xref: /dragonfly/share/man/man7/security.7 (revision dc71b7ab)
1.\" Copyright (c) 1998 Matthew Dillon.  Terms and conditions are those of
2.\" the BSD Copyright as specified in the file "/usr/src/COPYRIGHT" in
3.\" the source tree.
4.\"
5.\" $FreeBSD: src/share/man/man7/security.7,v 1.13.2.11 2002/04/13 02:04:44 keramida Exp $
6.\"
7.Dd September 18, 1999
8.Dt SECURITY 7
9.Os
10.Sh NAME
11.Nm security
12.Nd introduction to security under DragonFly
13.Sh DESCRIPTION
14Security is a function that begins and ends with the system administrator.
15While all
16.Bx
17multi-user systems have some inherent security, the job of building and
18maintaining additional security mechanisms to keep users
19.Sq honest
20is probably
21one of the single largest undertakings of the sysadmin.  Machines are
22only as secure as you make them, and security concerns are ever competing
23with the human necessity for convenience.
24.Ux
25systems,
26in general, are capable of running a huge number of simultaneous processes
27and many of these processes operate as servers - meaning that external entities
28can connect and talk to them.  As yesterday's mini-computers and mainframes
29become today's desktops, and as computers become networked and internetworked,
30security becomes an ever bigger issue.
31.Pp
32Security is best implemented through a layered onion approach.  In a nutshell,
33what you want to do is to create as many layers of security as are convenient
34and then carefully monitor the system for intrusions.  You do not want to
35overbuild your security or you will interfere with the detection side, and
36detection is one of the single most important aspects of any security
37mechanism.  For example, it makes little sense to set the
38.Pa schg
39flags
40(see
41.Xr chflags 1 )
42on every system binary because while this may temporarily protect the
43binaries, it prevents a hacker who has broken in from making an
44easily detectable change that may result in your security mechanisms not
45detecting the hacker at all.
46.Pp
47System security also pertains to dealing with various forms of attack,
48including attacks that attempt to crash or otherwise make a system unusable
49but do not attempt to break root.  Security concerns can be split up into
50several categories:
51.Bl -enum -offset indent
52.It
53Denial of service attacks
54.It
55User account compromises
56.It
57Root compromise through accessible servers
58.It
59Root compromise via user accounts
60.It
61Backdoor creation
62.El
63.Pp
64A denial of service attack is an action that deprives the machine of needed
65resources.  Typically, D.O.S. attacks are brute-force mechanisms that attempt
66to crash or otherwise make a machine unusable by overwhelming its servers or
67network stack.  Some D.O.S. attacks try to take advantages of bugs in the
68networking stack to crash a machine with a single packet.  The latter can
69only be fixed by applying a bug fix to the kernel.  Attacks on servers can
70often be fixed by properly specifying options to limit the load the servers
71incur on the system under adverse conditions.  Brute-force network
72attacks are harder to deal with.  A spoofed-packet attack, for example, is
73nearly impossible to stop short of cutting your system off from the Internet.
74It may not be able to take your machine down, but it can fill up Internet
75pipe.
76.Pp
77A user account compromise is even more common than a D.O.S. attack.  Many
78sysadmins still run standard telnetd, rlogind, rshd, and ftpd servers on their
79machines.  These servers, by default, do not operate over encrypted
80connections.  The result is that if you have any moderate-sized user base,
81one or more of your users logging into your system from a remote location
82(which is the most common and convenient way to login to a system)
83will
84have his or her password sniffed.  The attentive system admin will analyze
85his remote access logs looking for suspicious source addresses
86even for successful logins.
87.Pp
88One must always assume that once an attacker has access to a user account,
89the attacker can break root.  However, the reality is that in a well secured
90and maintained system, access to a user account does not necessarily give the
91attacker access to root.  The distinction is important because without access
92to root the attacker cannot generally hide his tracks and may, at best, be
93able to do nothing more than mess with the user's files or crash the machine.
94User account compromises are very common because users tend not to take the
95precautions that sysadmins take.
