sshguard-setup - Man Page

For syslog-ng 2.x, add the following lines to syslog-ng.conf:

# pass only entries with auth+authpriv facilities from programs other than sshguard
filter sshlogs { facility(auth, authpriv) and not match("sshguard"); };
# pass to this process with this template (avoids <ID> prefixes)
destination sshguardproc {
program("/usr/local/sbin/sshguard"
template("$DATE $FULLHOST $MESSAGE\n"));
};
log { source(src); filter(sshlogs); destination(sshguardproc); };

For syslog-ng 3.x, add the following lines to syslog-ng.conf:

# enable 3.x mode
@version:3.0

# pass only entries with auth+authpriv facilities from programs other than sshguard
filter f_sshguard { facility(auth, authpriv) and not program("sshguard"); };
# pass entries built with this format
destination sshguard {
program("/usr/sbin/sshguard"
template("$DATE $FULLHOST $MSGHDR$MESSAGE\n")
);
};
log { source(src); filter(f_sshguard); destination(sshguard); };

After restarting syslog-ng, SSHGuard should start as soon as a log entry with facility auth or authpriv arrives. If you are monitoring services other than sshd, add the appropriate log facilities to syslog-ng.conf.

Add the following lines to metalog.conf:

Stuff to protect from brute force attacks :
    # for ssh
    facility = "*"
    program = "sshd"
    # other services ...
    # log to /var/log/sshguard directory
    logdir = "/var/log/sshguard"

After restarting metalog, log entries will appear in /var/log/sshguard.  Use log polling to monitor the current log.

SSHGuard adds attackers to table <sshguard>. Create the table and block attackers by adding the following lines to the end of pf.conf:

table <sshguard> persist
block in proto tcp from <sshguard>

After reloading the pf configuration, you can inspect the contents of the table using:

# pfctl -t sshguard -T show

SSHGuard creates and adds attackers to table 22. The table can be used to block attackers in your ruleset. For example:

# ipfw add 5000 reset ip from table\(22\) to me

You can inspect the contents of the table using:

# ipfw table 22 list

Blocked attackers are added to two ipsets named sshguard4 and sshguard6. The entries in the ipsets are blocked by default in the default firewall zone. Additional firewall zones can be configured using:

# firewall-cmd --zone=zone-name --permanent \
    --add-rich-rule="rule source ipset=sshguard4 drop"
# firewall-cmd --zone=zone-name --permanent \
    --add-rich-rule="rule source ipset=sshguard6 drop"

You can inspect the entries in the two ipsets using:

# firewall-cmd --permanent --info-ipset=sshguard4
# firewall-cmd --permanent --info-ipset=sshguard6

Blocked attackers are added to two ipsets named sshguard4 and sshguard6. Nothing is blocked by default, but can used as a source for iptables and other tools. E.g.:

# iptables  -I INPUT -m set --match-set sshguard4 src -j DROP
# ip6tables -I INPUT -m set --match-set sshguard6 src -j DROP

Create a chain for SSHGuard:

# iptables -N sshguard      # for IPv4
# ip6tables -N sshguard     # for IPv6

Update the INPUT chain to also pass the traffic to the sshguard chain at the very end of its processing. Specify in --dport all the ports of services your sshguard protects. If you want to prevent attackers from doing any traffic to the host, remove the option completely:

# block any traffic from abusers
iptables -A INPUT -j sshguard
ip6tables -A INPUT -j sshguard

Or:

# block abusers only for SSH, FTP, POP, IMAP services (use "multiport" module)
iptables -A INPUT -m multiport -p tcp --destination-ports 21,22,110,143 -j sshguard
ip6tables -A INPUT -m multiport -p tcp --destination-ports 21,22,110,143 -j sshguard

Verify that you have NOT a default allow rule passing all ssh traffic higher in the chain. Verify that you have NOT a default deny rule blocking all ssh traffic in your firewall. In either case, you already have the skill to adjust your firewall setup.

Here is a sample ruleset that makes sense:

iptables -N sshguard
# block whatever SSHGuard says be bad ...
iptables -A INPUT -j sshguard
# enable ssh, dns, http, https
iptables -A INPUT -p tcp --dport 22 -j ACCEPT
iptables -A INPUT -p udp --dport 53 -j ACCEPT
iptables -A INPUT -p tcp --dport 80 -j ACCEPT
iptables -A INPUT -p tcp --dport 443 -j ACCEPT
# and block everything else (default deny)
iptables -P INPUT DROP

When rebooting, most systems reset the firewall configuration by default. To preserve your configuration, you usually use the iptables-save and iptables-restore utilities. However, each Linux variant has its own "right way".

SSHGuard creates tables with a high priority and adds attackers to a set automatically.

You can inspect the contents of the sets using:

# nft list set ip sshguard attackers
# nft list set ip6 sshguard attackers

Moreover, you can display sshguard's tables with:

# nft list table ip sshguard
# nft list table ip6 sshguard

Check that the correct FILES (or LOGREADER) and BACKEND are set in sshguard.conf.

SSHGuard spawns a pipeline of cooperating processes. You can verify that SSHGuard is started correctly by viewing your process list using ps or top. Shown here are the processes associated with an example configuration:

/bin/sh /opt/sshguard/sbin/sshguard
tail -F -n 0 /tmp/log.txt
/opt/sshguard/libexec/sshg-parser
/opt/sshguard/libexec/sshg-blocker -a 30 -p 120 -s 1800 -N 128 -n 32
/bin/sh /usr/local/libexec/sshg-fw-null

In order:

• SSHGuard, installed in /opt/sshguard, is being interpreted by /bin/sh.
• SSHGuard launched tail -F -n 0, which is monitoring /tmp/log.txt.
• The default parser sshg-parser is running.
• The blocker is running with options -a 30 -p 120 -s 1800 -N 128 -n 32.
• The firewall sshg-fw-null is running. The null backend doesn't actually block any attacks.

SSHGuard recognizes attacks by parsing log messages. The format of log messages can occasionally change. If you are using the default, built-in attack parser, you can check if SSHGuard recognizes your attacks by running:

$ cat /var/log/auth.log | %PREFIX%/libexec/sshg-parser -a

Log messages that are recognized as attacks are prefixed with a '*' at the beginning of each line. If a log message that should be recognized as an attack is not, consider reporting it.