Citylink's official Debian mirror (ftp.nz.debian.org) is apparently
again broken (not even listening on port 80 from where I'm sitting).
Would anyone be seriously interested in donating rack space, national
transit, or hardware to the cause of another official Debian mirror for
Having reliable and cheap access to free software projects like Debian
is critically important, and having a second official mirror seems like
a no-brainer to me.
I am very happy to donate my time to admin and monitor any equipment
provided for this, and to liase with the Debian project as required.
Perhaps Citylink could give us some insight in to the traffic levels
their existing mirrors see?
We have started receiving complaints from several of our customers for whom we are the MX for their domain. Some addresses in the domain are delivered locally, some are relayed to other mailboxes e.g. gmail and xtra.
It seems that there are folks who still believe that SPF is a good thing and have ‘-all’ on their record. Xtra it would appear is now enforcing the rejection on email we relay to them. We are placed in an awkward position. Our customer wants us to fix something that we are not really able to. The SPF policy belongs to the sender and Xtra are doing the rejection.
I have come across another add on technology for postfix called SRS for rewriting senders but that seems to be broken from all accounts as well.
Is anyone else having these sort of issues, is there something I should be doing in the ‘relay’ process that will prevent the rejections I’d love to hear.
GodZone Internet Services, a division of AGRE Enterprises Ltd.,
P.O. Box 8020, Palmerston North, New Zealand 4446
Ph +64 6 357 8168, Mob +64 27 542 4015
“Specialising in providing low-cost professional Internet Services since 1997"
Given the importance of Citylink's IX services in the community, I feel
that it is valuable to notify this communication from them to the entire
I would note that they are explicitly stating in this communication they
will not ever forcibly require consistent behavior. I would note that this
is in direct violation of RFC7947;
*22.214.171.124. Route Server AS_PATH Management*
* As a route server does not participate in the process of forwarding*
* data between client routers, and because modification of the AS_PATH*
* attribute could affect the route server client BGP Decision Process,*
* the route server SHOULD NOT prepend its own AS number to the AS_PATH*
* segment nor modify the AS_PATH segment in any other way.*
I would like to encourage (publicly) that Citylink implement this
consistently, rather than in a haphazard way that varies customer by
customer. Consistency is the key to supporting the NZ internet
RFC7947 doesn't say "do this when your customer can be bothered")*.
Additionally, I would like to encourage Citylink to communicate with the
NZNOG community when publishing updates that could have an operational
impact on the internet in NZ as a whole.
####### FOREWARDED MESSAGE BELOW #######
CityLink would like to inform you of some exciting NZIX developments that
concern your existing ExchangeNET connection(s). We’re introducing some
changes to the way we operate the NZIX route servers in order to improve
our compliance with recent proposed standards (RFC7947 and RFC7948), and to
minimise the disruption caused by routine administrative changes going
CityLink provides two route servers at each of the NZIX exchanges
- currently Auckland (APE), Hamilton (HIX), Wellington (WIX), Christchurch
(CHIX) and Dunedin (DPE). The route servers facilitate
multilateral interconnection between the participants at each exchange by
simplifying the BGP configuration and admin overhead otherwise needed
to exchange routing information between multiple eBGP speakers. The NZIX
route servers have traditionally prepended the exchange’s Autonomous System
(AS) number to the AS path of the prefixes they advertise back out to peers.
*What is changing?*
A number of ExchangeNET customers have expressed a preference that the NZIX
route servers no longer prepend the exchange AS number to the AS path of
advertisements sent to them. This preference is consistent with recent
proposed standards (RFC7947 and RFC7948), and is the recommended best
current practice for Internet exchange route server operation. However, in
order to put this change into effect, ExchangeNET customers who peer with
the NZIX route servers may need to make a (potentially disruptive) change
to their existing BGP configuration.
While CityLink would prefer all ExchangeNET customers peering with the NZIX
route servers to support this alignment with best current practice, we
recognise that the timing to implement such changes may differ based on
your individual business needs. Therefore, we have decided to implement the
changes on a per-participant basis, and we invite those
ExchangeNET customers who would like to “opt-in” to the changes (that is,
for the NZIX route servers to no longer prepend the exchange AS number to
the AS path of the advertisements sent to them), to contact us. We will
endeavor to accommodate requests for configuration changes to be made at
times convenient for each ExchangeNET customer, commencing from *13th March
*Note:* We will *not* be changing the configuration of any existing
sessions without that participant’s agreement.
