Ah, very cool. I was thinking more of the SRV record stuff
in FTS-5004 so you could get, e.g., binkd addresses and
things like that. Using TXT and HINFO RRs in the zone
files would also get you all sorts of interesting information.
Is any of that data in place?
What do you use to create your nodelist? If you use something like
makenl, it will do everything for you with regards to nodelist creation,
Mystic processes the nodelist, but husky for example, follows the FTSC spec and won't reprocess/replace the nodelist unless the day number
equals or matches the next release date. My Mystic setup saw him, but RNTrack and husky didn't as the nodelist you hatched out had the same
day number.
One way around this, if you do use makenl, is switch it's settings to daily instead of weekly. This way, when you release a new nodelist, it will contain that days number and will be processed properly by systems who use FTSC spec nodelist processing.
I have a shit ton of experience with makenl which, IMO, is the best of
the best for this. If you want, I can give you a hand scripting and setting it up so it just runs.
Did I send you the Perl script that I use? It creates the following
from a standard nodelist:
One way around this, if you do use makenl, is switch it's settings to daily instead of weekly. This way, when you release a new nodelist, it will contain that days number and will be processed properly by systems who use FTSC spec nodelist processing.
Hmm yep that could be the happy middle ground to assist all. Are there any downsides to this?
Just to chime in on this from my experience - Daily MakeNL nodelists
solve two problems. First it eliminates a bit of complexity with maintaining traditional list+differentials and the second is that any changes are obviously "daily".
Thanks for doing so :) I'm very open to guidance.
So would it become that I hatched daily nodelists or are we saying we run MakeNL in a fashion whereby it creates a format of the nodelist that can be zipped but when processed by systems is accepted as the correct
version as of that day vs a Friday only one?
On 02-15-20 20:54, Avon wrote to Vk3jed <=-
On 15 Feb 2020 at 05:02p, Vk3jed pondered and said...
Did I send you the Perl script that I use? It creates the following
from a standard nodelist:
I think you have... I think... but it's been a while since I looked at this stuff and to be honest it's probably buried in a folder
somewhere...
Hey Siri.... google search Tony's cool Perl script....
On 02-15-20 22:14, Avon wrote to Atreyu <=-
I should clarify, is there an option to process as daily but still
hatch as weekly? Or am I down the path of just changing a setting in MakeNL and then sending out daily nodelists that are/are not
compressed? I'm a tad confused.
Hmm yep that could be the happy middle ground to assist all. Are there
any downsides to this?
Thanks Frank, appreciate the offer. If all I need to do is to change that switch then I think I may be OK. But having said that if there's other stuff I should be cognizant of I'm totally open to your wisdom, and
So would it become that I hatched daily nodelists or are we saying we run MakeNL in a fashion whereby it creates a format of the nodelist that can be zipped but when processed by systems is accepted as the correct
version as of that day vs a Friday only one?
I should clarify, is there an option to process as daily but still hatch as weekly? Or am I down the path of just changing a setting in MakeNL
and then sending out daily nodelists that are/are not compressed? I'm a tad confused.
So would it become that I hatched daily nodelists or are we saying we run MakeNL in a fashion whereby it creates a format of the nodelist that can be zipped but when processed by systems is accepted as the correct
version as of that day vs a Friday only one?
Either, but I think if you can make the whole thing automatic like v??? (sorry can't remember the handle atm) suggested via crontab you can
remove one more thing you have to deal with and solve the issue at the same time.
On 02-15-20 16:25, Embalmed wrote to Avon <=-
So would it become that I hatched daily nodelists or are we saying we run MakeNL in a fashion whereby it creates a format of the nodelist that can be zipped but when processed by systems is accepted as the correct
version as of that day vs a Friday only one?
Either, but I think if you can make the whole thing automatic like v??? (sorry can't remember the handle atm) suggested via crontab you can
remove one more thing you have to deal with and solve the issue at the same time.
