country.

You can also add countries for one Website protection domain on its Countries tab, and a domain owner can add countries on their own Website protection domain if you allow it under Owner access. Those requests are also answered with 403 by the sentry's web proxy, with tier HTTP, and logged the same way. Owners can't change a country you set: it shows as Blocked by your host, Checked by your host or Logged by your host in their area.

A rule can be a Captcha Protection check instead of a block. The visitor is asked to complete a quick browser check before the page opens, logged as Browser check: country rule. Checks only work on domains with Website protection: on domains without Website protection, a check country is not checked or blocked.

Every new country starts in log mode: matching visitors are logged as "would have blocked" and counted, but let through. Switch the country to Enforce when you're happy with what you see.

A country's log mode is separate from a domain's Website protection log mode. Once you click Enforce, a Block always blocks, even while that domain's Website protection rules are in log mode. An enforced Captcha Protection check is only logged while the domain's Website protection rules are in log mode. See Country blocking. Certificate checks are covered on that page too, under Never blocked by country.

Never blocked or checked by country: verified Identified traffic, your never-check addresses (Captcha Protection), your own sentries, origin servers and cluster VIP, and your SSH allow-list. Country rules need the latest sentry release: older sentries ignore them.

What you can do: to stop blocking a country, remove it, or switch it back to log mode, on the tab where it was set. Countries are matched with a country database that's updated monthly, so an address that recently moved country can be matched to its old country for a while.

HTTP reasons​

These come from the sentry's web proxy, so they usually apply only to domains with Website protection enabled. Fake-crawler is also checked on plain HTTP for other domains. On Blocked IPs they have tier HTTP (filter HTTP denial). The request is answered by the sentry with an error status and never reaches your origin, except where noted.

In log mode, path-traversal, sensitive-path and WAF rules don't block: the request is let through and listed under Would have blocked, last 24 hours on the domain's Blocks tab. The automatic Captcha Protection checks (suspicious traffic, chosen pages, under load and country rules) are also only logged in log mode. Every other reason here still blocks in log mode, including Under Attack.

fake-crawler​

What triggers it: Block fake crawlers is on in Settings › Identified traffic, and a request's user agent said it was a known crawler (for example Googlebot, Bingbot, GPTBot or ClaudeBot) but it came from an IP address that crawler doesn't use. Answered with 403, in log mode too. This is checked on domains with Website protection, and on plain HTTP (port 80) for other domains. HTTPS on domains without Website protection isn't checked, because the sentry can't read the request. Real crawlers from their official addresses are never blocked by this check. On Logs the reason is fake-crawler; domain owners see it as Fake crawler.

What you can do: nothing for fake crawlers. If you run your own crawler that uses one of these names, change its user agent, or turn off Block fake crawlers.

unknown-host​

What triggers it: a request reached the sentry for a hostname that is not one of your domains with Website protection on that cluster (for example a scanner hitting the bare IP address, or a domain whose DNS still points at the sentry after it was removed). Answered with 403.

The Domain column on Logs shows the hostname the request asked for (or the IP address, if that's what the visitor used).

What you can do: nothing for scanners. If it is a site you host, add the hostname (including www. variants) on Domains with Website protection enabled, or point its DNS elsewhere.

path-traversal​

What triggers it: the URL contains ../, /.. or the encoded form %2e%2e, which is used to read files outside the website folder. Answered with 403.

What you can do: nothing; this is attack traffic. Consider adding persistent sources to Deny list.

sensitive-path​

What triggers it: a request for a file that usually holds secrets or backups: anything containing .env, .git or /debug.log, or ending in .sql or .bak. Answered with 403.

What you can do: nothing; these probes are common. If your site really serves such a URL, rename it. Never keep these files in your web root.

xmlrpc​

What triggers it: a request for /xmlrpc.php on a domain with type WordPress. XML-RPC is a common target for password guessing and pingback abuse. Answered with 403.

What you can do: if you rely on XML-RPC (for example the old Jetpack connection or an old mobile app), change the domain type to Generic, which removes all WordPress rules, or contact support.

uploads-exec​

What triggers it: on a WordPress domain, a request for a script file (.php, .phtml, .phar, .php5, .cgi, .pl) inside /wp-content/uploads/. Uploads should only contain media; scripts there are a sign of a planted backdoor. Answered with 403.

