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.\" $DragonFly: src/share/man/man7/security.7,v 1.2 2003/06/17 04:37:00 dillon Exp $ 7.\" 8.Dd September 18, 1999 9.Dt SECURITY 7 10.Os 11.Sh NAME 12.Nm security 13.Nd introduction to security under FreeBSD 14.Sh DESCRIPTION 15Security is a function that begins and ends with the system administrator. 16While all 17.Bx 18multi-user systems have some inherent security, the job of building and 19maintaining additional security mechanisms to keep users 20.Sq honest 21is probably 22one of the single largest undertakings of the sysadmin. Machines are 23only as secure as you make them, and security concerns are ever competing 24with the human necessity for convenience. 25.Ux 26systems, 27in general, are capable of running a huge number of simultaneous processes 28and many of these processes operate as servers - meaning that external entities 29can connect and talk to them. As yesterday's mini-computers and mainframes 30become today's desktops, and as computers become networked and internetworked, 31security becomes an ever bigger issue. 32.Pp 33Security is best implemented through a layered onion approach. In a nutshell, 34what you want to do is to create as many layers of security as are convenient 35and then carefully monitor the system for intrusions. You do not want to 36overbuild your security or you will interfere with the detection side, and 37detection is one of the single most important aspects of any security 38mechanism. For example, it makes little sense to set the 39.Pa schg 40flags 41(see 42.Xr chflags 1 ) 43on every system binary because while this may temporarily protect the 44binaries, it prevents a hacker who has broken in from making an 45easily detectable change that may result in your security mechanisms not 46detecting the hacker at all. 47.Pp 48System security also pertains to dealing with various forms of attack, 49including attacks that attempt to crash or otherwise make a system unusable 50but do not attempt to break root. Security concerns can be split up into 51several categories: 52.Bl -enum -offset indent 53.It 54Denial of service attacks 55.It 56User account compromises 57.It 58Root compromise through accessible servers 59.It 60Root compromise via user accounts 61.It 62Backdoor creation 63.El 64.Pp 65A denial of service attack is an action that deprives the machine of needed 66resources. Typically, D.O.S. attacks are brute-force mechanisms that attempt 67to crash or otherwise make a machine unusable by overwhelming its servers or 68network stack. Some D.O.S. attacks try to take advantages of bugs in the 69networking stack to crash a machine with a single packet. The latter can 70only be fixed by applying a bug fix to the kernel. Attacks on servers can 71often be fixed by properly specifying options to limit the load the servers 72incur on the system under adverse conditions. Brute-force network 73attacks are harder to deal with. A spoofed-packet attack, for example, is 74nearly impossible to stop short of cutting your system off from the Internet. 75It may not be able to take your machine down, but it can fill up Internet 76pipe. 77.Pp 78A user account compromise is even more common then a D.O.S. attack. Many 79sysadmins still run standard telnetd, rlogind, rshd, and ftpd servers on their 80machines. These servers, by default, do not operate over encrypted 81connections. The result is that if you have any moderate-sized user base, 82one or more of your users logging into your system from a remote location 83(which is the most common and convenient way to login to a system) 84will 85have his or her password sniffed. The attentive system admin will analyze 86his remote access logs looking for suspicious source addresses 87even for successful logins. 88.Pp 89One must always assume that once an attacker has access to a user account, 90the attacker can break root. However, the reality is that in a well secured 91and maintained system, access to a user account does not necessarily give the 92attacker access to root. The distinction is important because without access 93to root the attacker cannot generally hide his tracks and may, at best, be 94able to do nothing more than mess with the user's files or crash the machine. 95User account compromises are very common because users tend not to take the 96precautions that sysadmins take. 97.Pp 98System administrators must keep in mind that there are potentially many ways 99to break root on a machine. The attacker may know the root password, 100the attacker 101may find a bug in a root-run server and be able to break root over a network 102connection to that server, or the attacker may know of a bug in an suid-root 103program that allows the attacker to break root once he has broken into a 104user's account. If an attacker has found a way to break root on a machine, 105the attacker may not have a need to install a backdoor. 