That would be me. All I have to think about with my nodelist processing is to ensure that the data is up to date, and the automatic processing does the rest. :)
So would it become that I hatched daily nodelists or are we saying we run MakeNL in a fashion whereby it creates a format of the nodelist that can be zipped but when processed by systems is accepted as the correct version as o that day vs a Friday only one?
You can do both. Keep the current filename and weekly production of the Fsxnet nodelist as-is so Sysops used to the current hatching don't need
to change anything. Make up a different nodelist filename for the Daily, ie. FSXDAILY and promote it accordingly; Sysops will eventually switch
and enjoy the benefits I pointed out earlier.
Fun fact, the reason why Fidonet has seperate nodelist generation per
zone and not one centrally-pubished nodelist dates back to the 80's and the days of IBM XT's and MFM hard drives. Compiling/indexing the
nodelist was painfully slow and disk-space very limited. When a Zone placed their segment first before other zones, indexing was faster and mailers had the option of dropping the other Zone segments if they were not required.
Long distance was also a factor... this also explained the rise of Zonegates.
On 02-16-20 03:13, Embalmed wrote to Vk3jed <=-
That would be me. All I have to think about with my nodelist processing is to ensure that the data is up to date, and the automatic processing does the rest. :)
Sorry I was going to attempt it but knew I would butcher the name ;) Sounds nice!
Yikes :) The joys of early tech. When I think back to how much money I spent on 14.4k modems, then 28.8k, 33.6k etc.. and the 20 megabyte HDD -
I might start to cry a little :)
Can you guide me please as to what I should add?
Do you want me to send you the perl script I use for SciNet? It does everything for you and creates a bind9 compatible hosts conf file.
If you want an example, use https://mxtoolbox.com/SRVLookup.aspx and lookup your node in SciNet's SRV record for binkp as you use 24555. It is _binkp._tcp.f102.n3.z77.scinet-ftn.org
I can also publish the DNS for you if you wish. I have a cluster of DNS servers and have the infrastructure all setup here and already get a
copy of the fsxnet nodelist since I am a member of fsxnet. It would be just one more script I run and no big deal. You would just have to point the domain you want to use to my dns servers.
The issue in this case is that you shouldn't add SRV
RR's for a hostname that is an alias (e.g., a CNAME).
Another workaround might be to have a dedicated
"nodelist" domain; suppose one ponies up the 20
euros a year for a Dutch domain like `fsxnet.nl`;
then that domain might consist only of SRV
records pointing to actual host names (not CNAMEs).
People who want port information do SRV queries
against that domain; if you just want to connect,
you do a query against the domain in fsxnet.org.
I have a DNS server running here for fsxnet.nz and Vorlon also gets a
copy of the DNS records and is listed as another name server. Would
you be happy to receive copies of a zone file and act as another name server?
I have a DNS server running here for fsxnet.nz and Vorlon also gets a
copy of the DNS records and is listed as another name server. Would you
be happy to receive copies of a zone file and act as another name server?
On 19 Feb 2020 at 04:16p, tenser pondered and said...
The issue in this case is that you shouldn't add SRV
RR's for a hostname that is an alias (e.g., a CNAME).
Forgive me I'm still learning all things DNS, so some of this is not
clear to me as I have not used some of the zone records to know them
well.
On 18 Feb 2020 at 05:45p, Oli pondered and said...
Paul, have you given any thought to doing DNS distribution
for nodelist data, a la http://ftsc.org/docs/fts-5004.001 ?}
I'd kind of like to play around with that.
IMHO that specification and most implementations are broken by design.
The problem are SRV
records, non-standard ports and different hostnames for different
protocols.
You'll get no argument from me that most of the FTSC
standards are broken, or at a minimum poorly specified.
That said, I'm not sure I follow your argument: could
you expand on it a bit? If one is going to use
non-standard ports (and there are actually some decent
reasons for that), then SRV records seem OK to me.
On 02-19-20 21:32, Avon wrote to Netsurge <=-
I have a DNS server running here for fsxnet.nz and Vorlon also gets a
copy of the DNS records and is listed as another name server. Would you
be happy to receive copies of a zone file and act as another name
server?
