[보안] UNIX IP Stack Tuning Guide v2.7

UNIX IP Stack Tuning Guide v2.7  By Rob Thomas, robt at cymru.com, 03 DEC 2000

Introduction

The purpose of this document is to strengthen the UNIX IP stack against

a variety

of attack types prevalent on the Internet today.  This document details

the

settings recommended for UNIX servers designed to provide network

intensive

services such as HTTP or routing (firewall services).  This document

covers

the following UNIX variants:

        A. IBM AIX 4.3.X

        B. Sun Solaris 7

        C. Compaq Tru64 UNIX 5.X

        D. HP HP-UX 11.0 (research ongoing)

        E. Linux kernel 2.2 (tested both SuSE Linux 7.0 and RedHat

7.0)

        F. FreeBSD

        G. IRIX 6.5.10

Windows NT/2000 will be covered in future releases.

The tuning changes noted below will not remain across reboots.

Thus, these changes should be added to a runtime commands file so

that the changes are enacted at each boot.

        AIX – /etc/rc.net

        Solaris – /etc/init.d/inetinit

        Tru64 UNIX – Use the sysconfigdb or

dxkerneltuner command

        HP-UX – /etc/rc.config.d/nddconf

        Linux kernel 2.2 – /etc/sysctl.conf

        FreeBSD – /etc/rc.conf

        IRIX – Use the systune command

No tuning steps should be undertaken without a keen insight into

the purpose and effect of such steps.  It is assumed that the

personnel who undertake such tuning are well versed in the UNIX

kernel with bit-level understanding of TCP/IP and the RFCs that

govern IP behavior.

Revision History

Version 1.0 – Created the AIX IP stack tuning guide

Version 2.0 – Added Solaris and additional AIX tuning tips, reformated

doc

Version 2.2 – Added ARP tuning tips

Version 2.3 – Added Compaq Tru64 UNIX SYN defense tips

Version 2.4 – Added HP HP-UX 11.0 tuning tips

Version 2.5 – Added Linux and FreeBSD tuning tips

Version 2.6 – Added IRIX 6.5.10 and additional FreeBSD tuning tips

Version 2.7 – Added additional HP-UX 11.0 tuning tips

Credits

This has become a global effort, and this document is made more rich and

valuable thanks to the input of several very helpful folks!  Kudos and

lauds for the folks below:

Thanks to the Compaq Software Security Response Team for the Tru64 UNIX

SYN flood defense pointers.

For HP-UX 11.0, thanks to Tom Harrold for running ndd over and over at my

every request. 🙂

Thanks to Klaus Moeller for a copious collection of Linux tuning tips.

Thanks to Andrew Korty for the FreeBSD tuning tips.

Thanks to Christian Lambert and Alex Icasiano of SGI for the IRIX tuning

tips.

Thanks, as always, to the FIRST

community for support, questions, and feedback.

General IP Stack Tuning Recommendations

1. TCP send and receive spaces

The TCP send and receive spaces directly effect the TCP window

size parameter.  An increased window size will allow for more

efficient transfers, particularly bulk transfers such as FTP

and HTTP.  The default for each is not optimal, and should be

increased to 32768 bytes.  This value should not be increased

above 64K bytes unless the implications of

RFC1323

and RFC2018

are fully understood and support for both is enabled.

Do not enable RFC1323 without also enabling support for RFC2018.

Remember, pipe drain is a Bad Thing[tm].

        A. AIX

        /usr/sbin/no -o tcp_sendspace=32768

        /usr/sbin/no -o tcp_recvspace=32768

        B. Solaris

        /usr/sbin/ndd -set /dev/tcp tcp_xmit_hiwat 32768

        /usr/sbin/ndd -set /dev/tcp tcp_recv_hiwat 32768

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        The TCP send and receive spaces are set to 32768 by default

        E. Linux kernel 2.2

        While Linux automagically assigns the TCP send and receive

spaces,

                support for both RFC1323 (large window support,

                net.ipv4.tcp_window_scaling) and RFC2018 (SACK

support,

                net.ipv4.tcp_sack) are enabled by default.

        F. FreeBSD

        sysctl -w net.inet.tcp.sendspace=32768

        sysctl -w net.inet.tcp.recvspace=32768

        G. IRIX

        The default settings for IRIX are 64Kbytes for both the

TCP send

                and receive spaces.

2. Socket queue defense against SYN attacks

While great effort is undertaken to defend any network from those

with malicious intent, several ports (largely TCP) must remain

open to conduct business.  Internet vandals may attempt to

exploit these ports to launch a denial of service attack.  One of

the most popular attacks remains the SYN flood, wherein the socket

queue of the attacked host is overwhelmed with bogus connection

requests.  To defend against such attacks, certain UNIX variants

maintain separate queues for inbound socket connection requests.

