ipsec_newhostkey(8) generate a new raw RSA authentication key for a host

SYNOPSIS

ipsec newhostkey [[--configdiranssdbdir] | [--password password]] [[--quiet] | [--verbose]] [--bits bits] [--hostname hostname] --output filename

DESCRIPTION

newhostkey

outputs (into filename, which can be '-' for standard output) an RSA private key suitable for this host, in /etc/ipsec.secrets format (see ipsec.secrets(5)) using the --quiet option per default.

The --output option is mandatory. The specified filename is created under umask 077 if nonexistent; if it already exists and is non-empty, a warning message about that is sent to standard error, and the output is appended to the file.

The --quiet option suppresses both the rsasigkey narrative and the existing-file warning message.

When compiled with NSS support, --configdir specifies the nss configuration directory where the certificate key, and modsec databases reside. There is no default value, though /etc/ipsec.d might be sensible choice.

When compiled with NSS support, --password specifies a module authentication password that may be required if FIPS mode is enabled

The --bits option specifies the number of bits in the key; the current default is 2192 and we do not recommend use of anything shorter unless unusual constraints demand it.

The --hostname option is passed through to rsasigkey to tell it what host name to label the output with (via its --hostname option).

The output format is that of rsasigkey, with bracketing added to complete the ipsec.secrets format. In the usual case, where ipsec.secrets contains only the host’s own private key, the output of newhostkey is sufficient as a complete ipsec.secrets file.

FILES

/dev/random, /dev/urandom

HISTORY

Written for the Linux FreeS/WAN project <m[blue]http://www.freeswan.orgm[]> by Henry Spencer.

BUGS

As with rsasigkey, the run time is difficult to predict, since depletion of the system’s randomness pool can cause arbitrarily long waits for random bits, and the prime-number searches can also take unpre dictable (and potentially large) amounts of CPU time. See ipsec_rsasigkey(8) for some typical performance numbers.

A higher-level tool which could handle the clerical details of changing to a new key would be helpful.

The requirement for --output is a blemish, but private keys are extremely sensitive information and unusual precautions seem justified.