On 02-20-20 03:41, tenser wrote to Avon <=-
So the designers of DNS chose, instead, to bypass
the entire problem by mandating that you couldn't
attach such records to a DNS alias. It's a
pragmatic decision that sidesteps lots of consistency
problems.
If one reads FTS-5004 carefully, one will notice
that for hosts that require SRV records for e.g.
non-standard port numbers, those entries have A
and AAAA records and no CNAMEs, while those that
use standard ports and don't need SRV records just
have CNAMEs. Of course, the problem with THAT is
that now you've got different IP addresses listed
in different domains, and one is under control
(presumably) two different parties: there's nothing
technically keeping them in sync anymore. If one
wants to move one's system from e.g. one's home
network to a cloud provider with an IP address
change, then you've got to update your local
domain _and_ the nodelist domain. That sounds
painful.
Well, a technical solution might be for the
nodelist->zone generation software to run as a
daemon and do continual zone updates. You could
do something like query each node in the list
and generate A and AAAA records in the nodelist
zone for that host, but keep track of the record
TTLs and, once they expire, query again and
update the zone if the IP address changed. That
would be robust against the sorts of consistency
issues mentioned above in the sense that latency
of nodelist domain changes would be more or less
that of cached result expiration from the source
domain, which an IP address change would have to
suffer through anyway, but it would be rather
more involved than just a simple script that read
and parsed one text file and generated another.
On 02-20-20 03:41, tenser wrote to Avon <=-
So the designers of DNS chose, instead, to bypass
the entire problem by mandating that you couldn't
attach such records to a DNS alias. It's a
pragmatic decision that sidesteps lots of consistency
problems.
I remember similar issues with MX records, and pointing an MX record to
a CNAME was/is a no no.
Well, a technical solution might be for the
nodelist->zone generation software to run as a
daemon and do continual zone updates. You could
do something like query each node in the list
and generate A and AAAA records in the nodelist
zone for that host, but keep track of the record
TTLs and, once they expire, query again and
update the zone if the IP address changed. That
Gets ugly when TTL is 60 seconds (common for DDNS services). ;)
On 02-20-20 15:18, tenser wrote to Vk3jed <=-
I remember similar issues with MX records, and pointing an MX record to
a CNAME was/is a no no.
Correct. The only other RRs you are permitted to
attach to a CNAME are all related to DNSSEC.
Gets ugly when TTL is 60 seconds (common for DDNS services). ;)
I actually don't think it would be that bad. That probably
amounts to a handful of domains per minute.
Another option, though really rather more involved, is to
write a special authoritative DNS server for the nodelist
domain. It can maintain the IP address information for
the nodelist domain and synthesize RRs on the fly, lazily
fetching the results from queries on the underlying domain
and its own cache of those responses and their expiration
times. Then the frequency of lookups is on the order of
the number of times a particular node is queried, modulo
the target node's TTL.
Hmm, maybe we can all help each other out here with name servers, and I can also make my third party slave available. :)
CNAMEOn 02-20-20 03:41, tenser wrote to Avon <=-
So the designers of DNS chose, instead, to bypass
the entire problem by mandating that you couldn't
attach such records to a DNS alias. It's a
pragmatic decision that sidesteps lots of consistency
problems.
I remember similar issues with MX records, and pointing an MX record to a
was/is a no no.
Another option, though really rather more involved, is to
write a special authoritative DNS server for the nodelist
domain. It can maintain the IP address information for
the nodelist domain and synthesize RRs on the fly, lazily
fetching the results from queries on the underlying domain
and its own cache of those responses and their expiration
times. Then the frequency of lookups is on the order of
the number of times a particular node is queried, modulo
the target node's TTL.
Another option, though really rather more involved, is to
write a special authoritative DNS server for the nodelist
domain. It can maintain the IP address information for
the nodelist domain and synthesize RRs on the fly, lazily
fetching the results from queries on the underlying domain
and its own cache of those responses and their expiration
times. Then the frequency of lookups is on the order of
the number of times a particular node is queried, modulo
the target node's TTL.
