NNTP access problem: 119: Connection timed out...

Hello. First let me say that I love the existence of the NNTP gateway to
the forum. I used to prefer mailing lists over forums because I liked the
clean text interface I get with alpine better than anything a web browser
can give me. But nowadays, most of the time when there exist both a mailing list
and a forum for anything technical, mailing lists tend to be just a little bit
less friendly than the forum.

Having an NNTP gateway to a user forum such as currently provided by
nntp.opensuze.org, is the best of both worlds… When it works.

I can’t really say when it started, I had to reinstall my Sabayon
installation due to a bad system upgrade. And another killed my Xubuntu,
which I decided to replace with Mageia. Both of which took my attention
away from other interests for a while. It’s only been this week that I
returned to the opensuze forum for the first time in months.

Now I access nntp.opensuze.org directly from alpine. It used to work sweet.
But now I’m finding that when I select nntp.opensuze.org from ALPINE 2.00’s
COLLECTION LIST screen, I’m seeing:

Code:

[Fetching folder data]

On the status line for a lot longer than I should. That at least the first
time I try, {And sometimes for the first two or three attempts} I get:

Code:

[Connection failed to nntp.opensuse.org,119: Connection timed out]

At which point I use the keyboard command keys “<” $ “>” to back out to the
COLLECTION LIST screen, and reopen the collection list. When I finally see
my subscribed subforums {including this one} they open so fast that I don’t
even see the " [Fetching folder data] " message. Then if I’m quick about it
I can read some messages. better results can be had by quickly saving all
new messages to a local folder to read later. But either way when I back
out to the list to select the next subforum, I’m usually having to reopen
the collection list {see above} first…

Then when it comes to posting a message to any of the subforums it didn’t
used to matter if the nntp.opensuze.org collection was open or not. As long
as that was the currently selected news server in the .pinerc, any properly
configured message containing {for example} these headers:

Code:

Newsgroups: opensuse.org.feedback.forums.support-information
From: JtWdyP <jtwdyp@no-mx.forums.opensuse.org>

would be posted even if the currently open folder was either a local one or
an IMAP folder.

But now, when I’m done composing, it behooves me to suspend it to the
postponed folder, then reopen the nntp.opensuse.org Collection {see above}
again. And as soon as I can see the list of my subscribed subforums, I need
to quickly select to compose from postponed, select the message and use
crtl+x to begin the “send message dialog”… I do have to be quick, If I
delay, the posting will fail, and I’ll need to postpone it again so that I
can once again tell alpine to open the nntp collection again…

I don’t think the problem is alpine as it used to work properly. And still
does when the nntp collection is set to gmane’s nntp mailing list mirror.

Also this occurs with nntp.opensuze.org now, regardless of which of my
installed Linux, I’m running Or on which of two computers I’m running it
on.

Is this part of a known problem? Or a new symptom. And is anybody else
having the same problem??


JtWdyP

I use NNTP as well across several domains and have never seen this (the
openSUSE forums are among the most active in terms of message count).
Since you have your game figured out (other thread) you obviously know
your way around Linux a little… could you do some troubleshooting with
us to identify the issue? A few questions to start:

Wired or wireless connection? If wireless, can you try wired? FYI: My
connection is wired; I have more issues when wireless at the other end of
my place, but that’s to be expected.

Second, do you have IPv6 enabled on your machine? Have you tried
disabling and rebooting, unless you really need it? nntp.opensuse.org
does not have an AAAA (IPv6) DNS record from what I can tell, so if your
system tries that first it may timeout before trying the A (IPv4) DNS
record. Shouldn’t work that way, but I don’t use clients other than
Thunderbird, or other Linux distributions, for NNTP.

It sounds like this happens regularly. Can you catch it in a LAN trace?

Code:

sudo /usr/sbin/tcpdump -n -s 0 -i any -w /tmp/nntp0.cap port 119

Let it run until you see the symptom and work past it. Hit ctrl+c to
cancel the trace and then post the resulting file somewhere after compressing:

Code:

bzip2 /tmp/nntp0.cap

If you cannot post it anywhere else, like a Dropbox share or something,
let us know and we’ll try something like ftp://ftp.novell.com/incoming and
see if somehow we can grab the file from there (it’s a place where files,
once uploaded, can only be downloaded by Novell employees, so posting the
name of the file here won’t risk anybody getting it unless they’re
looking; Novell is affiliated with openSUSE, just so you know why their
FTP site is mentioned).