96.Pp
97System administrators must keep in mind that there are potentially many ways
98to break root on a machine.  The attacker may know the root password,
99the attacker
100may find a bug in a root-run server and be able to break root over a network
101connection to that server, or the attacker may know of a bug in an suid-root
102program that allows the attacker to break root once he has broken into a
103user's account.  If an attacker has found a way to break root on a machine,
104the attacker may not have a need to install a backdoor.
105Many of the root holes found and closed to date involve a considerable amount
106of work by the hacker to cleanup after himself, so most hackers do install
107backdoors.  This gives you a convenient way to detect the hacker.  Making
108it impossible for a hacker to install a backdoor may actually be detrimental
109to your security because it will not close off the hole the hacker found to
110break in the first place.
111.Pp
112Security remedies should always be implemented with a multi-layered
113.Sq onion peel
114approach and can be categorized as follows:
115.Bl -enum -offset indent
116.It
117Securing root and staff accounts
118.It
119Securing root - root-run servers and suid/sgid binaries
120.It
121Securing user accounts
122.It
123Securing the password file
124.It
125Securing the kernel core, raw devices, and filesystems
126.It
127Quick detection of inappropriate changes made to the system
128.It
129Paranoia
130.El
131.Sh SECURING THE ROOT ACCOUNT AND SECURING STAFF ACCOUNTS
132Don't bother securing staff accounts if you haven't secured the root
133account.  Most systems have a password assigned to the root account.  The
134first thing you do is assume that the password is
135.Sq always
136compromised.  This does not mean that you should remove the password.  The
137password is almost always necessary for console access to the machine.
138What it does mean is that you should not make it possible to use the password
139outside of the console or possibly even with a
140.Xr su 1
141command.
142For example, make sure that your pty's are specified as being unsecure
143in the
144.Sq Pa /etc/ttys
145file
146so that direct root logins via telnet or rlogin are disallowed.  If using
147other login services such as sshd, make sure that direct root logins are
148disabled there as well.  Consider every access method - services such as
149ftp often fall through the cracks.  Direct root logins should only be allowed
150via the system console.
151.Pp
152Of course, as a sysadmin you have to be able to get to root, so we open up
153a few holes.  But we make sure these holes require additional password
154verification to operate.  One way to make root accessible is to add appropriate
155staff accounts to the wheel group
156(in
157.Pa /etc/group ) .
158The staff members placed
159in the wheel group are allowed to
160.Sq su
161to root.  You should never give staff
162members native wheel access by putting them in the wheel group in their
163password entry.  Staff accounts should be placed in a
164.Sq staff
165group, and then added to the wheel group via the
166.Sq Pa /etc/group
167file.  Only those staff members who actually need to have root access
168should be placed in the wheel group.  It is also possible, when using an
169authentication method such as kerberos, to use kerberos's
170.Sq Pa .k5login
171file in the root account to allow a
172.Xr ksu 1
173to root without having to place anyone at all in the wheel group.  This
174may be the better solution since the wheel mechanism still allows an
175intruder to break root if the intruder has gotten hold of your password
176file and can break into a staff account.  While having the wheel mechanism
177is better than having nothing at all, it isn't necessarily the safest
178option.
179.Pp
180An indirect way to secure the root account is to secure your staff accounts
181by using an alternative login access method and *'ing out the crypted password
182for the staff accounts.  This way an intruder may be able to steal the password
183file but will not be able to break into any staff accounts (or, indirectly,
184root, even if root has a crypted password associated with it).  Staff members
185get into their staff accounts through a secure login mechanism such as
186.Xr kerberos 8
187or
188.Xr ssh 1
189using a private/public
190key pair.  When you use something like kerberos you generally must secure
191the machines which run the kerberos servers and your desktop workstation.
192When you use a public/private key pair with ssh, you must generally secure
193the machine you are logging in FROM
194(typically your workstation),
195but you can
196also add an additional layer of protection to the key pair by password
197protecting the keypair when you create it with
198.Xr ssh-keygen 1 .
199Being able
200to *-out the passwords for staff accounts also guarantees that staff members
201can only login through secure access methods that you have setup.  You can
202thus force all staff members to use secure, encrypted connections for
203all their sessions which closes an important hole used by many intruders:  That
204of sniffing the network from an unrelated, less secure machine.
