mirror of https://github.com/fail2ban/fail2ban
Created How to ban something other as host (IP-address), like user or e-mail, etc. (markdown)
parent
ea487f6393
commit
a03e79f184
|
@ -0,0 +1,77 @@
|
||||||
|
**[Q]** Can I ban something other as host (IP-address), like user or e-mail, etc.<br/>
|
||||||
|
**[A]** Yes, it is theoretically possible with fail2ban, since no-host banning was implemented (v. 0.9.5 or 0.10). See [fail2ban/gh-1454](https://github.com/fail2ban/fail2ban/pull/1454) for more implementation details.
|
||||||
|
|
||||||
|
* You should create your own `action` and specify there which command(s) should be executed by ban/unban
|
||||||
|
* Version 0.10 allows you to define failure-ID in `failregex`:
|
||||||
|
- use `<F-ID/>` for failure-ID as no-space tag (equivalent to `(?P<fid>\S+)`), example:
|
||||||
|
```
|
||||||
|
failregex = ^authentication failure; login=<F-ID/>
|
||||||
|
```
|
||||||
|
- use `<F-ID>...</F-ID>` for own regex contains failure-ID (equivalent to `(?P<fid>...)`), example:
|
||||||
|
```
|
||||||
|
failregex = ^authentication failure; login=<F-ID>[^@]+@\S+</F-ID/>
|
||||||
|
```
|
||||||
|
* In version 0.9 you should use `(?P<host>...)` to define failure-ID and implicitly reset all host-related features (e. g. dns resolving) for this jail, so define `usedns = raw`, `ignoreip =`, `ignorecommand =`
|
||||||
|
|
||||||
|
Example for test jail to ban users, config `jail.local`:
|
||||||
|
```bash
|
||||||
|
[test]
|
||||||
|
# don't use dns, because host group is not hostname (and not resolvable ip):
|
||||||
|
usedns = raw
|
||||||
|
ignoreip =
|
||||||
|
ignorecommand =
|
||||||
|
# if used some filter:
|
||||||
|
#filter = no-host-filter[HOST=(?P<host>\S+)]
|
||||||
|
#
|
||||||
|
# if used failregex:
|
||||||
|
filter =
|
||||||
|
# v. 0.10:
|
||||||
|
failregex = ^\s*(?:\S+\s+)?(?:[^:]+:auth\[\d+\]:\s+)?pam_unix(?:\(\S+\))?:?\s+authentication failure; login=<F-ID/>
|
||||||
|
# action that bans users
|
||||||
|
banaction = test-ban-user[name=%(__name__)s]
|
||||||
|
#
|
||||||
|
logpath = %(syslog_authpriv)s
|
||||||
|
enabled = true
|
||||||
|
```
|
||||||
|
For fail2ban version 0.9, you should define `failregex` like below:
|
||||||
|
```bash
|
||||||
|
[test]
|
||||||
|
...
|
||||||
|
# v. 0.9.5:
|
||||||
|
failregex = ^\s*(?:\S+\s+)?(?:[^:]+:auth\[\d+\]:\s+)?pam_unix(?:\(\S+\))?:?\s+authentication failure; login=(?P<host>\S+)
|
||||||
|
```
|
||||||
|
|
||||||
|
Action config file `action.d/test-ban-user.local`:
|
||||||
|
``` bash
|
||||||
|
[Definition]
|
||||||
|
actionstart =
|
||||||
|
actionstop =
|
||||||
|
actioncheck =
|
||||||
|
actionban = echo 'ban f2b-<name> --user <ip>'
|
||||||
|
actionunban = echo 'unban f2b-<name> --user <ip>'
|
||||||
|
```
|
||||||
|
|
||||||
|
To test, the user "xxx" will be banned, just execute following commands (3 times if `maxretry = 3` for this jail):
|
||||||
|
``` bash
|
||||||
|
logger -t 'test:auth' -i -p auth.info "pam_unix(test:auth): authentication failure; login=xxx"
|
||||||
|
logger -t 'test:auth' -i -p auth.info "pam_unix(test:auth): authentication failure; login=xxx"
|
||||||
|
logger -t 'test:auth' -i -p auth.info "pam_unix(test:auth): authentication failure; login=xxx"
|
||||||
|
```
|
||||||
|
|
||||||
|
To test regular expression, use new option `--raw` or `-r`, to prevent dns resolving errors:
|
||||||
|
|
||||||
|
``` bash
|
||||||
|
# v. 0.10:
|
||||||
|
fail2ban-regex --raw /var/log/auth.log '^\s*(?:\S+\s+)?(?:[^:]+:auth\[\d+\]:\s+)?pam_unix(?:\(\S+\))?:?\s+authentication failure; login=<F-ID/>'
|
||||||
|
# v. 0.9.5:
|
||||||
|
fail2ban-regex --raw /var/log/auth.log '^\s*(?:\S+\s+)?(?:[^:]+:auth\[\d+\]:\s+)?pam_unix(?:\(\S+\))?:?\s+authentication failure; login=(?P<host>\S+)'
|
||||||
|
```
|
||||||
|
|
||||||
|
**[Q]** I don't have any failure-ID in the log-entry, can I nevertheless configure the fail2ban, that it should simply execute some command if some message will be found in observed log-file<br/>
|
||||||
|
**[A]** Yes, if you've no failure-id at all (no user-id, e-mail or something other), but you'll that fail2ban execute some shell script after failure occurrence, you should additionally:
|
||||||
|
* set empty or something other as match for failure-id (still `<host>` in 0.9th-branch) in `failregex`;
|
||||||
|
* set `maxretry = 1` and `findtime = 1` (ban after first occurrence in 1 seconds);
|
||||||
|
* set small `bantime` (e. g. 1 second) to this "jail" (otherwise no "ban" action will be executed in this time, because "already banned" occurs), e. g. `bantime = 1`
|
||||||
|
* you need to specify only `actionban` parameter in your custom action file (`actionban = /user/bin/command.sh`);
|
||||||
|
* `actionban` script will be executed as root (or with user, fail2ban running), so use `su` if other/restricted user needed;
|
||||||
|
- set `usedns`, `ignoreip`, `ignorecommand` as suggested above, otherwise you can get error by comparison with empty/illegal host (that will be found by "failure");
|
Loading…
Reference in New Issue