problems observed with IRRToolSet-4.8.2 - and the mailing list!?

Ruediger Volk, Deutsche Telekom T-Com - TE14 P1 rv at nic.dtag.de
Fri Feb 10 11:01:07 UTC 2006


Trying to install and use a new copy of IRRToolSet 4.8.2 we recently run
into several problems;  some of which are not really new.

The set of problems also includes the working status of the mailing list
irrtoolset at isc.org, which had NO traffic since early November, and
some messages sent to it since from members of my team to this address
seem to have met a black hole.  So this message is ALSO meant to test
the mailing list again (listmaster: we know your address works, please check
arrival/redistribution/archiving of this message).

list of problems found:

(1) http://ftp.isc.org/isc/IRRToolSet/IRRToolSet-4.8.2/IRRToolSet-4.8.2.tar.gz
    is different from the version available from 
    ftp://ftp.ripe.net/tools/IRRToolSet/IRRToolSet-4.8.2/
    We understand that the ISC version will be updated in the near future.

(2) APNIC's whois server (RIPE NCC software with RPSS functions disabled)
    has changed syntax of some embedded comments in a way that violates
    the protocol assumptions in IRRToolSet 4.8.2
    (required/allowed "empty lines" ~ "\n\n" in the context of "%" comment
    lines).
    We will offer a fix for this (in birdwhoisc.cc) separately.
    (This problem seems to be around since APNIC switched to a database
    version supporting IPv6/RPSLng)

(3) We are seeing cases where resolving to route-sets
    (a) an ASnnn and
    (b) an AS-set AS-xxxx which includes ASnnn as a member
    lists some prefixes
	as members of ASnnn
	but NOT as members of AS-xxxx
    (Which turns something like "125.16.0.0/14^14-24"
    into some thousand separately listed 125.16.x.y/n^n-n or 125.16.x.y/n^m-m
    that don't combine into any 125.16.x.y/n^m-l unless m=l)
    
    We have verified that peval and the database server (using the standard
    RIPE NCC server whois.RIPE.NET) actually include the complete query
    for ASnnn (including all results) as a proper substring of the
    complete query for AS-xxxx;  so the loss seems to happen somewhere
    in the IRRtoolSet code.
    Does anybody remember seeing something similar?
    We will come back with examples to reproduce the problem separately.

Best regards,
  Ruediger


Ruediger Volk

Deutsche Telekom AG -- Internet Backbone Engineering

E-Mail:  rv at NIC.DTAG.DE
Phone:   +49 251 7985 200	fax:  +49 251 7985 109


More information about the irrtoolset mailing list