106Many of the root holes found and closed to date involve a considerable amount 107of work by the hacker to cleanup after himself, so most hackers do install 108backdoors. This gives you a convenient way to detect the hacker. Making 109it impossible for a hacker to install a backdoor may actually be detrimental 110to your security because it will not close off the hole the hacker found to 111break in the first place. 112.Pp 113Security remedies should always be implemented with a multi-layered 114.Sq onion peel 115approach and can be categorized as follows: 116.Bl -enum -offset indent 117.It 118Securing root and staff accounts 119.It 120Securing root - root-run servers and suid/sgid binaries 121.It 122Securing user accounts 123.It 124Securing the password file 125.It 126Securing the kernel core, raw devices, and filesystems 127.It 128Quick detection of inappropriate changes made to the system 129.It 130Paranoia 131.El 132.Sh SECURING THE ROOT ACCOUNT AND SECURING STAFF ACCOUNTS 133Don't bother securing staff accounts if you haven't secured the root 134account. Most systems have a password assigned to the root account. The 135first thing you do is assume that the password is 136.Sq always 137compromised. This does not mean that you should remove the password. The 138password is almost always necessary for console access to the machine. 139What it does mean is that you should not make it possible to use the password 140outside of the console or possibly even with a 141.Xr su 1 142command. 143For example, make sure that your pty's are specified as being unsecure 144in the 145.Sq Pa /etc/ttys 146file 147so that direct root logins via telnet or rlogin are disallowed. If using 148other login services such as sshd, make sure that direct root logins are 149disabled there as well. Consider every access method - services such as 150ftp often fall through the cracks. Direct root logins should only be allowed 151via the system console. 152.Pp 153Of course, as a sysadmin you have to be able to get to root, so we open up 154a few holes. But we make sure these holes require additional password 155verification to operate. One way to make root accessible is to add appropriate 156staff accounts to the wheel group 157(in 158.Pa /etc/group ) . 159The staff members placed 160in the wheel group are allowed to 161.Sq su 162to root. You should never give staff 163members native wheel access by putting them in the wheel group in their 164password entry. Staff accounts should be placed in a 165.Sq staff 166group, and then added to the wheel group via the 167.Sq Pa /etc/group 168file. Only those staff members who actually need to have root access 169should be placed in the wheel group. It is also possible, when using an 170authentication method such as kerberos, to use kerberos's 171.Sq Pa .k5login 172file in the root account to allow a 173.Xr ksu 1 174to root without having to place anyone at all in the wheel group. This 175may be the better solution since the wheel mechanism still allows an 176intruder to break root if the intruder has gotten hold of your password 177file and can break into a staff account. While having the wheel mechanism 178is better then having nothing at all, it isn't necessarily the safest 179option. 180.Pp 181An indirect way to secure the root account is to secure your staff accounts 182by using an alternative login access method and *'ing out the crypted password 183for the staff accounts. This way an intruder may be able to steal the password 184file but will not be able to break into any staff accounts (or, indirectly, 185root, even if root has a crypted password associated with it). Staff members 186get into their staff accounts through a secure login mechanism such as 187.Xr kerberos 1 188or 189.Xr ssh 1 190using a private/public 191key pair. When you use something like kerberos you generally must secure 192the machines which run the kerberos servers and your desktop workstation. 193When you use a public/private key pair with ssh, you must generally secure 194the machine you are logging in FROM 195(typically your workstation), 196but you can 197also add an additional layer of protection to the key pair by password 198protecting the keypair when you create it with 199.Xr ssh-keygen 1 . 200Being able 201to *-out the passwords for staff accounts also guarantees that staff members 202can only login through secure access methods that you have setup. You can 203thus force all staff members to use secure, encrypted connections for 204all their sessions which closes an important hole used by many intruders: That 205of sniffing the network from an unrelated, less secure machine. 