Otherwise there are always fun little network tests assuming you have
‘netcat’ installed:

Code:

while 1 ] ; do date; netcat -zv nntp.opensuse.org 119; sleep 5; done

Feel free to run that in a shell somewhere and see if you ever get any
failures connecting. If not, maybe the timeout is because of the amount
of data coming down from the NNTP server when you get new messages, but
that seems pretty unlikely too since you should only be getting what’s
new, and that’s not going to be measured in megabytes unless you’re out
for quite a while.

Good luck.

It would appear that on Jul 13, ab did say:

> I use NNTP as well across several domains and have never seen this (the
> openSUSE forums are among the most active in terms of message count).
> Since you have your game figured out (other thread) you obviously know
> your way around Linux a little… could you do some troubleshooting with
> us to identify the issue? A few questions to start:

A little… maybe some, If I can figure out what you want…

> Wired or wireless connection? If wireless, can you try wired? FYI: My
> connection is wired; I have more issues when wireless at the other end of
> my place, but that’s to be expected.

Both, My laptop is connected via wireless. But I have same problem with my
desktop which is connected to the router via Ethernet cable…

> Second, do you have IPv6 enabled on your machine? Have you tried
> disabling and rebooting, unless you really need it? nntp.opensuse.org
> does not have an AAAA (IPv6) DNS record from what I can tell, so if your
> system tries that first it may timeout before trying the A (IPv4) DNS
> record. Shouldn’t work that way, but I don’t use clients other than
> Thunderbird, or other Linux distributions, for NNTP.

I’m not sure… I’ve seen this happening with at least three of the Linux
installed to my desktop system. But lets focus on the opensuze installation.
I think opensuze may enable IPv6 by default. If you can tell me the cli
command to disable it, I’ll test that from the desktop. NOTE, I’m not using
nether kde nor gnome, I prefer e17… So if possible make that cli something
that works from a console (runlevel3 aka multi-user.target)…

> It sounds like this happens regularly. Can you catch it in a LAN trace?
>
> Code:
> ----------
> sudo /usr/sbin/tcpdump -n -s 0 -i any -w /tmp/nntp0.cap port 119
> ----------
>
> Let it run until you see the symptom and work past it. Hit ctrl+c to
> cancel the trace and then post the resulting file somewhere after compressing:

I’m guessing that I should run that from a different console or xterm than I’m
going to run alpine in? Then when I get past the first 119 timeout well enough
to open a forum message, I would switch back to the console or xterm where I
ran the above command before I hit ctrl+c

> Code:
> ----------
> bzip2 /tmp/nntp0.cap
> ----------
>
> If you cannot post it anywhere else, like a Dropbox share or something,
> let us know and we’ll try something like ftp://ftp.novell.com/incoming and
> see if somehow we can grab the file from there (it’s a place where files,
> once uploaded, can only be downloaded by Novell employees, so posting the
> name of the file here won’t risk anybody getting it unless they’re
> looking; Novell is affiliated with openSUSE, just so you know why their
> FTP site is mentioned).

Sounds like you think the data in the nntp0.cap file could be sensitive?

So since my only option would be to upload it to my ISP provided web-space,
which would mean that anybody who read the message with the link could access
the file, perhaps I should hear more about this ftp://ftp.novell.com/incoming
option…

When I check it out with my favorite ftp client I can’t navigate to any of the
subdirectories, not even pub. nor can I view the readme…

http://i43.tinypic.com/333bdkj.png

> Otherwise there are always fun little network tests assuming you have
> ‘netcat’ installed:
>
> Code:
> ----------
> while 1 ] ; do date; netcat -zv nntp.opensuse.org 119; sleep 5; done
> ----------
>
> Feel free to run that in a shell somewhere and see if you ever get any
> failures connecting. If not, maybe the timeout is because of the amount
> of data coming down from the NNTP server when you get new messages, but
> that seems pretty unlikely too since you should only be getting what’s
> new, and that’s not going to be measured in megabytes unless you’re out
> for quite a while.

