Skip to content

Reading the WAF log and fixing false alarms

How to read the web application firewall log of a site, and what to do when a legitimate request is blocked — switch the site to detect, add a narrow rule exclusion or lower the sensitivity.

The web application firewall (WAF) of a site writes an entry every time a request looks suspicious. This page shows how to read the log and what to do when it blocks something that is legitimate, which is called a false alarm. How the WAF modes work is in the Security center.

  1. Open the account and choose its Security tab.
  2. Scroll to WAF log.

You can see the log if you may manage the account’s security: administrators, and resellers and customers for their own accounts. Entries are kept for 30 days. Administrators also see the latest blocks of every site under Security → WAF overview.

Each row is one request:

Column Meaning
Time When it happened.
Host The site name the request was sent to.
Client The address the request came from.
Request The method and the path, for example POST /contact.
Rules The number or numbers of the rules that matched.
Outcome Blocked (the visitor got a 403 answer) or Logged (detect mode, let through).
Message A short description of what looked suspicious.

To narrow the list, use Site, Outcome (All, Blocked, Logged only) and the From and To dates, which are in UTC. Choose Load more for older entries.

A visitor, or you, sees a 403 page on something that should work: a form, a login or an editor that cannot save. Typical sources are rich-text editors such as the WordPress block editor, forms that send code or markup, and tools that call the site’s API.

  1. Reproduce the problem and note the time, your address and the page.
  2. In the WAF log, set Outcome to Blocked and find the request: same host, path and client address. If nothing matches, the 403 did not come from the WAF. Check the site itself and the country and firewall rules of the server.
  3. Note the Rules value of the entry.
  4. Fix it with the smallest change that works, from the options below.

Best when one rule blocks one feature. In the account’s Security tab, pick the site and go to Rule exclusions:

  1. Enter the number from the Rules column in Rule ID.
  2. Enter a Path prefix, for example /contact/. A path starts with / and uses only letters, digits and / . _ ~ % -. Leave it empty to exclude the rule for the whole site.
  3. Choose Add exclusion.

The rule is then off for that path only; every other rule still applies. Repeat for each rule that appears in the entry. Remove deletes an exclusion you no longer need. A few rules, those that set the firewall up and decide whether a request is blocked, cannot be excluded: the panel refuses them, because excluding one would switch the WAF off.

Best while you do not yet know which rules are involved, or when the site is down for users. Set WAF mode to Detect (log only) and save. Nothing is blocked any more, and every suspicious request still lands in the log, so you can collect the rule numbers and then add exclusions. Switch back to Block when you are done. While the site is in detect mode, attacks go through unblocked: keep that time short.

If many different legitimate requests are blocked, lower the Paranoia level of the site (1 gives the fewest false alarms). This reduces protection for the whole site, so try the first two options before this one.

Repeat the action. The entry should now be Logged or absent, and the page should work. New entries appear in the log shortly after the request. Changes to the mode or the exclusions reach the server within a moment; the site page shows the mode that is Running now.