206.Pp 207The more indirect security mechanisms also assume that you are logging in 208from a more restrictive server to a less restrictive server. For example, 209if your main box is running all sorts of servers, your workstation shouldn't 210be running any. In order for your workstation to be reasonably secure 211you should run as few servers as possible, up to and including no servers 212at all, and you should run a password-protected screen blanker. 213Of course, given physical access to 214a workstation an attacker can break any sort of security you put on it. 215This is definitely a problem that you should consider but you should also 216consider the fact that the vast majority of break-ins occur remotely, over 217a network, from people who do not have physical access to your workstation or 218servers. 219.Pp 220Using something like kerberos also gives you the ability to disable or 221change the password for a staff account in one place and have it immediately 222effect all the machine the staff member may have an account on. If a staff 223member's account gets compromised, the ability to instantly change his 224password on all machines should not be underrated. With discrete passwords, 225changing a password on N machines can be a mess. You can also impose 226re-passwording restrictions with kerberos: not only can a kerberos ticket 227be made to timeout after a while, but the kerberos system can require that 228the user choose a new password after a certain period of time 229(say, once a month). 230.Sh SECURING ROOT - ROOT-RUN SERVERS AND SUID/SGID BINARIES 231The prudent sysadmin only runs the servers he needs to, no more, no less. Be 232aware that third party servers are often the most bug-prone. For example, 233running an old version of imapd or popper is like giving a universal root 234ticket out to the entire world. Never run a server that you have not checked 235out carefully. Many servers do not need to be run as root. For example, 236the ntalk, comsat, and finger daemons can be run in special user 237.Sq sandboxes . 238A sandbox isn't perfect unless you go to a large amount of trouble, but the 239onion approach to security still stands: If someone is able to break in 240through a server running in a sandbox, they still have to break out of the 241sandbox. The more layers the attacker must break through, the lower the 242likelihood of his success. Root holes have historically been found in 243virtually every server ever run as root, including basic system servers. 244If you are running a machine through which people only login via sshd and 245never login via telnetd or rshd or rlogind, then turn off those services! 246.Pp 247.Fx 248now defaults to running ntalkd, comsat, and finger in a sandbox. 249Another program which may be a candidate for running in a sandbox is 250.Xr named 8 . 251The default rc.conf includes the arguments necessary to run 252named in a sandbox in a commented-out form. Depending on whether you 253are installing a new system or upgrading an existing system, the special 254user accounts used by these sandboxes may not be installed. The prudent 255sysadmin would research and implement sandboxes for servers whenever possible. 256.Pp 257There are a number of other servers that typically do not run in sandboxes: 258sendmail, popper, imapd, ftpd, and others. There are alternatives to 259some of these, but installing them may require more work then you are willing 260to put 261(the convenience factor strikes again). 262You may have to run these 263servers as root and rely on other mechanisms to detect break-ins that might 264occur through them. 265.Pp 266The other big potential root hole in a system are the suid-root and sgid 267binaries installed on the system. Most of these binaries, such as rlogin, 268reside in 269.Pa /bin , 270.Pa /sbin , 271.Pa /usr/bin , 272or 273.Pa /usr/sbin . 274While nothing is 100% safe, 275the system-default suid and sgid binaries can be considered reasonably safe. 276Still, root holes are occasionally found in these binaries. A root hole 277was found in Xlib in 1998 that made xterm 278(which is typically suid) 279vulnerable. 