I assume netcat is in the repos…

Should all of these tests wait till I’ve disabled IPv6???

Be advised:
It might take me a couple days to buy enough 'puter time from my lady.
{She does get jealous of the time I spend on them…} And she’s already heated
with me for spending the time with that [other thread] amoebax thing…


JtWdyP

On 07/13/2013 03:39 AM, JtWdyP wrote:
>> Second, do you have IPv6 enabled on your machine? Have you tried
>> disabling and rebooting, unless you really need it? nntp.opensuse.org
>> does not have an AAAA (IPv6) DNS record from what I can tell, so if your
>> system tries that first it may timeout before trying the A (IPv4) DNS
>> record. Shouldn’t work that way, but I don’t use clients other than
>> Thunderbird, or other Linux distributions, for NNTP.
>
> I’m not sure… I’ve seen this happening with at least three of the Linux
> installed to my desktop system. But lets focus on the opensuze installation.
> I think opensuze may enable IPv6 by default. If you can tell me the cli
> command to disable it, I’ll test that from the desktop. NOTE, I’m not using
> nether kde nor gnome, I prefer e17… So if possible make that cli something
> that works from a console (runlevel3 aka multi-user.target)…

Yes, all commands work from anywhere. To disable IPv6 run yast (from the
command line or GUI) and go to Network Settings and modify your network
card settings.

Code:

sudo /sbin/yast lan

Under Global Options uncheck ‘Enable IPv6’ and then reboot after
saving/exiting.

>> It sounds like this happens regularly. Can you catch it in a LAN trace?
>>
>> Code:
>> ----------
>> sudo /usr/sbin/tcpdump -n -s 0 -i any -w /tmp/nntp0.cap port 119
>> ----------
>>
>> Let it run until you see the symptom and work past it. Hit ctrl+c to
>> cancel the trace and then post the resulting file somewhere after compressing:
>
> I’m guessing that I should run that from a different console or xterm than I’m
> going to run alpine in? Then when I get past the first 119 timeout well enough
> to open a forum message, I would switch back to the console or xterm where I
> ran the above command before I hit ctrl+c

If you can get the command there you could run this in a TTY (Ctrl+Alt+F2)
so that it’s always there and out of the way. Go back to your desktop
(Ctrl+Alt+F7) and work until the issue happens, then finish duplicating
and go back to the TTY and cancel the command. You could also do this
from any xterm or equivalent.

>> Code:
>> ----------
>> bzip2 /tmp/nntp0.cap
>> ----------
>>
>> If you cannot post it anywhere else, like a Dropbox share or something,
>> let us know and we’ll try something like ftp://ftp.novell.com/incoming and
>> see if somehow we can grab the file from there (it’s a place where files,
>> once uploaded, can only be downloaded by Novell employees, so posting the
>> name of the file here won’t risk anybody getting it unless they’re
>> looking; Novell is affiliated with openSUSE, just so you know why their
>> FTP site is mentioned).
>
> Sounds like you think the data in the nntp0.cap file could be sensitive?

The data should not be sensitive since the forums are public. Some people
are paranoid about data they send out, so this option works. Basically
point an FTP client of some sort to the server mentioned and upload the
file, or use your own. Limited to port 119 nothing at all sensitive
should show up. I’d be fine pulling from your server. Hopefully the
problem isn’t DNS or we won’t catch it with that command, but you could
try to get both:

Code:

sudo /usr/sbin/tcpdump -n -s 0 -i any -w /tmp/nntp0.cap port 119 or port 53

> When I check it out with my favorite ftp client I can’t navigate to any of the
> subdirectories, not even pub. nor can I view the readme…
>
> http://i43.tinypic.com/333bdkj.png

“Helpful” FTP clients often fail on this site because of the restrictions
imposed. If you use the command line it works just fine:

Code:

ftp -a ftp.novell.com #-a automates an anonymous login
binary
cd incoming
put /tmp/nntp0.cap.bz2
bye

> I assume netcat is in the repos…

Yes, indeed. It may be called netcat-openbsd or something, but you can
find it with one of the following:

Code:

zypper se netcat
zypper se nc

> Should all of these tests wait till I’ve disabled IPv6???