205.Pp
206The more indirect security mechanisms also assume that you are logging in
207from a more restrictive server to a less restrictive server.  For example,
208if your main box is running all sorts of servers, your workstation shouldn't
209be running any.  In order for your workstation to be reasonably secure
210you should run as few servers as possible, up to and including no servers
211at all, and you should run a password-protected screen blanker.
212Of course, given physical access to
213a workstation an attacker can break any sort of security you put on it.
214This is definitely a problem that you should consider but you should also
215consider the fact that the vast majority of break-ins occur remotely, over
216a network, from people who do not have physical access to your workstation or
217servers.
218.Pp
219Using something like kerberos also gives you the ability to disable or
220change the password for a staff account in one place and have it immediately
221affect all the machines the staff member may have an account on.  If a staff
222member's account gets compromised, the ability to instantly change his
223password on all machines should not be underrated.  With discrete passwords,
224changing a password on N machines can be a mess.  You can also impose
225re-passwording restrictions with kerberos:  not only can a kerberos ticket
226be made to timeout after a while, but the kerberos system can require that
227the user choose a new password after a certain period of time
228(say, once a month).
229.Sh SECURING ROOT - ROOT-RUN SERVERS AND SUID/SGID BINARIES
230The prudent sysadmin only runs the servers he needs to, no more, no less.  Be
231aware that third party servers are often the most bug-prone.  For example,
232running an old version of imapd or popper is like giving a universal root
233ticket out to the entire world.  Never run a server that you have not checked
234out carefully.  Many servers do not need to be run as root.  For example,
235the ntalk, comsat, and finger daemons can be run in special user
236.Sq sandboxes .
237A sandbox isn't perfect unless you go to a large amount of trouble, but the
238onion approach to security still stands:  If someone is able to break in
239through a server running in a sandbox, they still have to break out of the
240sandbox.  The more layers the attacker must break through, the lower the
241likelihood of his success.  Root holes have historically been found in
242virtually every server ever run as root, including basic system servers.
243If you are running a machine through which people only login via sshd and
244never login via telnetd or rshd or rlogind, then turn off those services!
245.Pp
246.Dx
247now defaults to running ntalkd, comsat, and finger in a sandbox.
248Another program which may be a candidate for running in a sandbox is
249.Xr named 8 .
250The default rc.conf includes the arguments necessary to run
251named in a sandbox in a commented-out form.  Depending on whether you
252are installing a new system or upgrading an existing system, the special
253user accounts used by these sandboxes may not be installed.  The prudent
254sysadmin would research and implement sandboxes for servers whenever possible.
255.Pp
256There are a number of other servers that typically do not run in sandboxes:
257sendmail, popper, imapd, ftpd, and others.  There are alternatives to
258some of these, but installing them may require more work than you are willing
259to put
260(the convenience factor strikes again).
261You may have to run these
262servers as root and rely on other mechanisms to detect break-ins that might
263occur through them.
264.Pp
265The other big potential root hole in a system are the suid-root and sgid
266binaries installed on the system.  Most of these binaries, such as rlogin,
267reside in
268.Pa /bin ,
269.Pa /sbin ,
270.Pa /usr/bin ,
271or
272.Pa /usr/sbin .
273While nothing is 100% safe,
274the system-default suid and sgid binaries can be considered reasonably safe.
275Still, root holes are occasionally found in these binaries.  A root hole
276was found in Xlib in 1998 that made xterm
277(which is typically suid)
278vulnerable.
279It is better to be safe than sorry and the prudent sysadmin will restrict suid
280binaries that only staff should run to a special group that only staff can
281access, and get rid of
282.Pq Li "chmod 000"
283any suid binaries that nobody uses.  A
284server with no display generally does not need an xterm binary.  Sgid binaries
285can be almost as dangerous.  If an intruder can break an sgid-kmem binary the
286intruder might be able to read
287.Pa /dev/kmem
288and thus read the crypted password
289file, potentially compromising any passworded account.  Alternatively an
290intruder who breaks group kmem can monitor keystrokes sent through pty's,
291including pty's used by users who login through secure methods.  An intruder
292that breaks the tty group can write to almost any user's tty.  If a user
293is running a terminal
294program or emulator with a keyboard-simulation feature, the intruder can
295potentially
296generate a data stream that causes the user's terminal to echo a command, which
297is then run as that user.