One queue is for half-open sockets (SYN received, SYN|ACK sent),

the other queue for fully-open sockets awaiting an accept()

call from the application.  These two queues should be increased so

that an attack of low to moderate intensity will have little to no

effect on the stability or availability of the server.

        A. AIX

        /usr/sbin/no -o clean_partial_conns=1

        This setting will instruct the kernel to randomly remove

                half-open sockets from the q0 queue to make room for

new

                sockets.

        B. Solaris

        /usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q 1024

        The q queue holds sockets awaiting an accept()

                call from the application.

        

        

        /usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q0 2048

        The q0 queue contains half-open sockets.

        C. Tru64 UNIX

        /sbin/sysconfig -r socket sominconn=65535

        The value of sominconn determines how many simultaneous

incoming

                SYN packets can be handled by the system.

        

        

        /sbin/sysconfig -r socket somaxconn=65535

        The value of somaxconn sets the maximum number of pending

TCP

                connections.

        D. HP-UX

        /usr/sbin/ndd -set tcp_syn_rcvd_max 1024

        /usr/sbin/ndd -set tcp_conn_request_max 200

        E. Linux kernel 2.2

        /sbin/sysctl -w net.ipv4.tcp_max_syn_backlog=1280

        Increases the size of the socket queue (effectively, q0).

        /sbin/sysctl -w net.ipv4.tcp_syn_cookies=1

        Enables support for TCP SYN cookies, which mitigates the

effectiveness

                of SYN floods.  However, this may cause performance

problems for

                large windows (see RFC1323 and RFC2018).  To read

more about SYN

                cookies, please review DJ Bernstein’s paper

                here.

        F. FreeBSD

        sysctl -w kern.ipc.somaxconn=1024

        G. IRIX

        The listen() queue is hardcoded to 32.  However,

the system

                actually enforces the limit of pending connections as

                ((3 * backlog) / 2) + 1.  This yields a maximum

backlog of 49

                connections.

3. Redirects

A miscreant can use IP redirects to modify the routing table on a

remote host.  In a well-designed network, redirects to the end

stations should not be required.  Both the sending and accepting

of redirects should be disabled.

        A. AIX

        /usr/sbin/no -o ipignoreredirects=1

        /usr/sbin/no -o ipsendredirects=0

        B. Solaris

        /usr/sbin/ndd -set /dev/ip ip_ignore_redirect 1

        /usr/sbin/ndd -set /dev/ip ip_send_redirects 0

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        /usr/sbin/ndd -set /dev/ip ip_send_redirects 0

        E. Linux kernel 2.2

        /sbin/sysctl -w net.ipv4.conf.all.send_redirects=0

        /sbin/sysctl -w net.ipv4.conf.all.accept_redirects=0

        F. FreeBSD

        sysctl -w net.inet.icmp.drop_redirect=1

        sysctl -w net.inet.icmp.log_redirect=1

        sysctl -w net.inet.ip.redirect=0

        sysctl -w net.inet6.ip6.redirect=0

        G. IRIX

        /usr/sbin/systune icmp_dropredirects to 1

4. ARP cleanup

It is possible for a miscreant to create a resource exhaustion or

performance degredation by filling the IP route cache with bogus

ARP entries.  In Solaris, there are two parameters that govern the

cleanup interval for the IP route cache.  For unsolicited ARP

responses, the parameter to be tuned is arp_cleanup_interval.  In

AIX, the cleanup interval is governed by the value of arpt_killc.

However, this parameter governs both solicited and unsolicited ARP

entries.  For this reason, it is likely best to leave the parameter

at the default setting of 20 minutes.

        A. AIX

        /usr/sbin/no -o arpt_killc=20

        B. Solaris

        /usr/sbin/ndd -set /dev/arp arp_cleanup_interval

60000

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        By default set to five minutes.

        E. Linux kernel 2.2

        No tuning recommendations.

        F. FreeBSD

        sysctl -w net.link.ether.inet.max_age=1200

        G. IRIX

        No tuning recommendations.

5. Source routing

With source routing, an attacker can attempt to reach internal IP

addresses – including RFC1918 addresses.  It is important to disable

the acceptance of source routed packets to prevent subtle probes of

your internal networks.

        A. AIX

        /usr/sbin/no -o ipsrcroutesend=0

        Disable the sending of source routed packets.

        /usr/sbin/no -o ipsrcrouteforward=0

        This is important if the box is routing, e.g. a firewall.

                Disable this feature to prevent the host from

forwarding source

                routed packets.        

        B. Solaris

        /usr/sbin/ndd -set /dev/ip ip_src_route_forward 0

        This is important if the box is routing, e.g. a firewall.  

                Disable this feature to prevent the host from

forwarding source

                routed packets.

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        ndd -set /dev/ip ip_forward_src_routed 0

        Disable this feature to prevent the host from forwarding

source

                routed packets.