I would, since IPv6 is a suspect and not something you’re likely
needing/using.

> Be advised:
> It might take me a couple days to buy enough 'puter time from my lady.
> {She does get jealous of the time I spend on them…} And she’s already heated
> with me for spending the time with that [other thread] amoebax thing…

She’s more important… take your time.

Good luck.

On 2013-07-13 11:39, JtWdyP wrote:
> I’m guessing that I should run that from a different console or xterm than I’m
> going to run alpine in? Then when I get past the first 119 timeout well enough
> to open a forum message, I would switch back to the console or xterm where I
> ran the above command before I hit ctrl+c

You are using alpine for nntp? wow. :-o

I use alpine for email, but not for nntp because they do it wrong:
alpine does not caches any message locallly, and does not keep record of
what posts are already read and which not. I have not found a good
enough text mode nntp client.

It means that alpine is more sensitive than other clients to network
problems.

Anyway, I think you should try any other nntp client to find out if they
exhibit the same problems or not.

Alternatively, you could use leafnode, as a local nntp proxy server. I
use it.


Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

On Fri, 12 Jul 2013 23:48:32 +0000, JtWdyP wrote:

> nntp.opensuze.org

Obviously, this isn’t the address you’re using - or is it?

You might look at using tcptraceroute to see if there’s something between
you and the server dropping packets. Been working fine here for a very
long time.

Jim

Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

Jim Henderson wrote:

>> nntp.opensuze.org
>

Uh…?

On 2013-07-14 05:45, caf4926 wrote:
> Jim Henderson wrote:
>
>>> nntp.opensuze.org
>>
>
> Uh…?

Look at the code section, the server name is correct there.


Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

It would appear that on Jul 13, ab did say:

> Code:
> ----------
> sudo /sbin/yast lan
> ----------
>
> Under Global Options uncheck ‘Enable IPv6’ and then reboot after
> saving/exiting.

Done…

> >> It sounds like this happens regularly. Can you catch it in a LAN trace?

BTW: Very regularly… Like every time lately.

>
> Code:
> ----------
> sudo /usr/sbin/tcpdump -n -s 0 -i any -w /tmp/nntp0.cap port 119 or port 53
> ----------

Done

> “Helpful” FTP clients often fail on this site because of the restrictions
> imposed. If you use the command line it works just fine:

For chucks and giggles, since mc appears to be able to get to
ftp.novell.com/incoming it’s just that once there it can’t get any ftp side
ls data to update it’s file list with etc.

I tried using it’s twin panel method of putting the file:
jtwdyp-nntp0.cap.bz2
But I had no feedback that it was successful so renamed the file:
jtwdyp_nntp0.cap.bz2
And did it this way:
> Code:
> ----------
> ftp -a ftp.novell.com #-a automates an anonymous login
> binary
> cd incoming
> put /tmp/nntp0.cap.bz2
> bye
> ----------

Got errors when I tried to absolute pathname reference in the put line…
So I repeated the ftp sequence from the directory containing the renamed
file…

150 Opening BINARY mode data connection for jtwdyp_nntp0.cap.bz2
100% |******************************************************| 519 KB
72.78 KB/s 00:00 ETA
226 Transfer complete

But I’d be curious to know if both the:
jtwdyp_nntp0.cap.bz2
And the:
jtwdyp-nntp0.cap.bz2
copies of the file got there…

The problem is definately related to how alpine talks to nntp servers… as
I can’t get a single failed connection out of netcat…

Any way to find out exactly where in the dialog alpine is screwing up so
that I can try to talk to the people who currently work on it…

Or, if it turns out that nntp.opensuse.org’s server is being overly picky
with what should be a legal communication dialog (according to the so called
{RFCs???}) could the nntp server maybe get tweaked to make it tolerate alpine?


