OpenBSD manual page server

Manual Page Search Parameters

ISAKMPD(8) System Manager's Manual ISAKMPD(8)

isakmpdISAKMP/Oakley a.k.a. IKEv1 key management daemon

isakmpd [-46adKLnSTv] [-c config-file] [-D class=level] [-f fifo] [-i pid-file] [-l packetlog-file] [-N udpencap-port] [-p listen-port] [-R report-file]

The isakmpd daemon establishes Security Associations (SAs) for encrypted and/or authenticated network traffic. At this moment, and probably forever, this means ipsec(4) traffic. Traditionally, isakmpd was configured using the isakmpd.conf(5) file format. A newer, much simpler format is now available: ipsec.conf(5).

isakmpd implements the IKEv1 protocol which is defined in the standards ISAKMP/Oakley (RFC 2408), IKE (RFC 2409), and the Internet DOI (RFC 2407). The newer IKEv2 protocol, as defined in RFC 5996, is not supported by isakmpd but by iked(8). It follows then that references to IKE in this document pertain to IKEv1 only, and not IKEv2.

The way isakmpd goes about its work is by maintaining an internal configuration as well as a policy database which describes what kinds of SAs to negotiate, and by listening for different events that trigger these negotiations. The events that control isakmpd consist of negotiation initiations from a remote party, user input via a FIFO or by signals, upcalls from the kernel via a PF_KEY socket, and lastly by scheduled events triggered by timers running out.

Most uses of isakmpd will be to implement so called "virtual private networks" (VPNs). The ability to provide redundancy is made available through carp(4) and sasyncd(8). For other uses, some more knowledge of IKEv1 as a protocol is required. The RFCs mentioned below are a possible starting point.

On startup isakmpd forks into two processes for privilege separation. The unprivileged child jails itself with chroot(8) to /var/empty. The privileged process communicates with the child, reads configuration files and PKI information, and binds to privileged ports on its behalf. See the CAVEATS section below.

The options are as follows:

|
These options control what address family (AF_INET and/or AF_INET6) isakmpd will use. The default is to use both IPv4 and IPv6.
If given, isakmpd does not set up flows automatically. Instead manual flows may be configured using ipsec.conf(5) or by programs such as bgpd(8). Thus isakmpd only takes care of SA establishment.
config-file
If given, the -c option specifies an alternate configuration file instead of /etc/isakmpd/isakmpd.conf. As this file may contain sensitive information, it must be readable only by the user running the daemon. isakmpd will reread the configuration file when sent a SIGHUP signal.

Note that this option applies only to configuration files in the isakmpd.conf(5) format, not those in the ipsec.conf(5) format.

class=level
Debugging class. It's possible to specify this argument many times. It takes a parameter of the form class=level, where both class and level are numbers. class denotes a debugging class, and level the level you want that debugging class to limit debug printouts at (i.e. all debug printouts above the level specified will not output anything). If class is set to ‘A’, then all debugging classes are set to the specified level.

Valid values for class are as follows:

0
Misc
1
Transport
2
Message
3
Crypto
4
Timer
5
Sysdep
6
SA
7
Exchange
8
Negotiation
9
Policy
10
FIFO user interface
A
All

Currently used values for level are 0 to 99.

The -d option is used to make the daemon run in the foreground, logging to stderr.
fifo
The -f option specifies the FIFO (a.k.a. named pipe) where the daemon listens for user requests. If the path given is a dash (‘-’), isakmpd will listen to stdin instead.
pid-file
By default the PID of the daemon process will be written to /var/run/isakmpd.pid. This path can be overridden by specifying another one as the argument to the -i option. Note that only paths beginning with /var/run are allowed.
When this option is given, isakmpd does not read the policy configuration file and no keynote(4) policy check is accomplished. This option can be used when policies for flows and SA establishment are arranged by other programs like ipsecctl(8) or bgpd(8).
Enable IKE packet capture. When this option is given, isakmpd will write an unencrypted copy of the negotiation packets it is sending and receiving to the file /var/run/isakmpd.pcap, which can later be read by tcpdump(8) and other utilities using pcap_open_offline(3).
packetlog-file
As option -L above, but capture to a specified file. Note that only paths beginning with /var/run are allowed.
udpencap-port
The -N option specifies the listen port for encapsulated UDP that the daemon will bind to.
When the -n option is given, the kernel will not take part in the negotiations. This is a non-destructive mode, so to speak, in that it won't alter any SAs in the IPsec stack.
listen-port
The -p option specifies the listen port the daemon will bind to.
report-file
When you signal isakmpd a SIGUSR1, it will report its internal state to a report file, normally /var/run/isakmpd.report, but this can be changed by feeding the file name as an argument to the -R flag. Note that only paths beginning with /var/run are allowed.
This option is used for setups using sasyncd(8) and carp(4) to provide redundancy. isakmpd starts in passive mode and will not initiate any connections or process any incoming traffic until sasyncd has determined that the host is the carp master. Additionally, isakmpd will not delete SAs on shutdown by sending delete messages to all peers.
When this option is given, NAT-Traversal will be disabled and isakmpd will not advertise support for NAT-Traversal to its peers.
Enables verbose logging. Normally, isakmpd is silent and outputs only messages when a warning or an error occurs. With verbose logging isakmpd reports successful completion of phase 1 (Main and Aggressive) and phase 2 (Quick) exchanges (Information and Transaction exchanges do not generate any additional status information).