That's the only reliable option. It could also query SRV records on the hostname in the nodelist (the uderlying domain).
But then we
still would run into problem if people will start using TLS and check
for the validity of the cert.
Nodelist distribution over DNS is also not very decentralized anymore,
but a single point of failure.
Btw, do you know how (or
if) the option can be disabled for binkp at compilation? I find it more annoying than helpful.
Netsurge wrote to Avon <=-
If you need a tertiary dns server I would be glad to offer you up
space.
Avon wrote to tenser <=-
Forgive me I'm still learning all things DNS, so some of this is not
clear to me as I have not used some of the zone records to know them
well.
I picked up the O'Reilly book "DNS and BIND" by Cricket Liu, came in to work on a Saturday morning, and brought up a BSD box on a PC, then read the book from cover to cover. By noon, we had a responding DNS server.
Great book, I thoroughly recommend it to anyone wanting to run DNS.
Let me know if you need more mirrors here in Sweden. I have three public DNS servers.
On 02-20-20 18:42, Avon wrote to Vk3jed <=-
On 20 Feb 2020 at 11:41a, Vk3jed pondered and said...
Hmm, maybe we can all help each other out here with name servers, and I can also make my third party slave available. :)
If I can suss how to, I'd be happy to support you both...
On 02-20-20 08:39, Oli wrote to Vk3jed <=-
My biggest mistake (a long time ago) was using a CNAME on the naked
domain ...
On 02-20-20 06:55, poindexter FORTRAN wrote to Avon <=-
I picked up the O'Reilly book "DNS and BIND" by Cricket Liu, came in to work on a Saturday morning, and brought up a BSD box on a PC, then read the book from cover to cover. By noon, we had a responding DNS server.
Great book, I thoroughly recommend it to anyone wanting to run DNS.
On 02-21-20 12:01, tenser wrote to poindexter FORTRAN <=-
On 20 Feb 2020 at 06:55a, poindexter FORTRAN pondered and said...
I picked up the O'Reilly book "DNS and BIND" by Cricket Liu, came in to work on a Saturday morning, and brought up a BSD box on a PC, then read the book from cover to cover. By noon, we had a responding DNS server.
I remember that book. For that matter, I remember
BIND version 4. Ah, the halcyon days of my youth....
Things were so much simpler then.
tenser wrote to poindexter FORTRAN <=-
I remember that book. For that matter, I remember
BIND version 4. Ah, the halcyon days of my youth....
Things were so much simpler then.
My biggest mistake (a long time ago) was using a CNAME on the naked domain ...Hmm, what do you mean by that exactly?
tenser wrote to poindexter FORTRAN <=-
I remember that book. For that matter, I remember
BIND version 4. Ah, the halcyon days of my youth....
Things were so much simpler then.
I'm dating myself here, but I remember when people ran open SMTP servers as a courtesy.
I'm dating myself here, but I remember when people ran open SMTP servers as a courtesy.
On 02-20-20 08:39, Oli wrote to Vk3jed <=-
My biggest mistake (a long time ago) was using a CNAME on the naked
domain ...
Hmm, what do you mean by that exactly?
I'm dating myself here, but I remember when people ran open SMTP servers as a courtesy.
I'm dating myself here, but I remember when people ran open SMTP server as a courtesy.
I'm pretty fussy and don't think I could put up with dating myself, which is why I just have a series of one night stands with myself
I carry a picture of my right hand in my wallet.
I have to hide it from the left hand; there are some serious jealousy issues.
On 02-21-20 08:44, Phoobar wrote to Vk3jed <=-
My biggest mistake (a long time ago) was using a CNAME on the naked domain ...Hmm, what do you mean by that exactly?