280It is better to be safe then sorry and the prudent sysadmin will restrict suid 281binaries that only staff should run to a special group that only staff can 282access, and get rid of 283.Pq Li "chmod 000" 284any suid binaries that nobody uses. A 285server with no display generally does not need an xterm binary. Sgid binaries 286can be almost as dangerous. If an intruder can break an sgid-kmem binary the 287intruder might be able to read 288.Pa /dev/kmem 289and thus read the crypted password 290file, potentially compromising any passworded account. Alternatively an 291intruder who breaks group kmem can monitor keystrokes sent through pty's, 292including pty's used by users who login through secure methods. An intruder 293that breaks the tty group can write to almost any user's tty. If a user 294is running a terminal 295program or emulator with a keyboard-simulation feature, the intruder can 296potentially 297generate a data stream that causes the user's terminal to echo a command, which 298is then run as that user. 299.Sh SECURING USER ACCOUNTS 300User accounts are usually the most difficult to secure. While you can impose 301Draconian access restrictions on your staff and *-out their passwords, you 302may not be able to do so with any general user accounts you might have. If 303you do have sufficient control then you may win out and be able to secure the 304user accounts properly. If not, you simply have to be more vigilant in your 305monitoring of those accounts. Use of ssh and kerberos for user accounts is 306more problematic due to the extra administration and technical support 307required, but still a very good solution compared to a crypted password 308file. 309.Sh SECURING THE PASSWORD FILE 310The only sure fire way is to *-out as many passwords as you can and 311use ssh or kerberos for access to those accounts. Even though the 312crypted password file 313.Pq Pa /etc/spwd.db 314can only be read by root, it may 315be possible for an intruder to obtain read access to that file even if the 316attacker cannot obtain root-write access. 317.Pp 318Your security scripts should always check for and report changes to 319the password file 320(see 321.Sq Checking file integrity 322below). 323.Sh SECURING THE KERNEL CORE, RAW DEVICES, AND FILESYSTEMS 324If an attacker breaks root he can do just about anything, but there 325are certain conveniences. For example, most modern kernels have a 326packet sniffing device driver built in. Under 327.Fx 328it is called 329the 330.Sq bpf 331device. An intruder will commonly attempt to run a packet sniffer 332on a compromised machine. You do not need to give the intruder the 333capability and most systems should not have the bpf device compiled in. 334.Pp 335But even if you turn off the bpf device, 336you still have 337.Pa /dev/mem 338and 339.Pa /dev/kmem 340to worry about. For that matter, 341the intruder can still write to raw disk devices. 342Also, there is another kernel feature called the module loader, 343.Xr kldload 8 . 344An enterprising intruder can use a KLD module to install 345his own bpf device or other sniffing device on a running kernel. 346To avoid these problems you have to run 347the kernel at a higher secure level, at least securelevel 1. The securelevel 348can be set with a sysctl on the kern.securelevel variable. Once you have 349set the securelevel to 1, write access to raw devices will be denied and 350special chflags flags, such as 351.Sq schg , 352will be enforced. You must also ensure 353that the 354.Sq schg 355flag is set on critical startup binaries, directories, and 356script files - everything that gets run up to the point where the securelevel 357is set. This might be overdoing it, and upgrading the system is much more 358difficult when you operate at a higher secure level. You may compromise and 359run the system at a higher secure level but not set the schg flag for every 360system file and directory under the sun. Another possibility is to simply 361mount / and /usr read-only. It should be noted that being too draconian in 362what you attempt to protect may prevent the all-important detection of an 363intrusion. 364.Sh CHECKING FILE INTEGRITY: BINARIES, CONFIG FILES, ETC 365When it comes right down to it, you can only protect your core system 366configuration and control files so much before the convenience factor 367rears its ugly head. For example, using chflags to set the schg bit 368on most of the files in / and /usr is probably counterproductive because 369while it may protect the files, it also closes a detection window. The 370last layer of your security onion is perhaps the most important - detection. 371The rest of your security is pretty much useless (or, worse, presents you with 372a false sense of safety) if you cannot detect potential incursions. Half 373the job of the onion is to slow down the attacker rather then stop him 374in order to give the detection side of the equation a chance to catch him in 375the act. 376.Pp 377The best way to detect an incursion is to look for modified, missing, or 378unexpected files. The best 379way to look for modified files is from another (often centralized) 380limited-access system. 