> I'm not sure it wasn't premature to go production on, but since this is
> also an experimental network, they could just be giving it a try.
I agree it is all about experimentation and learning. Learning that adding
an IPv6 address may unexpectedly break some things is part of that, and it
is better to experience it here that at work on some client visible site :-)
Rob
> >/I get 404 too, and I am IPV6. /> Same here, looks like an issue with incorrect IPv6 DNS information?
Same here on my dual-stack network. It is probably a result of the recent changes
that allow registration of AAAA records in ampr.org, and people experimenting by
adding an AAAA record for their hostname to see if that is working.
Not a good idea to do that when it either does not point to the same system, or
the services on that system (like http/https) are not configured to actually handle
the IPv6 case. I have seen this problem before when I added IPv6 capability to
my own network and the proxy at work: even some of the "big guys" got this wrong
especially in the early days. AAAA record in DNS but IPv6 not actually working,
or only for some protocols and not for others. And often no monitoring for IPv6,
so nobody noticing.
Usually, when this became obvious a slightly different domain name was used for
IPv6 testing (e.g. adding a 6 or .ipv6. somewhere), and that is probably what
people testing IPv6 on .ampr.org should do as well.
Rob
> The reason for all this complication is that I've had to condense two
> machines, hamradio and amprgw, into one because hamradio is going away.
> The reason for using 'mailman' as the hostname in the mailing list headers
> is that we may someday move the mailing lists to their own machine with
> that name (and it's own IP address!)
It should not be that difficult to run a mailservice on the same machine but
with a different address and name. I.e. add 44.0.0.2 to the machine, move
mailman.ampr.org there, and run all SMTP services on that address and corresponding
name. Then the move to the new machine is transparent.
But otherwise, it is a good idea to use virtualization as much as possible.
This can be classical hardware virtualization like we use (VMware ESXi)
but there are also leaner methods. That way, you can keep the different
tasks on separate virtual machines, on separate addresses with their
corresponding hostname, and there is also a lot less impact when one
time you want to update the OS or switch to another OS.
(e.g. you could run the mailman on Linux and the gw on BSD, and maybe
experiment with a new gw running on Linux, all on the same hardware)
Rob
Hi,
I'm moving my 44net connection from an OpenWRT device to a BeagleBone Black, because I've got a TNCblack to connect directly to a radio.
However, in testing I'm finding the 44net connection doesn't stay up, max about 36hours before a reboot is required.
Any ideas why this is, or how I could diagnose further. The device is always accessible via the local WiFi connection.
Currently I've just got this tacked onto the end of my rc.local
#Start 44net
/usr/sbin/amprd -d -v 2> /var/log/amprd/amprd_err.log > /var/log/amprd/amprd_std.log &
sleep 30
ip route add default via 169.228.34.84 dev ampr0 onlink table default
ip rule add from 44.131.170.1 table default
ip rule add from 44.131.170.1 to 44.131.170.0/30 table main
sleep 10
ping -Iampr0 -c10 gw.ampr.org > /dev/null
Device is on m1bkf.ampr.org - sometimes!
Thanks. Bill (M1BKF)
--
Red to red, black to black, switch it on, but stand well back.
> Because of the high volume of traffic on the 44net list as of late,
> Jann suggested creating an announce-only moderated list. That list is
> now online at44-announce at mailman.ampr.org. <https://mailman.ampr.org/mailman/listinfo/44net> It's separate from this
> list, so subscribers to 44-announce only won't see all the traffic on
> the 44net list.
And how about the reverse? Will announcement messages be sent to 44net as
well or do we all have to subscribe to the announce list to still get them?
In that case, it may be better to subscribe everyone to 44-announce that is
now subscribed to 44net?
Rob
The host 'hamradio.ucsd.edu', which is the current home of the '44net'
mailing list, is planned to be decommissioned in the next few months.
If everything goes right, on September 19th (Tuesday) morning (Pacific
time), I will move this mailing list from hamradio.ucsd.edu to its
new home on mailman.ampr.org, which is currently an alias for amprgw.