Turn away...unless you are ready for something no one should see. ;)
Am in a decent mood right now...so I had to take it. ;)
On 02-21-20 18:15, Oli wrote to Vk3jed <=-
21 Feb 20 19:43, you wrote to me:
On 02-20-20 08:39, Oli wrote to Vk3jed <=-
My biggest mistake (a long time ago) was using a CNAME on the naked
domain ...
Hmm, what do you mean by that exactly?
something like
example.com CNAME example.gitlab.io
A good explanation why this doesn't work: https://www.isc.org/blogs/cname-at-the-apex-of-a-zone/
On 02-21-20 07:07, poindexter FORTRAN wrote to tenser <=-
tenser wrote to poindexter FORTRAN <=-
I remember that book. For that matter, I remember
BIND version 4. Ah, the halcyon days of my youth....
Things were so much simpler then.
I'm dating myself here, but I remember when people ran open SMTP
servers as a courtesy.
On 02-22-20 06:14, Avon wrote to poindexter FORTRAN <=-
On 21 Feb 2020 at 07:07a, poindexter FORTRAN pondered and said...
I'm dating myself here, but I remember when people ran open SMTP servers as a courtesy.
I hope the date goes well... oh wait a minute :)
Joking aside I think it's neat that we have such a nice mix of age and experience with respect to the early and later Internet eras along with those of us old enough to remember dial up times, the hey day of BBS
and earlier still.
I always enjoy your thoughtful posts. Thanks for writing/sharing them
:)
On 02-21-20 13:41, ryan wrote to poindexter FORTRAN <=-
I'm dating myself here, but I remember when people ran open SMTP servers as a courtesy.
I'm pretty fussy and don't think I could put up with dating myself,
which is why I just have a series of one night stands with myself
Hmm, what do you mean by that exactly?Can I point and giggle? :P
Turn away...unless you are ready for something no one should see. ;)
On 02-21-20 22:57, Phoobar wrote to Vk3jed <=-
Hmm, what do you mean by that exactly?Can I point and giggle? :P
Turn away...unless you are ready for something no one should see. ;)
One good thing...I have the electron magnifying glass racket for you to have something to point to & giggle.
Avon wrote to poindexter FORTRAN <=-
Joking aside I think it's neat that we have such a nice mix of age and experience with respect to the early and later Internet eras along with those of us old enough to remember dial up times, the hey day of BBS
and earlier still.
ryan wrote to poindexter FORTRAN <=-
I'm pretty fussy and don't think I could put up with dating myself,
which is why I just have a series of one night stands with myself
Atreyu wrote to Ryan <=-
I carry a picture of my right hand in my wallet.
I have to hide it from the left hand; there are some serious jealousy issues.
I've spent a fair amount of time with a bunch of
FTSC documents lately as I wrote my own binkp
implementation and am writing my own FTN<->NNTP
gateway software (binkd was too buggy and the
upstream maintainer isn't terribly responsive;
existing FTN<->NNTP gateways have bit rotted and
latency). Anyway, I think there's significant
room for improvement and innovation. It sounds
fun. It sounds experimental. It might be
much simpler in terms of implementation than
what's out there now.
Haha small targets? ;)Hmm, what do you mean by that exactly?Can I point and giggle? :P
Turn away...unless you are ready for something no one should see
One good thing...I have the electron magnifying glass racket for you have something to point to & giggle.
On 02-22-20 08:42, Phoobar wrote to Vk3jed <=-
Haha small targets? ;)Hmm, what do you mean by that exactly?Can I point and giggle? :P
Turn away...unless you are ready for something no one should see
One good thing...I have the electron magnifying glass racket for you have something to point to & giggle.
Always been true...nuke 'em till they glow & shoot 'em in the dark.
But the cool thing about FTN is that it just needs a plaintext file to connect nodes. No domain registrations that cost money and make some people rich. DNS is already a good target for censorship and stuff is happening that is outside the control of the user (sysop).
Yes, the decentralization thing is interesting, but there might be better solutions nowadays (distributed hash table, ...). I'm more interested in getting rid of the DNS dependency in FTN.