        E. Linux kernel 2.2

        /sbin/sysctl -w

net.ipv4.conf.all.accept_source_route=0

        Drop all source route packets.

        /sbin/sysctl -w net.ipv4.conf.all.forwarding=0

        /sbin/sysctl -w net.ipv4.conf.all.mc_forwarding=0

        Do not forward source routed frames.

        F. FreeBSD

        sysctl -w net.inet.ip.sourceroute=0

        sysctl -w net.inet.ip.accept_sourceroute=0

        G. IRIX

        /usr/sbin/systune ipforward to 2

6. TIME_WAIT setting

On a busy web server, many sockets may linger in the TIME_WAIT state.

This is caused by improperly coded client applications that do not

properly shut down a socket.  This can also be used as a type of DDoS

attack.

        A. AIX

        No tuning recommendations.

        B. Solaris

        /usr/sbin/ndd -set /dev/tcp tcp_time_wait_interval

60000

        This parameter effects the amount of time a TCP socket will

                remain in the TIME_WAIT state.  The default is quite

high

                for a busy web server, so it should be lowered to

60000

                milliseconds (60 seconds).  The parameter name was

corrected

                in Solaris 7 and higher.  Prior to Solaris 7, the

parameter

                was incorrectly labeled as tcp_close_wait_interval.

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        ndd -set /dev/tcp tcp_time_wait_interval 60000

        Sockets will linger in TIME_WAIT state no more than 60

seconds.

        E. Linux kernel 2.2

        /sbin/sysctl -w net.ipv4.vs.timeout_timewait=60

        Sockets will linger in TIME_WAIT state for no more than 60

seconds.

        F. FreeBSD

        No tuning recommendations.

        G. IRIX

        /usr/sbin/systune tcp_2msl to 60

7. Broadcast ECHO response

Smurf attacks work by sending ICMP 8 0 (ECHO REQUEST) messages to a

broadcast address from a spoofed address.  Some IP stacks will respond,

by default, to such messages.  This should be disabled.  Further, if the

host is a firewall (router), it should not propogate directed broadcasts.

        A. AIX

        /usr/sbin/no -o directed_broadcast=0

        Do not respond to directed broadcasts.

        B. Solaris

        /usr/sbin/ndd -set /dev/ip ip_respond_to_echo_broadcast

0

        Do not respond to directed broadcasts.

        /usr/sbin/ndd -set /dev/ip

ip_forward_directed_broadcasts 0

        Do not forward directed broadcasts.

        C. Tru64 UNIX

        No tuning recommendations.

        D. HP-UX

        ndd -set /dev/ip ip_respond_to_echo_broadcast 0

        Do not respond to directed broadcasts.

        ndd -set /dev/ip ip_forward_directed_broadcasts 0

        Do not forward directed broadcasts.

        E. Linux kernel 2.2

        /sbin/sysctl -w

net.ipv4.icmp_echo_ignore_broadcasts=1

        Do not respond to directed broadcasts.

        F. FreeBSD

        sysctl -w net.inet.icmp.bmcastecho=0

        G. IRIX

        /usr/sbin/systune allow_brdaddr_srcaddr to 0

8. Other broadcast probes

There are two other broadcast probes that a miscreant could utilize

against a network.  The address mask query can be used to map out the

size of the netblock, and set a range for further probes.  The

timestamp broadcast is another means of mapping and fingerprinting

hosts.

        A. AIX

        /usr/sbin/no -o icmpaddressmask=0

        Prevent address mask queries.

        B. Solaris

        /usr/sbin/ndd -set /dev/ip

ip_respond_to_address_mask_broadcast 0

        Prevent address mask queries.

        /usr/sbin/ndd -set /dev/ip

ip_respond_to_timestamp_broadcast 0

        Disable timestamp broadcast query response.

        C. Tru64 UNIX

        No tuning recomendations.

        D. HP-UX

        ndd -set /dev/ip ip_respond_to_address_mask_broadcast

0

        Prevent the host from revealing the configured netmask.

        ndd -set /dev/ip ip_respond_to_timestamp_broadcast 0

    Disable timestamp broadcast query response.

        E. Linux kernel 2.2

        No tuning recommendations.

        F. FreeBSD

        sysctl -w net.inet.icmp.maskrepl=0

        G. IRIX

        Use ipfilterd to block unwanted ICMP types.

9. Support for RFC1948

This will utilize RFC1948

sequence number generation techniques to ensure that the sequence

number for a given TCP socket is very difficult to guess.  This

tactic prevents TCP session hijacking, which could be quite an issue

with e-commerce.

        B. Solaris

        Set TCP_STRONG_ISS=2 in /etc/default/inetinit.

        This will require a reboot to take effect.

        G. IRIX

        /usr/sbin/systune tcpiss_md5 to 1