JtWdyP

It would appear that on Jul 13, Carlos E. R. did say:

> You are using alpine for nntp? wow. :-o
>
> I use alpine for email, but not for nntp because they do it wrong:
> alpine does not caches any message locallly, and does not keep record of
> what posts are already read and which not. I have not found a good
> enough text mode nntp client.

You mean remember the difference between read messages and expunged ones the
next time the nttp “folder” is opened right? Your right that alpine doesn’t
keep a database for that. The nntp servers certainly are not going to hold
they kind of flag data for each and every user on the server.
I “save” the “interesting” posts to a local folder where alpine does track
that data because with local and/or imap folders such data is saved in the
folder with the message.

> It means that alpine is more sensitive than other clients to network
> problems.

That I’ve noticed. I suspect that the alpine team has always insisted on
following the letter of what they call {is it "RFC"s?} rather than conforming
to any defacto standards commonly in place with the servers it interacts with.
Still, I agree they could/should have done a better job on the topic of silently
reconnecting to a server, when something hiccups. That and when downloading
a chunk of messages (say 100 for example) where the index data didn’t have the
right content length for a few of them, including zero length deleted articles,
it would be nice if alpine would continue processing the rest of the
messages…

But that said, while Alpine has always been sensitive to time out on long
periods of inactivity, (like while composing a reply) I never had trouble
just opening nntp.opensuze.org with it before… Unless you count the short
period when I didn’t know I needed to stop using forums.opensuze.org

And Gmane’s nntp mailing list gateway is STILL working as expected. Which
for my purposes is good enough.

> Anyway, I think you should try any other nntp client to find out if they
> exhibit the same problems or not.

I think that any client that does try to track the read state for messages kept ON
an nntp would have to do so in a separate {local?} file or internal program data.
Right? In which case I don’t think it would be suitable for my needs where I
sometimes find a reason ( in the middle of composing a message) to suddenly
postpone/close any open edits/files and reboot into one of the other 4 Linux
currently installed on my PC, and then reopen said files/compositions from there.
When that happens I need the mail/news client on say opensuse to have the same
message flags and/or expunged state, as I had on say Sabayon a couple minutes ago.

I also don’t want to deal with the {to me} horrid gui user interface I’ve found
with any of the gui clients I’ve ever looked at.

> Alternatively, you could use leafnode, as a local nntp proxy server. I
> use it.

Now THAT would be a possibility. However…

I’ve never been interested in maintaining a true local news spool on my Linux. {If
I went there for my desktop I’d have to do it for all 5 distros.} And
somehow all 5 would have to share the same spool data. Perhaps in similar
fashion to how alpine on all 5 distro share the same local mail folders via

Code:

ln - s ~/mail /path/to/personal/data/partition/maildir

I also sync my assorted newsrc files {also on data partition} with simlinks…

I did look at a wiki on leafnode, it implied that I’d need to bother with
setting up a fully qualified domain name. Which I don’t want, and otherwise
it looked like too much work to get it all set up for what little use I’d
get from it… 99% of the time my only use for nntp is gmane & opensuse.
Alpine doesn’t have any trouble with gmane. And I no how to make it
connect to opensuse. It just takes more effort than it should.

But thanks for the suggestions.


JtWdyP

It would appear that on Jul 14, Jim Henderson did say:

> On Fri, 12 Jul 2013 23:48:32 +0000, JtWdyP wrote:
>
> > nntp.opensuze.org
>
> Obviously, this isn’t the address you’re using - or is it?

It would appear that on Jul 14, caf4926 did say:

> >> nntp.opensuze.org
> >
>
> Uh…?

It would appear that on Jul 14, Carlos E. R. did say:

> Look at the code section, the server name is correct there.

And that’s because I used the clipboard to copy the actual server name from
my .pinerc to the code section. Az to the other ztringz I plead inzanity. :wink:

It would appear that on Jul 14, Jim Henderson continued to speak:

> You might look at using tcptraceroute to see if there’s something between
> you and the server dropping packets. Been working fine here for a very
> long time.