Always been true...nuke 'em till they glow & shoot 'em in the dark.Hahaha ouch. :D
On 02-24-20 16:20, Phoobar wrote to Vk3jed <=-
Always been true...nuke 'em till they glow & shoot 'em in the dark.Hahaha ouch. :D
The only bad thing is if we can make them glow red...rather than
blue...we will have an easier time finding them.
The only bad thing is if we can make them glow red...rather than blue...we will have an easier time finding them.Haha OK. :)
I believe that is my queue to say something ...
"Something" :)
(I'm not gonna quote the entire message chain here but bare with
me...)
One of the things I learnt while (foolishly, since it's like pouring gazoline on a fire trying to put out the fire) trying to help Ward
move, update and take control over fidonet.org and it's website is
this: hacks like updating DNS zonefiles from nodelists or vice-versa
are nice as long as anyone is interested in maintaining the whole
thing. In the case of fidonet.org someone (I can't remember who)
hosted the site and setup everything many many years ago and now,
when Ward came knocking, was happy to have someone else take it off
his hands but at the same time wasn't that interested in explaining
how stuff was setup.
What I'm trying to say is this: if possible, don't mess with a
working system. Nodelists work, let's keep it that way.
On 02-24-20 20:19, Phoobar wrote to Vk3jed <=-
The only bad thing is if we can make them glow red...rather than blue...we will have an easier time finding them.Haha OK. :)
Now...if we can teach them to ride the storm like Slim Pickens from Dr. Strangelove...be a real money maker.
On 02-25-20 12:50, Joacim Melin wrote to Joacim Melin <=-
Just so I'm being clear: I'm not discouraging any kind of new stuff, as long as
the old stuff still works as a fallback should things go south.
Now...if we can teach them to ride the storm like Slim Pickens from D Strangelove...be a real money maker.Since I haven't seen that, the reference is lost on me. :P
On 02-25-20 17:18, Phoobar wrote to Vk3jed <=-
Now...if we can teach them to ride the storm like Slim Pickens from D Strangelove...be a real money maker.Since I haven't seen that, the reference is lost on me. :P
It's a classic done by Stanley Kubric. Has George C Scott...Peter
Sellers doing 3 roles. Even has James Earl Jones in I believe his 1st role. Hell of a movie & more scary in times like these.
There's a conference room they built for the movie that the US military called them to ask about the plans of it. You will love it...but
remember it takes place in the early '60's.
Cool, sounds like one to watch.Strangelove...be a real money maker.It's a classic done by Stanley Kubric. Has George C Scott...Peter Sellers doing 3 roles. Even has James Earl Jones in I believe his 1st role. Hell of a movie & more scary in times like these.
On 02-26-20 14:50, Phoobar wrote to Vk3jed <=-
Be prepared for an escalating situation which goes more crazy the movie goes. The one about the race for caves/mines is one which had me rolling...but when you remember this is the way things work out.
Phoobar wrote to Vk3jed <=-
Be prepared for an escalating situation which goes more crazy the movie goes. The one about the race for caves/mines is one which had me rolling...but when you remember this is the way things work out.
DNow...if we can teach them to ride the storm like Slim Pickens from
Strangelove...be a real money maker.Since I haven't seen that, the reference is lost on me. :P
It's a classic done by Stanley Kubric. Has George C Scott...Peter
Sellers
doing 3 roles. Even has James Earl Jones in I believe his 1st role.
Hell of a
movie & more scary in times like these.
There's a conference room they built for the movie that the US
military
called them to ask about the plans of it. You will love it...but
remember it
takes place in the early '60's.
Mein Fuhrer! I can walk!
(mushroom cloud)
When Reagan took office in the 80s, one of the first things he demanded was to be shown the War Room. He thought it was all true since he saw it in Dr Strangelove.
Sysop: | echicken |
---|---|
Location: | Toronto, Ontario |
Users: | 2,224 |
Nodes: | 6 (0 / 6) |
Uptime: | 09:10:25 |
Calls: | 14,143 |
Files: | 295 |
Messages: | 551,261 |