298.Sh SECURING USER ACCOUNTS
299User accounts are usually the most difficult to secure.  While you can impose
300Draconian access restrictions on your staff and *-out their passwords, you
301may not be able to do so with any general user accounts you might have.  If
302you do have sufficient control then you may win out and be able to secure the
303user accounts properly.  If not, you simply have to be more vigilant in your
304monitoring of those accounts.  Use of ssh and kerberos for user accounts is
305more problematic due to the extra administration and technical support
306required, but still a very good solution compared to a crypted password
307file.
308.Sh SECURING THE PASSWORD FILE
309The only sure-fire way is to *-out as many passwords as you can and
310use ssh or kerberos for access to those accounts.  Even though the
311crypted password file
312.Pq Pa /etc/spwd.db
313can only be read by root, it may
314be possible for an intruder to obtain read access to that file even if the
315attacker cannot obtain root-write access.
316.Pp
317Your security scripts should always check for and report changes to
318the password file
319(see
320.Sq Checking file integrity
321below).
322.Sh SECURING THE KERNEL CORE, RAW DEVICES, AND FILESYSTEMS
323If an attacker breaks root he can do just about anything, but there
324are certain conveniences.  For example, most modern kernels have a
325packet sniffing device driver built in.  Under
326.Dx
327it is called
328the
329.Sq bpf
330device.  An intruder will commonly attempt to run a packet sniffer
331on a compromised machine.  You do not need to give the intruder the
332capability and most systems should not have the bpf device compiled in.
333.Pp
334But even if you turn off the bpf device,
335you still have
336.Pa /dev/mem
337and
338.Pa /dev/kmem
339to worry about.  For that matter,
340the intruder can still write to raw disk devices.
341Also, there is another kernel feature called the module loader,
342.Xr kldload 8 .
343An enterprising intruder can use a KLD module to install
344his own bpf device or other sniffing device on a running kernel.
345To avoid these problems you have to run
346the kernel at a higher secure level, at least securelevel 1.  The securelevel
347can be set with a sysctl on the
348.Va kern.securelevel
349variable.  Once you have
350set the securelevel to 1, write access to raw devices will be denied and
351special chflags flags, such as
352.Sq schg ,
353will be enforced.  You must also ensure
354that the
355.Sq schg
356flag is set on critical startup binaries, directories, and
357script files - everything that gets run up to the point where the securelevel
358is set.  This might be overdoing it, and upgrading the system is much more
359difficult when you operate at a higher secure level.  You may compromise and
360run the system at a higher secure level but not set the schg flag for every
361system file and directory under the sun.  Another possibility is to simply
362mount / and /usr read-only.  It should be noted that being too draconian in
363what you attempt to protect may prevent the all-important detection of an
364intrusion.
365.Sh CHECKING FILE INTEGRITY: BINARIES, CONFIG FILES, ETC
366When it comes right down to it, you can only protect your core system
367configuration and control files so much before the convenience factor
368rears its ugly head.  For example, using chflags to set the schg bit
369on most of the files in / and /usr is probably counterproductive because
370while it may protect the files, it also closes a detection window.  The
371last layer of your security onion is perhaps the most important - detection.
372The rest of your security is pretty much useless (or, worse, presents you with
373a false sense of safety) if you cannot detect potential incursions.  Half
374the job of the onion is to slow down the attacker rather than stop him
375in order to give the detection side of the equation a chance to catch him in
376the act.
377.Pp
378The best way to detect an incursion is to look for modified, missing, or
379unexpected files.  The best
380way to look for modified files is from another (often centralized)
381limited-access system.
382Writing your security scripts on the extra-secure limited-access system
383makes them mostly invisible to potential hackers, and this is important.