*What are the implications of the change?*
· ExchangeNET customers who peer with the NZIX route servers will
see the AS path for prefixes received shortened by one hop (the exchange AS
number). This may alter the current choice of best path for those prefixes.
· If you had been using the presence of an exchange AS in the AS
path to influence your organisation’s routing policy, you may need to amend
your configuration in order to achieve your routing policy objectives. It
may not be quite so obvious that a given route has been learnt from a
particular NZIX exchange point going forward.
· Your router may need to be configured with its equivalent of *no
bgp enforce-first-as* to restore its BGP session(s) with the NZIX route
servers after the change is made.
*I’d like to “opt-in” to the change. What do I need to do?*
If you’d like us to remove the exchange AS number from the AS path of the
advertisements you receive from the NZIX route servers, please email
us at *peering(a)citylink.co.nz
<peering(a)citylink.co.nz>* with your AS number, the contact details for the
person(s) from your organisation that will be involved in making the
changes, and an initial indication of when you’d like to implement the
change (if known). Changes will only be implemented at a mutually-agreed
time to minimise any disruption, commencing from *13th March 2017* onwards.
*I’d prefer not to “opt-in” to the change right now. Do I need to do
If you’d prefer that the NZIX route servers continue prepending the
exchange AS number to the AS path of advertisements sent to you, you don’t
need to take any action right now. However, we would appreciate a return
email confirming the contact details for the most suitable person(s) from
your organisation whom we should engage with on these matters going forward.
*I’d prefer not to “opt-in” to the change at all. Will I be encouraged to
It is hoped that over time all ExchangeNET customers will choose to
“opt-in” and thus support CityLink’s efforts to align the NZIX exchanges
with best current practices, to the benefit of the wider NZ Internet
*What will be the “default” configuration for new ExchangeNET customers?*
Going forward, all new BGP sessions configured on the NZIX route servers
will conform to the recommend best current practice (that is, the route
servers will *not* prepend the exchange AS number to the AS path of
prefixes sent to peers).
*Is anything else changing?*
Yes! The method by which configuration changes have historically been
“pushed” to the NZIX route servers has necessitated a “hard reset” of the
BGP session(s) with all existing peers. Although the two route servers at
each exchange are reloaded independently (ensuring that each peer has an
established session with at least one NZIX route server at all times), the
overheads associated with performing the hard reset are higher than they
need to be. In order to minimise the impact of routine administrative
changes going forward, we are changing the way in which configuration
changes are pushed to the route servers. Note that changes will continue to
be applied to each route server independently, however, thereby ensuring
optimal availability for NZIX.
*I have a question about these changes. Who should I contact?*
Please email us at *peering(a)citylink.co.nz <peering(a)citylink.co.nz>* in the
first instance. We’ll be happy to help!
[image: CityLink Ltd] <http://www.citylink.co.nz/>
*Toll Free* 0800 424 895
Level 6, 25 Cambridge Terrace
This mail message contains information that is confidential and which may
be subject to legal privilege. If you are not the intended recipient, you
must not use, distribute or copy this message. If you have received this
message in error, please notify the sender immediately and erase this mail
and its attachments.
> On 8/02/2017, at 1:15 PM, Criggie <criggie(a)criggie.org.nz> wrote:
> Its not really an answer, but SPF is doing what its told.
> So either you have to convince Xtra's mail to disable their rejections
> (unlikely) or you have to get something into the customer's SPF record
> to permit your hosts to send.
The domains with the hard fail SPFs are not my customers so I have no control over them at all.
The issue is people emailing from those domains to domains we do host and they are getting the bounce and blaming us. (perhaps rightly so, but this has only become an issue in the last week and even then for only 2 customers)
> Dirty hack that won't work much is to IPv6-enable your mail relay. There
> seem to be fewer restrictions on that side, at least google is a little
> more accepting.
Already fully IPv6 enabled.
So it looks like I need to go back and have another look at SRS. It would seem that either few organisations hosting email are ‘relaying’ like we do or have already implemented SRS. If that is the case, I’d be interested in knowing whether SRS creates its own set of problems.