381Writing your security scripts on the extra-secure limited-access system 382makes them mostly invisible to potential hackers, and this is important. 383In order to take maximum advantage you generally have to give the 384limited-access box significant access to the other machines in the business, 385usually either by doing a read-only NFS export of the other machines to the 386limited-access box, or by setting up ssh keypairs to allow the limit-access 387box to ssh to the other machines. Except for its network traffic, NFS is 388the least visible method - allowing you to monitor the filesystems on each 389client box virtually undetected. If your 390limited-access server is connected to the client boxes through a switch, 391the NFS method is often the better choice. If your limited-access server 392is connected to the client boxes through a hub or through several layers 393of routing, the NFS method may be too insecure (network-wise) and using ssh 394may be the better choice even with the audit-trail tracks that ssh lays. 395.Pp 396Once you give a limit-access box at least read access to the client systems 397it is supposed to monitor, you must write scripts to do the actual 398monitoring. Given an NFS mount, you can write scripts out of simple system 399utilities such as 400.Xr find 1 401and 402.Xr md5 1 403It is best to physically md5 the client-box files boxes at least once a 404day, and to test control files such as those found in 405.Pa /etc 406and 407.Pa /usr/local/etc 408even more often. When mismatches are found relative to the base md5 409information the limited-access machine knows is valid, it should scream at 410a sysadmin to go check it out. A good security script will also check for 411inappropriate suid binaries and for new or deleted files on system partitions 412such as 413.Pa / 414and 415.Pa /usr 416.Pp 417When using ssh rather then NFS, writing the security script is much more 418difficult. You essentially have to 419.Pa scp 420the scripts to the client box in order to run them, making them visible, and 421for safety you also need to scp the binaries (such as find) that those scripts 422use. The ssh daemon on the client box may already be compromised. All in all, 423using ssh may be necessary when running over unsecure links, but it's also a 424lot harder to deal with. 425.Pp 426A good security script will also check for changes to user and staff members 427access configuration files: 428.Pa .rhosts , 429.Pa .shosts , 430.Pa .ssh/authorized_keys 431and so forth... files that might fall outside the purview of the MD5 check. 432.Pp 433If you have a huge amount of user disk space it may take too long to run 434through every file on those partitions. In this case, setting mount 435flags to disallow suid binaries and devices on those partitions is a good 436idea. The 437.Sq nodev 438and 439.Sq nosuid 440options 441(see 442.Xr mount 8 ) 443are what you want to look into. I would scan them anyway at least once a 444week, since the object of this layer is to detect a break-in whether or 445not the breakin is effective. 446.Pp 447Process accounting 448(see 449.Xr accton 8 ) 450is a relatively low-overhead feature of 451the operating system which I recommend using as a post-break-in evaluation 452mechanism. It is especially useful in tracking down how an intruder has 453actually broken into a system, assuming the file is still intact after 454the break-in occurs. 455.Pp 456Finally, security scripts should process the log files and the logs themselves 457should be generated in as secure a manner as possible - remote syslog can be 458very useful. An intruder tries to cover his tracks, and log files are critical 459to the sysadmin trying to track down the time and method of the initial 460break-in. One way to keep a permanent record of the log files is to run 461the system console to a serial port and collect the information on a 462continuing basis through a secure machine monitoring the consoles. 463.Sh PARANOIA 464A little paranoia never hurts. As a rule, a sysadmin can add any number 465of security features as long as they do not effect convenience, and 466can add security features that do effect convenience with some added 467thought. Even more importantly, a security administrator should mix it up 468a bit - if you use recommendations such as those given by this manual 469page verbatim, you give away your methodologies to the prospective 470hacker who also has access to this manual page. 471.Sh SPECIAL SECTION ON D.O.S. ATTACKS 472This section covers Denial of Service attacks. A DOS attack is typically 473a packet attack. While there isn't much you can do about modern spoofed 474packet attacks that saturate your network, you can generally limit the damage 475by ensuring that the attacks cannot take down your servers. 