What you can do: check the uploads folder on your origin for unexpected script files and remove them.

wp-includes-php​

What triggers it: on a WordPress domain, a direct request for a .php file inside /wp-includes/. WordPress never needs visitors to load these directly. Answered with 403.

What you can do: nothing; this is probing.

wp-install​

What triggers it: on a WordPress domain, a request for wp-config.php, wp-admin/install.php or wp-admin/upgrade.php. Answered with 403.

What you can do: nothing for probes. If you need to run a WordPress database upgrade, do it from the origin directly or ask support.

user-enum​

What triggers it: on a WordPress domain, a request to /wp-json/wp/v2/users, which lists user names that attackers then use for password guessing. Answered with 403.

What you can do: nothing usually. If a plugin or headless front end needs this endpoint, change the domain type to Generic or contact support.

wp-admin-lock​

What triggers it: on a WordPress domain with Website protection on and its Admin login lock turned on, a request for /wp-login.php or anything under /wp-admin/ from an IP address that isn't on the lock's list. /wp-admin/admin-ajax.php stays open, because themes and plugins call it from public pages. Answered with 403, in log mode too. The lock is off unless you turn it on, or the domain owner does if you allow it on Owner access.

What you can do: if this is you or your customer logging in, open the domain's page and add the address under WordPress rules › Admin login lock, then click Save addresses. Or click Turn off lock. Separately, everyone is limited to about 4 requests a minute for wp-login.php per IP address (a few extra are allowed in a burst); going over that is logged as http-429.

wp-login-acl​

Only sentries on an older version set this reason. Current sentries leave the login page open unless the domain's admin login lock is on.

What triggers it: on a WordPress domain, a request for /wp-login.php on a sentry that hasn't updated yet, from an address that isn't on the cluster's old admin IP list. Answered with 403.

What you can do: update the sentry. Click Force on its row on the cluster's Edit page; see Sentry updates.

wp-admin-acl​

Only sentries on an older version set this reason; see wp-login-acl.

What triggers it: on a WordPress domain, a request under /wp-admin/ (except admin-ajax.php) on a sentry that hasn't updated yet, from an address that isn't on the cluster's old admin IP list. Answered with 403.

What you can do: as for wp-login-acl: update the sentry.

WAF rules​

The WAF is off until you turn it on for the cluster. Tick Use the WAF (OWASP Core Rule Set) on domains with Website protection from the cluster's Edit page. Until you do, your sentries don't check requests with the WAF, and the WAF card on each Website protection domain in that cluster shows Off and only this note: The WAF is off in this domain's cluster settings, so it isn't checking requests to this domain. The controls to turn the WAF off for a domain, or to ignore rules, stay hidden.

What triggers it: on a domain with Website protection with the WAF on, the request looked like a common attack, such as SQL injection or cross-site scripting. The WAF uses the OWASP Core Rule Set at paranoia level 1, the level designed to stop attacks with the fewest false alarms. Each rule that matches adds points, depending on how serious the match is, and a request is blocked once it reaches 5 points. One serious match is enough on its own. Answered with 403, or with 400 when the data sent with the request couldn't be read.

The reason shows the kind of attack and the main rule that matched, for example Database attack (rule 942100). If the rule isn't known, the reason is Attack blocked by the WAF. The WAF checks requests, not your site's responses. It records the request path without its query string, and it doesn't record the visitor's user agent.

On domains with type WordPress or Drupal, the WAF also uses the official OWASP Core Rule Set tuning for that platform, so everyday actions such as saving a post aren't mistaken for attacks. Joomla and Generic domains get the standard rules. Make sure each domain's Type is set correctly on Domains.

Like the other Website protection rules, the WAF can only check requests the sentry can read: plain HTTP, and HTTPS once it's inspected.

In log mode, the request is let through and listed under Would have blocked, last 24 hours on the domain's Blocks tab.