Hi New Zealand internet community!
I wanted to remind everyone that the APNIC elections
are coming up, with voting
closing in 6 days (3pm Monday, 27 February 2017).
It's important that we all get involved in the democratic process that
APNIC has and vote - those elected will determine policies that impact us
all (for instance around IP address transfers and justification of IP space
as we ramp down IPv4).
Specifically, I wanted to highlight two candidates from the list of nominees
<https://2017.apricot.net/elections/nominations/> that you may want to
consider voting for to represent us. Both have had ties to the New Zealand
internet community for over a decade, understand the challenges we face,
and I believe would represent our needs well.
Jon Brewer <https://2017.apricot.net/elections/nominations/jonathan-brewer/> is
well known to our community. Wellington based, he founded Araneo and ran
this company for some years, then since selling Araneo to TeamTalk has
focused on helping developing communities (particularly in the pacific
islands) establish better connectivity locally and to the world.
*Gaurab Raj Upadhaya*
had involvement in many countries whose internet infrastructures have been
developing over the past decade. He has strong ties to the networking
community in New Zealand; these began in the early FX days. Starting his
career in Nepal, he has experienced challenges greater than those we face
in New Zealand around being a remote internet community connected to a
US-centric internet first hand. He now leads the network organization at
*Don't forget to vote!*
Tim Hoffman | Twitter, Inc.
1355 Market St. | San Francisco, CA | 94103
I'll be over in NZ (Auckland) early next week with the boss and thought a few beers were in order to catch up with people that we may not have seen in a while as well as to meet more of the NZNOG folk.
We'll be putting on a few drinks at The Bluestone Room (9-11 Durham Ln, Auckland) on the 14th Feb from about 5:30pm, feel free to come down and join us for a few beers.
Drinks - Auckland CBD 14/2
Fraser McGlinn - Senior Network Operations Officer
Main Office: +61356224600
Level 1, 5/15 Phoenix Street Warragul, Victoria 3820, Australia
This communication, including any attachments, is confidential. If you are not the intended recipient, you should not read it - please contact me immediately, destroy it, and do not copy or use any part of this communication or disclose anything about it.
Is anyone else noticing high packet loss to AWS Sydney this morning?
PING ap-southeast-2.console.aws.amazon.com (126.96.36.199) 56(84) bytes of data.
64 bytes from 188.8.131.52 (184.108.40.206): icmp_seq=12 ttl=240 time=40.4 ms
64 bytes from 220.127.116.11 (18.104.22.168): icmp_seq=17 ttl=240 time=39.6 ms
64 bytes from 22.214.171.124 (126.96.36.199): icmp_seq=20 ttl=240 time=40.7 ms
64 bytes from 188.8.131.52 (184.108.40.206): icmp_seq=32 ttl=240 time=161 ms
64 bytes from 220.127.116.11 (18.104.22.168): icmp_seq=34 ttl=240 time=40.2 ms
64 bytes from 22.214.171.124 (126.96.36.199): icmp_seq=39 ttl=240 time=294 ms
64 bytes from 188.8.131.52 (184.108.40.206): icmp_seq=41 ttl=240 time=40.7 ms
64 bytes from 220.127.116.11 (18.104.22.168): icmp_seq=47 ttl=240 time=40.4 ms
64 bytes from 22.214.171.124 (126.96.36.199): icmp_seq=54 ttl=240 time=65.5 ms
> On 8/02/2017, at 1:03 PM, Jasper Bryant-Greene <jbg(a)rf.net.nz> wrote:
> SRS is the solution to this and is a standard, available on most mail systems. SPF is a good idea according to most of the internet, and is very widely implemented, so I suggest just getting with the times and fixing your mail relay to rewrite sender addresses according to the SRS standard rather than claiming to send mail for domains it's not authorized to do so for.
I have found several articles saying that SRS is, like SPF, basically broken in that it doesn’t handle any situation where multiple relays are involved (not prepared to comment on whether that is right or not), hence my reluctance at this stage to using it.
I agree that in principal SPF is a good idea but it breaks mailing lists, forwarding and relays. It has been my believe that using ~all (soft fail) is better than -all (hard fail) so that various anti-spam systems can use the soft fail as a contributing weighting, along with other things to determine what should be done with a message and to not rely on just the SPF.