384In order to take maximum advantage you generally have to give the
385limited-access box significant access to the other machines in the business,
386usually either by doing a read-only NFS export of the other machines to the
387limited-access box, or by setting up ssh keypairs to allow the limit-access
388box to ssh to the other machines.  Except for its network traffic, NFS is
389the least visible method - allowing you to monitor the filesystems on each
390client box virtually undetected.  If your
391limited-access server is connected to the client boxes through a switch,
392the NFS method is often the better choice.  If your limited-access server
393is connected to the client boxes through a hub or through several layers
394of routing, the NFS method may be too insecure (network-wise) and using ssh
395may be the better choice even with the audit-trail tracks that ssh lays.
396.Pp
397Once you give a limit-access box at least read access to the client systems
398it is supposed to monitor, you must write scripts to do the actual
399monitoring.  Given an NFS mount, you can write scripts out of simple system
400utilities such as
401.Xr find 1
402and
403.Xr md5 1 .
404It is best to physically md5 the client-box files boxes at least once a
405day, and to test control files such as those found in
406.Pa /etc ,
407.Pa /usr/local/etc
408and
409.Pa /usr/pkg/etc
410even more often.  When mismatches are found relative to the base md5
411information the limited-access machine knows is valid, it should scream at
412a sysadmin to go check it out.  A good security script will also check for
413inappropriate suid binaries and for new or deleted files on system partitions
414such as
415.Pa /
416and
417.Pa /usr
418.Pp
419When using ssh rather than NFS, writing the security script is much more
420difficult.   You essentially have to
421.Pa scp
422the scripts to the client box in order to run them, making them visible, and
423for safety you also need to scp the binaries (such as find) that those scripts
424use.  The ssh daemon on the client box may already be compromised.  All in all,
425using ssh may be necessary when running over unsecure links, but it's also a
426lot harder to deal with.
427.Pp
428A good security script will also check for changes to user and staff members
429access configuration files:
430.Pa .rhosts ,
431.Pa .shosts ,
432.Pa .ssh/authorized_keys
433and so forth... files that might fall outside the purview of the MD5 check.
434.Pp
435If you have a huge amount of user disk space it may take too long to run
436through every file on those partitions.  In this case, setting mount
437flags to disallow suid binaries and devices on those partitions is a good
438idea.  The
439.Sq nodev
440and
441.Sq nosuid
442options
443(see
444.Xr mount 8 )
445are what you want to look into.  I would scan them anyway at least once a
446week, since the object of this layer is to detect a break-in whether or
447not the breakin is effective.
448.Pp
449Process accounting
450(see
451.Xr accton 8 )
452is a relatively low-overhead feature of
453the operating system which I recommend using as a post-break-in evaluation
454mechanism.  It is especially useful in tracking down how an intruder has
455actually broken into a system, assuming the file is still intact after
456the break-in occurs.
457.Pp
458Finally, security scripts should process the log files and the logs themselves
459should be generated in as secure a manner as possible - remote syslog can be
460very useful.  An intruder tries to cover his tracks, and log files are critical
461to the sysadmin trying to track down the time and method of the initial
462break-in.  One way to keep a permanent record of the log files is to run
463the system console to a serial port and collect the information on a
464continuing basis through a secure machine monitoring the consoles.
465.Sh PARANOIA
466A little paranoia never hurts.  As a rule, a sysadmin can add any number
467of security features as long as they do not affect convenience, and
468can add security features that do affect convenience with some added
469thought.  Even more importantly, a security administrator should mix it up
470a bit - if you use recommendations such as those given by this manual
471page verbatim, you give away your methodologies to the prospective
472hacker who also has access to this manual page.
473.Sh SPECIAL SECTION ON D.O.S. ATTACKS
474This section covers Denial of Service attacks.  A DOS attack is typically
475a packet attack.  While there isn't much you can do about modern spoofed
476packet attacks that saturate your network, you can generally limit the damage
477by ensuring that the attacks cannot take down your servers.
478.Bl -enum -offset indent
479.It
480Limiting server forks
481.It
482Limiting springboard attacks (ICMP response attacks, ping broadcast, etc...)
483.It
484Kernel Route Cache
485.El
486.Pp
487A common D.O.S. attack is against a forking server that attempts to cause the
488server to eat processes, file descriptors, and memory until the machine
489dies.  Inetd
490(see
491.Xr inetd 8 )
492has several options to limit this sort of attack.