When isakmpd starts, it creates a FIFO (named pipe) where it listens for user requests. All commands start with a single letter, followed by command-specific options. Available commands are:

[section]:tag=value
 
[section]:tag=value
 
[section]:tag
 
[section]
 
[section]:tag=value [force]
Update the running isakmpd configuration atomically. ‘set’ sets a configuration value consisting of a section, tag, and value triplet. ‘set’ will fail if the configuration already contains a section with the named tag; use the ‘force’ option to change this behaviour. ‘add’ appends a configuration value to the named configuration list tag, unless the value is already in the list. ‘rm’ removes a tag in a section. ‘rms’ removes an entire section. ‘rmv’ removes an entry from a list, thus reversing an ‘add’ operation.

NOTE: Sending isakmpd a SIGHUP or an "R" through the FIFO will void any updates done to the configuration.

[section]:tag
Get the configuration value of the specified section and tag. The result is stored in /var/run/isakmpd.result.

name
Start the named connection, if stopped or inactive.

class level
 
D A level
 
D T
Set debug class class to level level. If class is specified as ‘A’, the level applies to all debug classes. D T toggles all debug classes to level zero. Another D T command will toggle them back to the earlier levels.

cookies msgid
Delete the specified SA from the system. Specify msgid as ‘-’ to match a Phase 1 SA.

 
Set isakmpd to active or passive mode. In passive mode no packets are sent to peers.

p on[=path]
 
p off
Enable or disable cleartext IKE packet capture. When enabling, optionally specify which file isakmpd should capture the packets to (the default is /var/run/isakmpd.pcap). Note that only paths beginning with /var/run are allowed.

Cleanly shutdown the daemon, as when sent a SIGTERM signal.

Reinitialize isakmpd, as when sent a SIGHUP signal.

Report isakmpd internal state to syslog(3). See the -R option. Same as when sent a SIGUSR1 signal.

Report information on all known SAs to the /var/run/isakmpd.result file.

Tear down all active quick mode connections.

[phase] name
Tear down the named connection, if active. For name, the tag specified in isakmpd.conf(5) or the IP address of the remote host can be used. The optional parameter phase specifies whether to delete a phase 1 or phase 2 SA. The value ‘main’ indicates a phase 1 connection; the value ‘quick’ a phase 2 connection. If no phase is specified, ‘quick’ will be assumed.

In order to use public key based authentication, there has to be an infrastructure managing the key signing. Either there is an already existing PKI isakmpd should take part in, or there will be a need to set one up. The procedures for using a pre-existing PKI varies depending on the actual Certificate Authority (CA) used, and is therefore not covered here, other than mentioning that openssl(1) needs to be used to create a Certificate Signing Request (CSR) that the CA understands.

A number of methods exist to allow authentication:

Passphrase:
This method does not use keys at all, but relies on a shared passphrase.
Host Keys:
Public keys are used to authenticate. See PUBLIC KEY AUTHENTICATION below.
X.509 Certificates:
X.509 Certificates are used to authenticate. See X.509 AUTHENTICATION below.
Keynote Certificates:
Keynote Certificates are used to authenticate. See KEYNOTE AUTHENTICATION below.