476.Bl -enum -offset indent 477.It 478Limiting server forks 479.It 480Limiting springboard attacks (ICMP response attacks, ping broadcast, etc...) 481.It 482Kernel Route Cache 483.El 484.Pp 485A common DOS attack is against a forking server that attempts to cause the 486server to eat processes, file descriptors, and memory until the machine 487dies. Inetd 488(see 489.Xr inetd 8 ) 490has several options to limit this sort of attack. 491It should be noted that while it is possible to prevent a machine from going 492down it is not generally possible to prevent a service from being disrupted 493by the attack. Read the inetd manual page carefully and pay specific attention 494to the 495.Fl c , 496.Fl C , 497and 498.Fl R 499options. Note that spoofed-IP attacks will circumvent 500the 501.Fl C 502option to inetd, so typically a combination of options must be used. 503Some standalone servers have self-fork-limitation parameters. 504.Pp 505Sendmail has its 506.Fl OMaxDaemonChildren 507option which tends to work much 508better than trying to use sendmail's load limiting options due to the 509load lag. You should specify a 510.Cm MaxDaemonChildren 511parameter when you start 512sendmail high enough to handle your expected load but no so high that the 513computer cannot handle that number of sendmails without falling on its face. 514It is also prudent to run sendmail in queued mode 515.Pq Fl ODeliveryMode=queued 516and to run the daemon 517.Pq Cm sendmail -bd 518separate from the queue-runs 519.Pq Cm sendmail -q15m . 520If you still want realtime delivery you can run the queue 521at a much lower interval, such as 522.Fl q1m , 523but be sure to specify a reasonable 524.Cm MaxDaemonChildren 525option for that sendmail to prevent cascade failures. 526.Pp 527Syslogd can be attacked directly and it is strongly recommended that you use 528the 529.Fl s 530option whenever possible, and the 531.Fl a 532option otherwise. 533.Pp 534You should also be fairly careful 535with connect-back services such as tcpwrapper's reverse-identd, which can 536be attacked directly. You generally do not want to use the reverse-ident 537feature of tcpwrappers for this reason. 538.Pp 539It is a very good idea to protect internal services from external access 540by firewalling them off at your border routers. The idea here is to prevent 541saturation attacks from outside your LAN, not so much to protect internal 542services from network-based root compromise. Always configure an exclusive 543firewall, i.e.\& 544.So 545firewall everything *except* ports A, B, C, D, and M-Z 546.Sc . 547This 548way you can firewall off all of your low ports except for certain specific 549services such as named 550(if you are primary for a zone), 551ntalkd, sendmail, 552and other internet-accessible services. 553If you try to configure the firewall the other 554way - as an inclusive or permissive firewall, there is a good chance that you 555will forget to 556.Sq close 557a couple of services or that you will add a new internal 558service and forget to update the firewall. You can still open up the 559high-numbered port range on the firewall to allow permissive-like operation 560without compromising your low ports. Also take note that 561.Fx 562allows you to 563control the range of port numbers used for dynamic binding via the various 564net.inet.ip.portrange sysctl's 565.Pq Li "sysctl -a | fgrep portrange" , 566which can also 567ease the complexity of your firewall's configuration. I usually use a normal 568first/last range of 4000 to 5000, and a hiport range of 49152 to 65535, then 569block everything under 4000 off in my firewall 570(except for certain specific 571internet-accessible ports, of course). 572.Pp 573Another common DOS attack is called a springboard attack - to attack a server 574in a manner that causes the server to generate responses which then overload 575the server, the local network, or some other machine. The most common attack 576of this nature is the ICMP PING BROADCAST attack. The attacker spoofs ping 577packets sent to your LAN's broadcast address with the source IP address set 578to the actual machine they wish to attack. If your border routers are not 579configured to stomp on ping's to broadcast addresses, your LAN winds up 580generating sufficient responses to the spoofed source address to saturate the 581victim, especially when the attacker uses the same trick on