Reason shownRule numbersWhat it means
Request type not allowed911xxxThe request used an HTTP method the site doesn't accept.
Attack tool or scanner913xxxThe request came from a tool that searches sites for weaknesses.
Malformed request920xxxThe request broke the rules for how web requests are built, which attack tools often do.
Request tampering921xxxThe request's headers were built to trick the server, for example to sneak a second request in (request smuggling or header injection).
Tampered form upload922xxxA form or file upload was built in a way that's used to slip attacks past security checks.
Attempt to read server files930xxxThe request tried to read files on the server, such as system or configuration files (local file inclusion).
Attempt to load outside code931xxxThe request tried to make the site load a file from another website (remote file inclusion).
Attempt to run server commands932xxxThe request tried to run commands on the server (remote code execution).
Attempt to run PHP code933xxxThe request tried to run PHP code on the site (PHP injection).
Attempt to run other code934xxxThe request tried to run other code on the server, such as Node.js or Python.
Script injection941xxxThe request tried to put a script into a page, to run in visitors' browsers (cross-site scripting, XSS).
Database attack942xxxThe request tried to read or change the site's database (SQL injection).
Session hijack attempt943xxxThe request tried to plant a login session so an attacker could take it over (session fixation).
Attempt to exploit Java944xxxThe request tried to exploit Java software on the server.
Unreadable request data200000–200099The data sent with the request, such as a form, couldn't be read, so it couldn't be checked.
Other security ruleAny otherAnother rule in the OWASP Core Rule Set.

What you can do: nothing for attack traffic. If a rule blocks real visitors, or would block them in log mode:

  • On Edit domain, under Website protection rules, click Ignore on this domain next to the rule under Would have blocked, last 24 hours. Or, under WAF, enter the rule number under Ignored rules and click Ignore rule. The rule number is in the reason, on Logs, or on the domain's Blocks tab.
  • If the request also matched other rules, they can still block it. Check the reason again, and ignore that rule too if it's also a false alarm. You can ignore up to 100 rules on a domain.
  • Only if that doesn't help, click Turn off WAF on the domain page. Rate limits and the other Website protection rules still apply.

Only an account owner can ignore rules or turn the WAF off for a domain. To turn the WAF off again for the whole cluster, untick Use the WAF (OWASP Core Rule Set) on domains with Website protection from the cluster's Edit page.

WAF rule updates​

The OWASP Core Rule Set gets new releases from time to time. On Settings › WAF rule updates, choose when your sentries install them: As soon as they're released, 7 days after release, 30 days after release or Don't install automatically. This is separate from automatic sentry updates. The default is 7 days after release.

With Don't install automatically, your sentries keep their current rules until an nginx update needs a new set, or until you press Install now. A new sentry installs the latest rules when it's set up. Install urgent fixes as soon as they're released is on by default. It installs urgent fixes at once, whichever option you choose above.

Under the update choice, Install now installs the newest rules on all your sentries straight away, whichever update option you chose above. It applies to every sentry in every cluster, and it shows for all four update choices whenever a sentry doesn't have the newest rules. It doesn't wait for the delay or the cluster's update window: each sentry starts within about a minute. It only covers rule sets already released when you press it; later releases follow the option you chose, as usual. Only an account owner can see and press it, and each press that queues an install is recorded in the audit log.

After a sentry installs new rules, it reloads nginx so the new rules take effect straight away.

The page also shows Newest rules installed on how many of your sentries, and a status for each one. Next to the status it may show the installed version, as Rules followed by the version. Statuses you may see:

  • Newest rules
  • Installing rules
  • Newer rules ready, waiting for your update setting
  • Offline. Installs the newest rules when it's back online.
  • Couldn't install the rules. The sentry is trying again.
  • Can't install yet: there are no rules for this sentry's nginx version
  • Can't install: the rules failed the security check, so nothing was changed
  • Can't install: updates are turned off in this sentry's own settings
  • Update this sentry to see its rules — the sentry needs a newer release before it can show its rules
  • Checking rules

bot-ai, bot-seo, bot-scripts​

What triggers it: the domain owner turned on bot blocking in the owner area, and the request came from that kind of bot. Answered with 403.

