- Judy dynamic arrays
- http://judy.sourceforge.net/
- FreeBSD: ports/devel/judy
- Debian/Ubuntu: libjudy-dev
- MacOS X: macports judy
- Test::Command::Simple perl module (for tests only)
- FreeBSD: ports/devel/p5-Test-Command-Simple
- anywhere: cpanm Test::Command::Simple
git clone https://github.com/DNS-OARC/validns
cd validns
sh autogen.sh
./configure
make
make install
Every record and every supported directive should be parsable,
which consitutes the most basic check of all.
The validns
program will report the exact reason
why it cannot parse a record or a directive.
Other basic checks include:
- there could only be one SOA in a zone;
- the first record in the zone must be an SOA record;
- a record outside the apex;
- TTL values differ within an RR set (excepting RRSIG);
- type exists, but NSEC does not mention it for name;
- NSEC mentions type, but no such record found for name;
- NSEC says x is the last name, but z exists;
- NSEC says z comes after x, but nothing does;
- NSEC says z comes after x, but y does;
- signature is too new;
- signature is too old;
- RRSIG exists for non-existing type type;
- RRSIG's original TTL differs from corresponding record's;
- RRSIG(type): cannot find a signer key;
- RRSIG(type): cannot verify the signature;
- RRSIG(type): cannot find the right signer key;
- NSEC3 record name is not valid;
- multiple NSEC3 with the same record name;
- no corresponding NSEC3 found for name;
- type exists, but NSEC3 does not mention it for name;
- NSEC3 mentions type, but no such record found for name;
- there are more record types than NSEC3 mentions for name;
- broken NSEC3 chain, expected name, but nothing found;
- broken NSEC3 chain, expected name1, but found name2;
- NSEC3 without a corresponding record (or empty non-terminal).
The following policy checks are understood:
single-ns
: Checks that there are at least two NS records per name (or zero).
cname-other-data
: Checks for CNAME and other data condition (excluding possible and NSEC).
dname
: DNAME checks: no multiple DNAMEs, no descendants of a node with a DNAME. Please note that DNAME/CNAME clash is handled by CNAME and other data check already.
dnskey
: DNSKEY checks: public key too short, leading zero octets in public key exponent or modulus.
nsec3param-not-apex
: NSEC3PARAM, if present, should only be at the zone apex.
mx-alias
: MX exchange should not be an alias.
ns-alias
: NS nsdname should not be an alias.
rp-txt-exists
: TXT domain name mentioned in RP record must have a corresponding TXT record if it is within the zone.
tlsa-host
: Domain name of a TLSA record must be a proper prefixed DNS name.
ksk-exists
: A KSK key must exist in a signed zone.
smimea-host
: Domain name must have the form which is proper for an SMIMEA record.
all
: Activates all of the above policy checks (but not the ones described below).
permit-starting-hyphen
: Starting hyphen (normally disallowed) is permitted in names.
A number of corners were cut to assume the most usual way of doing things.
Therefore, in many cases, validns
currently does not strictly adhere to various standards.
In particular, it should be possible (and easy) to construct a perfectly valid zone file which validns
will report as problematic.
It is expected that those cases will be all fixed over time.
But if you have a valid zone which validns
cannot parse, please do report this fact to the author, with examples.
If there is a need for the community to fix a particular omission, it will be fixed sooner.
Needless to say, patches are always welcome.
- textual segments in TXT and HINFO must be enclosed in double quotes;
- a dot within a label is not currently supported;
If at least one NSEC3 record uses opt-out flag, validns
assumes it is used as much as possible, that is, every unsigned delegation does not have a corresponding NSEC3 record.
This is done for reasons of efficiency, to avoid calculating cryptographic hashes of every unsigned delegation.
If this assumption is wrong for a zone, validns
will produce spurious validation errors.
Thanks go to Andy Holdaway, Daniel Stirnimann, Dennis Kjaer Jensen, Goran Bengtson, Hirohisa Yamaguchi, Hugo Salgado, Jake Zack, Jakob Schlyter, Koh-ichi Ito, Mathieu Arnold, Miek Gieben, Patrik Wallstrom, Paul Wouters, Ryan Eby, Tony Finch, Willem Toorop, and YAMAGUCHI Takanori for bug reports, testing, discussions, and occasional patches.
Special thanks to Stephane Bortzmeyer and Phil Regnauld.
Thanks for AFNIC which funded major portion of the development. Thanks for SWITCH for additional funding.