493It should be noted that while it is possible to prevent a machine from going
494down it is not generally possible to prevent a service from being disrupted
495by the attack.  Read the inetd manual page carefully and pay specific attention
496to the
497.Fl c ,
498.Fl C ,
499and
500.Fl R
501options.  Note that spoofed-IP attacks will circumvent
502the
503.Fl C
504option to inetd, so typically a combination of options must be used.
505Some standalone servers have self-fork-limitation parameters.
506.Pp
507Sendmail has its
508.Fl OMaxDaemonChildren
509option which tends to work much
510better than trying to use sendmail's load limiting options due to the
511load lag.  You should specify a
512.Cm MaxDaemonChildren
513parameter when you start
514sendmail high enough to handle your expected load but no so high that the
515computer cannot handle that number of sendmails without falling on its face.
516It is also prudent to run sendmail in queued mode
517.Pq Fl ODeliveryMode=queued
518and to run the daemon
519.Pq Cm sendmail -bd
520separate from the queue-runs
521.Pq Cm sendmail -q15m .
522If you still want realtime delivery you can run the queue
523at a much lower interval, such as
524.Fl q1m ,
525but be sure to specify a reasonable
526.Cm MaxDaemonChildren
527option for that sendmail to prevent cascade failures.
528.Pp
529Syslogd can be attacked directly and it is strongly recommended that you use
530the
531.Fl s
532option whenever possible, and the
533.Fl a
534option otherwise.
535.Pp
536You should also be fairly careful
537with connect-back services such as tcpwrapper's reverse-identd, which can
538be attacked directly.  You generally do not want to use the reverse-ident
539feature of tcpwrappers for this reason.
540.Pp
541It is a very good idea to protect internal services from external access
542by firewalling them off at your border routers.  The idea here is to prevent
543saturation attacks from outside your LAN, not so much to protect internal
544services from network-based root compromise.  Always configure an exclusive
545firewall, i.e.\&
546.So
547firewall everything *except* ports A, B, C, D, and M-Z
548.Sc .
549This
550way you can firewall off all of your low ports except for certain specific
551services such as named
552(if you are primary for a zone),
553ntalkd, sendmail,
554and other internet-accessible services.
555If you try to configure the firewall the other
556way - as an inclusive or permissive firewall, there is a good chance that you
557will forget to
558.Sq close
559a couple of services or that you will add a new internal
560service and forget to update the firewall.  You can still open up the
561high-numbered port range on the firewall to allow permissive-like operation
562without compromising your low ports.  Also take note that
563.Dx
564allows you to
565control the range of port numbers used for dynamic binding via the various
566net.inet.ip.portrange sysctl's
567.Pq Li "sysctl -a | fgrep portrange" ,
568which can also
569ease the complexity of your firewall's configuration.  I usually use a normal
570first/last range of 4000 to 5000, and a hiport range of 49152 to 65535, then
571block everything under 4000 off in my firewall
572(except for certain specific
573internet-accessible ports, of course).
574.Pp
575Another common D.O.S. attack is called a springboard attack - to attack a server
576in a manner that causes the server to generate responses which then overload
577the server, the local network, or some other machine.  The most common attack
578of this nature is the ICMP PING BROADCAST attack.  The attacker spoofs ping
579packets sent to your LAN's broadcast address with the source IP address set
580to the actual machine they wish to attack.  If your border routers are not
581configured to stomp on ping's to broadcast addresses, your LAN winds up
582generating sufficient responses to the spoofed source address to saturate the
583victim, especially when the attacker uses the same trick on several dozen
584broadcast addresses over several dozen different networks at once.  Broadcast
585attacks of over a hundred and twenty megabits have been measured.  A second
586common springboard attack is against the ICMP error reporting system.  By
587constructing packets that generate ICMP error responses, an attacker can
588saturate a server's incoming network and cause the server to saturate its
589outgoing network with ICMP responses.  This type of attack can also crash the
590server by running it out of mbuf's, especially if the server cannot drain the
591ICMP responses it generates fast enough.  The
592.Dx
593kernel has a new kernel
594compile option called ICMP_BANDLIM which limits the effectiveness of these
595sorts of attacks.  The last major class of springboard attacks is related to
596certain internal inetd services such as the udp echo service.  An attacker
597simply spoofs a UDP packet with the source address being server A's echo port,
598and the destination address being server B's echo port, where server A and B
599are both on your LAN.  The two servers then bounce this one packet back and
600forth between each other.  The attacker can overload both servers and their
601LANs simply by injecting a few packets in this manner.  Similar problems
602exist with the internal chargen port.  A competent sysadmin will turn off all
603of these inetd-internal test services.