  • bot-ai: an AI crawler, such as GPTBot, ClaudeBot or CCBot.
  • bot-seo: an SEO or marketing crawler, such as AhrefsBot, SemrushBot or MJ12bot.
  • bot-scripts: a command-line tool, code library or headless browser, such as curl or python-requests. This can also block uptime monitors and other services that connect to the site.

Search engines such as Google and Bing, and certificate checks, are never blocked. Bot blocking always blocks, even while Website protection rules are in log mode.

What you can do: if the owner is blocking something they need, such as your uptime monitor, ask them to change their bot blocking. To stop owners blocking bots, untick Block AI crawlers, SEO crawlers and scripts on Owner access. Their choice is then paused (see Paused settings).

Browser check, Browser check passed, Browser check failed​

What triggers it: Captcha Protection asked the visitor to complete a quick browser check before the page opened. On Logs the reason says why:

  • Browser check: chosen page: the visitor asked for one of the chosen pages, such as a login page.
  • Browser check: country rule: a country rule on your network or on the domain is set to a Captcha Protection check. See Country blocking.
  • Browser check: naughty list: the address is on the stillorigin naughty list, and you aren't already dropping those addresses with Use the stillorigin naughty list.
  • Browser check: site under load: the domain's origin is slow or failing to answer.
  • Browser check: suspected bot: the request looked like a bot but didn't come from a known service.
  • Browser check: too many requests: the visitor went over the rate limit, so they got the check instead of an error.
  • Browser check: Under Attack: Under Attack is on for the domain, its cluster or all clusters, or the domain owner turned it on.

Browser check passed means the visitor completed the check; they can then keep browsing without another check, and the pass ends after 30 minutes without opening a page. Browser check failed means the check wasn't completed correctly. Scripts and other non-browser clients get a short plain-text answer (403) instead of the check page.

In log mode, Suspicious traffic, Chosen pages and When the site is under load don't check visitors. The check they would have got is only logged, and shows under Would have blocked, last 24 hours on the domain's Blocks tab. Under Attack still checks visitors in log mode.

What you can do: if real visitors are checked too often, turn off the trigger in Settings › Captcha Protection, or add their addresses to Never check these addresses. Turn off Under Attack once the attack is over.

http-429​

What triggers it: too many HTTP requests from one IP: more requests per second than the cluster New connections per second per IP value plus Burst on a domain with Website protection (on domains with Website protection the same numbers also limit HTTP requests), or, on a WordPress domain, more than about 4 wp-login.php requests per minute. If the domain owner has set their own rate limit, their limit applies to that site instead of the cluster's. Answered with 429 (Too Many Requests). Visitors see the stillorigin block page, unless Captcha Protection is set to check suspicious traffic and turns that 429 into a browser check.

What you can do: as for excessive traffic: if real visitors are hitting it, raise New connections per second per IP / Burst on the cluster.

http-NNN (other statuses)​

What triggers it: the sentry answered with an error status without one of the reasons above. The number is the status, for example http-403.

Your origin's own error responses, such as a 404 for a missing page, a 500 from the site, or a 502 or 504 when the sentry couldn't reach your origin, are not blocks. They're counted as allowed traffic and aren't listed here. Older rows can still show them, for example http-404 or http-502.

What you can do: usually nothing. For many 502 or 504 errors, check that the origin web server is running, reachable from the cluster and not overloaded.

Auth fails​

The Auth fails tab is for failed logins reported from your origin servers. Each row is one report: the IP, the service (for example ssh), and how many failures were seen in the reporting window.

Reporting failed logins from origin servers is not available yet, so this tab is normally empty.

What happens: each report also adds the IP to the cluster Deny list as a timed entry with the reason <service> auth fail, for your deny expiry hours (24 hours when it's not set or 0). After that, the IP's traffic shows on Blocked IPs as deny-list.

What you can do: if a legitimate user mistyped a password, remove their IP on Deny list. If you see many reports, make sure the service uses key-based or two-factor login and is not open to the internet more than it needs to be.

Exporting​

Export CSV under each Logs table downloads every row that matches the current tab, filters, search and sort (up to 100,000 rows; the file ends with a note if that limit was reached). Times in the file are UTC.