When configuring isakmpd for key- and certificate-based authentication, the “Transforms” tag in isakmpd.conf(5) should include “RSA_SIG”. For example, the transform “3DES-SHA-RSA_SIG” means: 3DES encryption, SHA hash, authentication using RSA signatures.

It is possible to store trusted public keys to make them directly usable by isakmpd, bypassing the need to use certificates. The keys should be saved in PEM format (see openssl(1)) and named and stored after this easy formula:

For IPv4 identities:
/etc/isakmpd/pubkeys/ipv4/A.B.C.D
For IPv6 identities:
/etc/isakmpd/pubkeys/ipv6/abcd:abcd::ab:bc
For FQDN identities:
/etc/isakmpd/pubkeys/fqdn/foo.bar.org
For UFQDN identities:
/etc/isakmpd/pubkeys/ufqdn/user@foo.bar.org

Depending on the ID-type field of isakmpd.conf(5), keys may be named after their IPv4 address (IPV4_ADDR or IPV4_ADDR_SUBNET), IPv6 address (IPV6_ADDR or IPV6_ADDR_SUBNET), fully qualified domain name (FDQN), user fully qualified domain name (USER_FQDN), or key ID (KEY_ID).

For example, isakmpd can authenticate using the pre-generated keys if the local public key, by default /etc/isakmpd/local.pub, is copied to the remote gateway as /etc/isakmpd/pubkeys/ipv4/local.gateway.ip.address and the remote gateway's public key is copied to the local gateway as /etc/isakmpd/pubkeys/ipv4/remote.gateway.ip.address. Of course, new keys may also be generated (the user is not required to use the pre-generated keys). In this example, ID-type would also have to be set to IPV4_ADDR or IPV4_ADDR_SUBNET in isakmpd.conf(5).

X.509 is a framework for public key certificates. Certificates can be generated using openssl(1) and provide a means for PKI authentication. In the following example, a CA is created along with host certificates to be signed by the CA.

  1. Create your own Certificate Authority (CA).

    First, create a private key for the CA, and a Certificate Signing Request (CSR) to enable the CA to sign its own key:

    # openssl genrsa -out /etc/ssl/private/ca.key 2048
    # openssl req -new -key /etc/ssl/private/ca.key \
    	-out /etc/ssl/private/ca.csr

    openssl req will prompt for information that will be incorporated into the certificate request. The information entered comprises a Distinguished Name (DN). There are quite a few fields, but some can be left blank. For some fields there will be a default value; if ‘.’ is entered, the field will be left blank.

    After the CSR has been generated, it is used to create and sign a certificate for the CA:

    # openssl x509 -req -days 365 -in /etc/ssl/private/ca.csr \
    	-signkey /etc/ssl/private/ca.key \
    	-extfile /etc/ssl/x509v3.cnf -extensions x509v3_CA \
    	-out /etc/ssl/ca.crt
  2. Create Certificate Signing Requests (CSRs) for IKE peers. The CSRs are signed with a pre-generated private key.

    This step, as well as the next one, needs to be done for every peer. Furthermore the last step will need to be done once for each ID you want the peer to have. The 10.0.0.1 below symbolizes that ID, in this case an IPv4 ID, and should be changed for each invocation. A fully qualified domain name (FQDN) may be used instead of an IPv4 ID. You will be asked for a DN for each run. Encoding the ID in the common name is recommended, as it should be unique.

    # openssl req -new -key /etc/isakmpd/private/local.key \
    	-out /etc/isakmpd/private/10.0.0.1.csr

    Now take these certificate signing requests to your CA and process them as below. A configuration file is used to add a extension field matching the ID used by isakmpd to the certificate.

    If using an IPv4 ID, copy /etc/ssl/x509v3.cnf to a temporary file and edit it to replace $ENV::CERTIP with the IP address (10.0.0.1 in this example), then generate a signed certificate:

    # sed 's,\$ENV::CERTIP,10.0.0.1,' \
    	< /etc/ssl/x509v3.cnf > ~/tmp_x509v3.cnf
    # openssl x509 -req \
    	-days 365 -in 10.0.0.1.csr \
    	-CA /etc/ssl/ca.crt -CAkey /etc/ssl/private/ca.key \
    	-CAcreateserial -extfile ~/tmp_x509v3.cnf \
    	-extensions x509v3_IPAddr -out 10.0.0.1.crt

    For an FQDN certificate, replace $ENV::CERTFQDN with the hostname and generate a signed certificate:

    # sed 's,\$ENV::CERTFQDN,somehost.somedomain,' \
    	< /etc/ssl/x509v3.cnf > ~/tmp_x509v3.cnf
    # openssl x509 -req \
    	-days 365 -in somehost.somedomain.csr \
    	-CA /etc/ssl/ca.crt -CAkey /etc/ssl/private/ca.key \
    	-CAcreateserial -extfile ~/tmp_x509v3.cnf \
    	-extensions x509v3_FQDN -out somehost.somedomain.crt

    If CERTFQDN is being used, make sure that the subjectAltName field of the certificate is specified using srcid in ipsec.conf(5). A similar setup will be required if isakmpd.conf(5) is being used instead.

    Put the certificate (the file ending in .crt) in /etc/isakmpd/certs/ on your local system. Also carry over the CA cert /etc/ssl/ca.crt and put it in /etc/isakmpd/ca/.

To revoke certificates, create a Certificate Revocation List (CRL) file and install it in the /etc/isakmpd/crls/ directory. See openssl(1) and the ‘crl’ subcommand for more info.

Keynote is a trust-management framework. Keys can be generated using keynote(1) and provide an alternative means for isakmpd to authenticate. See keynote(4) for further information.

/etc/isakmpd/ca/
The directory where CA certificates are kept.
/etc/isakmpd/certs/
The directory where IKE certificates are kept, both the local certificate(s) and those of the peers, if a choice to have them kept permanently has been made.
/etc/isakmpd/crls/
The directory where CRLs are kept.
/etc/isakmpd/isakmpd.conf
The configuration file. As this file can contain sensitive information it must not be readable by anyone but the user running isakmpd.
/etc/isakmpd/isakmpd.policy
The keynote policy configuration file. The same mode requirements as isakmpd.conf.
/etc/isakmpd/keynote/
The directory where KeyNote credentials are kept.
/etc/isakmpd/private/
The directory where local private keys used for public key authentication are kept. By default, the system startup script rc(8) generates a key-pair when starting, if one does not already exist. The entire keypair is in local.key, and a copy of the public key suitable for transferring to other hosts is extracted into /etc/isakmpd/local.pub. There has to be a certificate for local.key in the certificate directory, /etc/isakmpd/certs/. local.key has the same mode requirements as isakmpd.conf.
/etc/isakmpd/pubkeys/
The directory in which trusted public keys are kept. The keys must be named in the fashion described above.
/var/run/isakmpd.fifo
The FIFO used to manually control isakmpd.
/var/run/isakmpd.pcap
The default IKE packet capture file.
/var/run/isakmpd.pid
The PID of the current daemon.
/var/run/isakmpd.report
The report file written when SIGUSR1 is received.
/var/run/isakmpd.result
The report file written when the ‘S’ or ‘C get’ command is issued in the command FIFO.

openssl(1), getnameinfo(3), pcap_open_offline(3), ipsec(4), ipsec.conf(5), isakmpd.conf(5), isakmpd.policy(5), iked(8), sasyncd(8), ssl(8), tcpdump(8)

D. Piper, The Internet IP Security Domain of Interpretation for ISAKMP, RFC 2407, November 1998.

D. Maughan, M. Schertler, M. Schneider, and J. Turner, Internet Security Association and Key Management Protocol (ISAKMP), RFC 2408, November 1998.

D. Harkins and D. Carrel, The Internet Key Exchange (IKE), RFC 2409, November 1998.

T. Kivinen, B. Swander, A. Huttunen, and V. Volpe, Negotiation of NAT-Traversal in the IKE, RFC 3947, January 2005.

This implementation of the ISAKMP/Oakley key management protocol was done in 1998 by Niklas Hallqvist and Niels Provos, sponsored by Ericsson Radio Systems.

When storing a trusted public key for an IPv6 identity, the form of address representation, i.e. "::" instead of ":0:0:0:", must be used or the matching will fail. isakmpd uses the output from getnameinfo(3) for the address-to-name translation. The privileged process only allows binding to the default port 500 or unprivileged ports (>1024). It is not possible to change the interfaces isakmpd listens on without a restart.

For redundant setups with carp(4) and sasyncd(8), sasyncd(8) must be manually restarted every time isakmpd is restarted, and isakmpd.conf(5) must explicitly configure isakmpd to listen on the virtual IP address of each carp(4) interface.

August 30, 2019 OpenBSD-current