604.Pp
605Spoofed packet attacks may also be used to overload the kernel route cache.
606Refer to the
607.Va net.inet.ip.rtexpire ,
608.Va net.inet.ip.rtminexpire ,
609and
610.Va net.inet.ip.rtmaxcache
611sysctl
612parameters.  A spoofed packet attack that uses a random source IP will cause
613the kernel to generate a temporary cached route in the route table, viewable
614with
615.Sq netstat -rna \&| fgrep W3 .
616These routes typically timeout in 1600
617seconds or so.  If the kernel detects that the cached route table has gotten
618too big it will dynamically reduce the rtexpire but will never decrease it to
619less than rtminexpire.  There are two problems:  (1) The kernel does not react
620quickly enough when a lightly loaded server is suddenly attacked, and (2) The
621rtminexpire is not low enough for the kernel to survive a sustained attack.
622If your servers are connected to the internet via a T3 or better it may be
623prudent to manually override both rtexpire and rtminexpire via
624.Xr sysctl 8 .
625Never set either parameter to zero
626(unless you want to crash the machine :-)).
627Setting both parameters to 2 seconds should be sufficient to protect the route
628table from attack.
629.Sh ACCESS ISSUES WITH KERBEROS AND SSH
630There are a few issues with both kerberos and ssh that need to be addressed
631if you intend to use them.  Kerberos V is an excellent authentication
632protocol but the kerberized telnet and rlogin suck rocks.  There are bugs that
633make them unsuitable for dealing with binary streams.  Also, by default
634kerberos does not encrypt a session unless you use the
635.Fl x
636option.  Ssh encrypts everything by default.
637.Pp
638Ssh works quite well in every respect except when it is set up to
639forward encryption keys.
640What this means is that if you have a secure workstation holding
641keys that give you access to the rest of the system, and you ssh to an
642unsecure machine, your keys becomes exposed.  The actual keys themselves are
643not exposed, but ssh installs a forwarding port for the duration of your
644login and if a hacker has broken root on the unsecure machine he can utilize
645that port to use your keys to gain access to any other machine that your
646keys unlock.
647.Pp
648We recommend that you use ssh in combination with kerberos whenever possible
649for staff logins.  Ssh can be compiled with kerberos support.  This reduces
650your reliance on potentially exposable ssh keys while at the same time
651protecting passwords via kerberos.  Ssh keys
652should only be used for automated tasks from secure machines (something
653that kerberos is unsuited to).  We also recommend that you either turn off
654key-forwarding in the ssh configuration, or that you make use of the
655.Pa "from=IP/DOMAIN"
656option that ssh allows in its
657.Pa authorized_keys
658file to make the key only usable to entities logging in from specific
659machines.
660.Sh SEE ALSO
661.Xr chflags 1 ,
662.Xr find 1 ,
663.Xr md5 1 ,
664.Xr netstat 1 ,
665.Xr openssl 1 ,
666.Xr ssh 1 ,
667.Xr xdm 1 Pq Pa x11/xdm ,
668.Xr group 5 ,
669.Xr ttys 5 ,
670.Xr firewall 7 ,
671.Xr accton 8 ,
672.Xr init 8 ,
673.Xr kerberos 8 ,
674.Xr sshd 8 ,
675.Xr sysctl 8 ,
676.Xr syslogd 8 ,
677.Xr vipw 8
678.Sh HISTORY
679The
680.Nm
681manual page was originally written by
682.An Matthew Dillon
683and first appeared
684in
685.Fx 3.1 ,
686December 1998.
687