several dozen 582broadcast addresses over several dozen different networks at once. Broadcast 583attacks of over a hundred and twenty megabits have been measured. A second 584common springboard attack is against the ICMP error reporting system. By 585constructing packets that generate ICMP error responses, an attacker can 586saturate a server's incoming network and cause the server to saturate its 587outgoing network with ICMP responses. This type of attack can also crash the 588server by running it out of mbuf's, especially if the server cannot drain the 589ICMP responses it generates fast enough. The 590.Fx 591kernel has a new kernel 592compile option called ICMP_BANDLIM which limits the effectiveness of these 593sorts of attacks. The last major class of springboard attacks is related to 594certain internal inetd services such as the udp echo service. An attacker 595simply spoofs a UDP packet with the source address being server A's echo port, 596and the destination address being server B's echo port, where server A and B 597are both on your LAN. The two servers then bounce this one packet back and 598forth between each other. The attacker can overload both servers and their 599LANs simply by injecting a few packets in this manner. Similar problems 600exist with the internal chargen port. A competent sysadmin will turn off all 601of these inetd-internal test services. 602.Pp 603Spoofed packet attacks may also be used to overload the kernel route cache. 604Refer to the net.inet.ip.rtexpire, rtminexpire, and rtmaxcache sysctl 605parameters. A spoofed packet attack that uses a random source IP will cause 606the kernel to generate a temporary cached route in the route table, viewable 607with 608.Sq netstat -rna \&| fgrep W3 . 609These routes typically timeout in 1600 610seconds or so. If the kernel detects that the cached route table has gotten 611too big it will dynamically reduce the rtexpire but will never decrease it to 612less then rtminexpire. There are two problems: (1) The kernel does not react 613quickly enough when a lightly loaded server is suddenly attacked, and (2) The 614rtminexpire is not low enough for the kernel to survive a sustained attack. 615If your servers are connected to the internet via a T3 or better it may be 616prudent to manually override both rtexpire and rtminexpire via 617.Xr sysctl 8 . 618Never set either parameter to zero 619(unless you want to crash the machine :-)). 620Setting both parameters to 2 seconds should be sufficient to protect the route 621table from attack. 622.Sh ACCESS ISSUES WITH KERBEROS AND SSH 623There are a few issues with both kerberos and ssh that need to be addressed 624if you intend to use them. Kerberos V is an excellent authentication 625protocol but the kerberized telnet and rlogin suck rocks. There are bugs that 626make them unsuitable for dealing with binary streams. Also, by default 627kerberos does not encrypt a session unless you use the 628.Fl x 629option. Ssh encrypts everything by default. 630.Pp 631Ssh works quite well in every respect except when it is set up to 632forward encryption keys. 633What this means is that if you have a secure workstation holding 634keys that give you access to the rest of the system, and you ssh to an 635unsecure machine, your keys becomes exposed. The actual keys themselves are 636not exposed, but ssh installs a forwarding port for the duration of your 637login and if a hacker has broken root on the unsecure machine he can utilize 638that port to use your keys to gain access to any other machine that your 639keys unlock. 640.Pp 641We recommend that you use ssh in combination with kerberos whenever possible 642for staff logins. Ssh can be compiled with kerberos support. This reduces 643your reliance on potentially exposable ssh keys while at the same time 644protecting passwords via kerberos. Ssh keys 645should only be used for automated tasks from secure machines (something 646that kerberos is unsuited to). We also recommend that you either turn off 647key-forwarding in the ssh configuration, or that you make use of the 648.Pa "from=IP/DOMAIN" 649option that ssh allows in its 650.Pa authorized_keys 651file to make the key only usable to entities logging in from specific 652machines. 653.Sh SEE ALSO 654.Xr chflags 1 , 655.Xr find 1 , 656.Xr kerberos 1 , 657.Xr md5 1 , 658.Xr netstat 1 , 659.Xr openssl 1 , 660.Xr ssh 1 , 661.Xr xdm 1 , 662.Xr group 5 , 663.Xr ttys 5 , 664.Xr accton 8 , 665.Xr init 8 , 666.Xr sshd 8 , 667.Xr sysctl 8 , 668.Xr syslogd 8 , 669.Xr vipw 8 670.Sh HISTORY 671The 672.Nm 673manual page was originally written by 674.An Matthew Dillon 675and first appeared 676in 677.Fx 3.1 , 678December 1998. 679