After the move, all future submissions should be sent to
44net(a)mailman.ampr.org, and the outgoing mail will be sent from that
host too.
Please be prepared to update your address books and spam filters.
All subscriptions will remain the same.
The archives will move there too.
I'll set up forwarding on 'hamradio' so that postings, replies, and
such sent to the old address will be forwarded to the new address for
some time, until 'hamradio' is decommissioned.
This is supposed to be simple. Wish me luck!
- Brian
Because of the high volume of traffic on the 44net list as of late,
Jann suggested creating an announce-only moderated list. That list is
now online at 44-announce(a)mailman.ampr.org. It's separate from this
list, so subscribers to 44-announce only won't see all the traffic on
the 44net list.
https://mailman.ampr.org/mailman/listinfo/44-announce
All postings to the 44-announce list will be held for review and
approval by the list moderators, so there will be some delay in message
distribution. I anticipate very low volume on that list.
I've set 44-announce to be a public list. The list will show up in
the general mailing list index on the mailman.ampr.org machine, and
the archives of the list will be available publicly so it's likely that
various search engines will index them. I've had significant problems
resulting from this policy in the past, but those were non-moderated lists
so maybe this will work better. The list of subscribers is not public.
Note that whether a list is public or private, fake names can subscribe
and harvest poster's addresses from postings and use them for spam.
Not much I can do about that if we're to still have the option of replying
to the individual poster.
It's a new list, so please be patient while I tune the parameters.
- Brian
I like the MIME version of the digest, but I am seeking advice on how to
add an option that I need.
On the "digest" version, when I turn on "Display Attachments Inline",
the emails appear looking like this:
... but in order to reply to an individual message, I have to find the
same email among the .eml files in the "Attachments" menu, and use that
to preserve the threading. I'd like to have a button I could click to
reply to an individual message from the screen shown above. I hope
someone else on the list has cured this in the past few days.
Bill, W4EWH
I like the MIME version of the digest, but I am seeking advice on how to
add an option that I need.
On the "digest" version, when I turn on "Display Attachments Inline",
the emails appear looking like this:
... but the .eml files which _are_ the individual emails do not appear
in the "attachments" bar unless I turn the option to display them in
line off, and there doesn't seem to be any way to reply to the
individual messages displayed as part of the digest, since when I click
the "reply" or "Reply to List" choices, I get a draft email which is a
copy of the digest, not the individual email I want to reply to.
The question is, then, how I can add (or turn on) an option to give me a
clickable link to reply to an individual post without having to turn off
the "display attachments inline" option.
Sorry to clutter the list: I hope someone else has come across this in
the past few days.
73,
Bill, W4EWH
Bill, W4EWH
> It's a single mbox format file that is broken up into individual messages
> which are stored in directories by date. A script to feed them into an
> NNTP posting program. There are currently over 10,000 such messages.
Well OK, it is not the kind of data volume that you could not afford to store in two different native formats.
(given today's size of disk drives etc)
> There are difficulties.
> The Mailman authentication data is stored as a Python 'pickle' (compressed
> dictionary file). There is a program to export a list of users, but
> it does not permit querying a single specific user, nor does it export
> passwords.
> NNTP has a Perl hook for authentication. Someone would have to write a
> Python program that could extract the necessary per-user data from the
> Python dictionary, and be called from the Perl NNTP authenticator.
Again a case where it would be so convenient to have a central store of authentication info that
could be queries by many different services...
E.g. when one of the APIs would be LDAP (as someone already suggested), it would be really
simple to write a Perl hook that does an LDAP query.
> Across the Atlantic. The NNTP server is in England, the Mailman mailing
> list is in San Diego.
My suggestion was to run INN on the gw server that now runs mailman, and keep it local.
You could run a couple of servers around the world, but of course each of them would
be facing the authentication issue. There should not be yet another "register here
for access" thing.
Rob