Maybe… But this is definitely an alpine problem because I’ve yet to have
a bad connection attempt with netcat… <sigh>


JtWdyP

On Sun, 14 Jul 2013 16:07:16 +0000, JtWdyP wrote:

>> You might look at using tcptraceroute to see if there’s something
>> between you and the server dropping packets. Been working fine here
>> for a very long time.
>
> Maybe… But this is definitely an alpine problem because I’ve yet to
> have a bad connection attempt with netcat… <sigh>

Sounds like you’re making progress, and may have found a bug with the
software you’re using. :slight_smile:

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On 2013-07-14 18:02, JtWdyP wrote:
>
> It would appear that on Jul 13, Carlos E. R. did say:
>
>> You are using alpine for nntp? wow. :-o
>>
>> I use alpine for email, but not for nntp because they do it wrong:
>> alpine does not caches any message locallly, and does not keep record of
>> what posts are already read and which not. I have not found a good
>> enough text mode nntp client.
>
> You mean remember the difference between read messages and expunged ones the
> next time the nttp “folder” is opened right? Your right that alpine doesn’t
> keep a database for that. The nntp servers certainly are not going to hold
> they kind of flag data for each and every user on the server.
> I “save” the “interesting” posts to a local folder where alpine does track
> that data because with local and/or imap folders such data is saved in the
> folder with the message.

I don’t expunge nntp posts. I like to go back and see the history of a
thread. I use thunderbird which keeps a local copy of the nntp headers
(not the bodies) for this. All nntp clients I tried, except alpine, do
this, they keep some kind of local cache. And that way they keep
automatically track of what you read.

The way alpine does with imap servers is fine, because the flags are
kept remotely. Then they do the same with nntp, not keep any local
storage, and they insists that’s the way to do it. I say it is wrong.

>> Anyway, I think you should try any other nntp client to find out if they
>> exhibit the same problems or not.
>
> I think that any client that does try to track the read state for messages kept ON
> an nntp would have to do so in a separate {local?} file or internal program data.
> Right?

Of course.

> In which case I don’t think it would be suitable for my needs where I
> sometimes find a reason ( in the middle of composing a message) to suddenly
> postpone/close any open edits/files and reboot into one of the other 4 Linux
> currently installed on my PC, and then reopen said files/compositions from there.
> When that happens I need the mail/news client on say opensuse to have the same
> message flags and/or expunged state, as I had on say Sabayon a couple minutes ago.

You mean that you stop composing in Alpine, and take it over in another
Linux? That can only happen if the postponed local folder is the same
across all your installs. Of course you can do that with other clients,
if you share that folder.

> I also don’t want to deal with the {to me} horrid gui user interface I’ve found
> with any of the gui clients I’ve ever looked at.

Well, I use alpine all the time, but I find the interface of the text
nntp clients I have tried awkward.

>> Alternatively, you could use leafnode, as a local nntp proxy server. I
>> use it.
>
> Now THAT would be a possibility. However…
>
> I’ve never been interested in maintaining a true local news spool on my Linux. {If

It is not a “true” news spool, it is rather a news proxy.

> I went there for my desktop I’d have to do it for all 5 distros.} And
> somehow all 5 would have to share the same spool data. Perhaps in similar
> fashion to how alpine on all 5 distro share the same local mail folders via

Or a dedicated server machine :slight_smile:

> I also sync my assorted newsrc files {also on data partition} with simlinks…
>
> I did look at a wiki on leafnode, it implied that I’d need to bother with
> setting up a fully qualified domain name.

Why? I don’t have one. If you say that “telcontar.valinor” is a fully
qualified name, then so it is. :slight_smile:

> Which I don’t want, and otherwise
> it looked like too much work to get it all set up for what little use I’d
> get from it… 99% of the time my only use for nntp is gmane & opensuse.
> Alpine doesn’t have any trouble with gmane. And I no how to make it
> connect to opensuse. It just takes more effort than it should.

It is little work, and once done, it is blazingly fast, posts light up
(even on alpine) in a split second. With nntp.opensuse.org I can count
the seconds sometimes.

