- Configuration and enabling of the clamd service.
- Updating the CLAMAV malware database and enabling the freshclam service
- Configuration and enabling of the clamav-milter service. Current versions of the clamav-milter require connectivity to the clamd daemon through either a local UNIX socker or a TCP/IP socket.
- Configuration of Postfix to utilize the available clamav-milter service.
Text 1: Typical settings overridden from defaults in /etc/clamd.conf
The clamd daemon typically reads its configuration from the /etc/clamd.conf file. Most importantly this file specifies, via the TCPSocket and TCPAddr directives, on what IP port and address the service listens for connections. These directives should be set to values appropriate for the host and which will be reachable by the clamav-milter. If the clamav-milter and the clamd daemon will be running on the same host the clamd service can be configured to listen to the localhost address [127.0.0.1] to avoid any potential network firewall and traffic filtering issues.
The clamd.conf file also provides many other tunable values but almost all of these should be appropriate at the distributions defaults.
Once configured the clamd service must be started and enabled for automatic start following the system's boot-up sequence; on RPM based systems this is typically achieved using the service and chkconfig commands.
Step #2 : Enabling the freshclam service
The freshclam service is an instance of the freshclam command line tool started with the “-d” option which runs the command in daemon mode. Whether started from the command-line or running in daemon mode freshclam will read its configuration from the /etc/freshclam.conf file. When running the freshclam daemon will periodically check the CLAMAV project mirrors for new malware signatures and update the local database used by the clamd scanning service. The freshclam daemon should run as the same user context as the clamd service; the typical way to ensure this is to synchronize the values of DatabaseOwner in /etc/freshclam.conf and User in /etc/clamd.conf. The frequency which freshclam will check for new patterns is controlled by the Checks directive – the default is 12 [times a day], this value should be sufficient in most cases. When database update succeeds the freshclam service will notify the clamd service that newer patterns are now available [for this to work the NotifyClamd directive must indicate the correct path to the current clamd configuration file].
Text 2: Example /etc/freshclam.conf file (comments removed)
The most important considerations in configuration of freshclam is if your network configuration requires use of an HTTP proxy server in order to access the CLAMAV mirrors for updates and if you need to configure some form of notification concerning success or failure of the pattern updates – a security focused service like a malware milter doesn't help anyone if it is silently failing in the background.
The HTTPProxyPort and HTTPProxyServer directives allow an HTTP proxy to be specified; freshclam will use this proxy for all mirror requests whether running as a command-line utility or in daemon mode. Should your proxy require a username/password for authentication these can be provided using the additional HTTPProxyUsername and HTTPProxyPassword directives. However it is simpler and more reliable to simply approve the “database.clamav.net” domain and sub-domains on your HTTP proxy service; all mirror requests will be made to those domains.
For notification of successful or failed updates the OnUpdateExecute and OnErrorExecute directives are used respectively. Whatever command is specified here will execute in the security context of the DatabaseOwner. A useful approach is to enable the log file via the UpdateLogFile directive and have the tail-end of that file mailed to a responsible party such as a help-desk or system-administrator for periodic verification that the service is operational.
#!/bin/shThe proper operation of freshclam can be tested by simply executing the freshclam utility on the command-line; it should check the mirrors and download any new patterns without an error message. Once configured and tested the freshclam service must be started and enabled for automatic start following the system's boot-up sequence.
tail -25 /var/log/freshclam.log \
| mail -s "[NOTICE] Malware Database Update Successful" \
-r email@example.com firstname.lastname@example.org
Text 3: A simple example script that might be used for OnUpdateExecute
Step #3 : Enabling the clamav-milter service
Text 4: Example clamav-milter.conf file (with comments removed)
Once the clamd scanning service is running and the freshclam service is maintaining the malware signatures the clamav-milter must be configured and started in order to connect the scanning service into Postfix's SMTP processing. The milter service is typically loads its configuration from the /etc/clamav-milter.conf file.
The service must be informed via the ClamdSocket directive where to find the clamd scanning service and via MilterSocket where to listen for connections from Postfix. The MitlerSocket directive is “inet:port@ip-address”. VirusAction and OnInfected directives can be used to control the behavior of the service when malware is identified; an OnInfected value of Quarantine will cause Postfix to hold the infected message in it's hold queue while a value of Reject will bounce the message with an SMTP error. Especially when used in Reject mode defining an appropriate VirusAction to notify the intended recipient of the message that a message has been discarded is important. The script named by VirusAction is executed in the security context of the scanning service and is provided seven parameters:
- Virus name-space
- Message queue id
- The sender's e-mail addres
- The e-mail address of the intended recipient.
- The subject of the message-id
- The message's Message-ID
- The date of the message.
#!/bin/bashConnecting the Postfix service to clamav-milter
# virus name, queue id, sender, destination, subject, message id, message date
echo " A message containing malware has been discarded.";
echo " Malware: $1";
echo " Sender: $3";
echo " Destination: $4";
echo " Subject: $5";
echo " Message-ID: $6";
echo " Date: $7";
echo " Queue-ID: $2";
) | \
mail -s '[ALERT] Infected Messages Discarded' \
-r email@example.com -c firstname.lastname@example.org $4
Text 5: A sample script for use as the VirusAction. This script notifies the intended recipient and help-desk that a message was identified as malware and discarded.
In order to integrate the scanning into Postfix the milter is configured in the main.cf file as an smtpd_milter. The default action of the milter should be set to “accept” so if for any reason the milter is unresponsive messages will still be delivered. As when connecting the other components it is important to verify that the Postfix service can reach the specified service [traffic is permitted by firewall's etc...].
smtpd_milters = inet:milter.example.com:32767
milter_default_action = accept
Text 6: Configuration directives from Postfix's main.cf
Upon modification of the main.cf file the Postfix service should be restarted.
Once configured the malware filtering service should be tested; this can be accomplished by acquiring a copy of the EICAR diagnostic virus and verifying that messages with this content attached are rejected and that the end-user's are notified of the rejection [according the clamav-milter's defined VirusAction].
clamd: instream(127.0.0.1@60469): Eicar-Test-Signature FOUND
Text 7: Example clamd log message for identified malware.
When malware is detected a message will be logged by clamd via syslog regarding the event; this will typically be logged under the “mail” service. Depending on the distribution messages logged as mail will be written to either /var/log/mail or /var/log/maillog [at least with the default syslog configuration].