no image caption

This howto does not cover installation of mod_security, it will only cover the finer points about rules and how to write your own custom ones 🙂
A ModSecurity rule does consist of the following.

Always consult the reference manual when writing rules, it covers everything you need. https://github.com/SpiderLabs/ModSecurity/wiki/Reference-Manual

Directive : Eg. SecRule for rules duh…
Scanfiler : REQUEST_URI or USERAGENT etc. full list here.
Scantarget : For example this can be googlebot when checking for useragent.
Action : What to do if filter fount a hit, Eg. Deny, Pass, Allow, log. It also has to contain a unique ID. full list here.

Some actions and scanfilters have prerequisites to work, and they are mostly documented in reference manual.
To start with you can add the following modsecurity rules in modsec config before including custom rules.

SecDataDir /tmp/sectmp
SecTmpDir /tmp
SecAuditLogDirMode 0777
SecPcreMatchLimit 1500
SecPcreMatchLimitRecursion 1500
SecCollectionTimeout 28800
SecRequestBodyAccess On
SecStreamInBodyInspection On
SecHashEngine On
SecRequestBodyInMemoryLimit 1310720

With this set you can forexample store connection ip adresses for up to 8 hours (28800 sec.) and to make a comparison you can say this is a form of session, which you can store session data upon.
With ModSecurity you can have a rule which store a variable on top of a ip, with this you can for example store how many times in the last 10 minutes someone has visited a page on you webpage.

Lets make an example like that.

First we need to start storing a “Session” for every visiting ip address.
This you can do with ACTION initcol like so.

SecAction phase:1,nolog,pass,initcol:ip=%{REMOTE_ADDR},id:1370

Then we can start storing information, lets say we will store how many times googlebot visit us total for up to 8 hours, when it will be reset.

SecRule REQUEST_HEADERS:User-agent "googlebot" "phase:1,nolog,id:1371,setvar:ip.counter=+1,pass"

We can then use this information to block googlebot requests if over 10 requests in last 8 hours.

SecRule ip:counter "@gt 10" "phase:1,log,msg:'googlebot has been blocked due to over 10 requests.',deny,id:1372"

You can also use AND in rules, in modsec this is done by using chains. Chained rules do not require ID.
Let say we will only block googlebot when it has over 10 requests and is visiting a specific file.

SecRule ip:counter "@gt 10" "phase:1,log,msg:'googlebot has been blocked due to over 10 requests.',deny,id:1372,chain"
SecRule REQUEST_URI "somefile.php" "t:none"

With this you should have grasped how modsecurity rules are written, and with the reference manual you can start doing magic 🙂