Look, this is my entire config:


cer@Telcontar:~> cat /etc/leafnode/config | egrep -v "^:space:]]*$|^#"
expire = 400
server = nntp.opensuse.org
server = nntp.novell.com
timeout = 30
groupexpire opensuse.org.help.howto = -1
groupexpire opensuse.org.help.howto.submissions = -1
groupexpire opensuse.org.* = 600
initialfetch = 1000
debugmode = 1
maxage = 0
maxbytes = 500000
timeout_long = 60
allow_8bit_headers = 1
cer@Telcontar:~>

Do you really think /that/ is difficult? :slight_smile:

The only snag is that it uses one file per post, and that’s a huge lot
of inodes - thus I make sure that /var/spool/news is on reiserfs.

Then you need a cronjob to do the hard work:


> -3,8,13,18,23,28,33,38,43,48,53,58 * * * * news /var/lib/news/bin/cronscriptparanovell


>
> cer@Telcontar:~> cat /var/lib/news/bin/cronscriptparanovell
> #!/bin/bash
> date --rfc-3339=seconds          >> /var/lib/news/fetchnews.cron.log 2>&1
> /usr/sbin/fetchnews -v           >> /var/lib/news/fetchnews.cron.log 2>&1
> date --rfc-3339=seconds          >> /var/lib/news/fetchnews.cron.log 2>&1
> echo -e "------------
">> /var/lib/news/fetchnews.cron.log 2>&1
> cer@Telcontar:~>

That’s all. And you can do without that script, just call fetchnews from
the cron.


Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

On 07/14/2013 09:13 AM, JtWdyP wrote:
>>>> It sounds like this happens regularly. Can you catch it in a LAN trace?
>
> BTW: Very regularly… Like every time lately.

Bleh… you’re alone (or nearly) in this symptom, so let’s figure out why.

>> “Helpful” FTP clients often fail on this site because of the restrictions
>> imposed. If you use the command line it works just fine:
>
> For chucks and giggles, since mc appears to be able to get to
> ftp.novell.com/incoming it’s just that once there it can’t get any ftp side
> ls data to update it’s file list with etc.
>
> I tried using it’s twin panel method of putting the file:
> jtwdyp-nntp0.cap.bz2
> But I had no feedback that it was successful so renamed the file:
> jtwdyp_nntp0.cap.bz2
> And did it this way:
>> Code:
>> ----------
>> ftp -a ftp.novell.com #-a automates an anonymous login
>> binary
>> cd incoming
>> put /tmp/nntp0.cap.bz2
>> bye
>> ----------
>
> Got errors when I tried to absolute pathname reference in the put line…
> So I repeated the ftp sequence from the directory containing the renamed
> file…
>
> 150 Opening BINARY mode data connection for jtwdyp_nntp0.cap.bz2
> 100% |******************************************************| 519 KB
> 72.78 KB/s 00:00 ETA
> 226 Transfer complete
>
> But I’d be curious to know if both the:
> jtwdyp_nntp0.cap.bz2
> And the:
> jtwdyp-nntp0.cap.bz2
> copies of the file got there…

Yes, both made it with matching md5 checksums so they both worked
perfectly. The tricky thing about not allowing listings (or at least
returning nothing) is that it fouls overly-clever clients up.
Overly-clever clients may upload and then try to check to ensure that the
file is there and is the right size. When that is disabled (even though
the file is there) those clients have a fit. Oh well… it worked anyway.

> The problem is definately related to how alpine talks to nntp servers… as
> I can’t get a single failed connection out of netcat…

Yes, I think it is largely Alpine’s fault, though you also have some
network packet drops that are a bit weird. That you see this with a wired
connection makes me wonder if this could somehow be a problem further
upstream with your ISP. That definitely shouldn’t happen, but there you
go. The reason I think this could be the case has to do with what happens
when packets are dropped in your trace. TCP retries things… if one
packet doesn’t get through then TCP tries again very quickly (tiny
fraction of a second) and then tries again after doubling that, then
doubling that… eventually taking multiple seconds between retries. None
of those show up in your LAN trace. Where did they go? Maybe the NNTP
server’s TCP implementation does not do that, but that’s insanely
unlikely; I’d bet your computer against it, in fact.

> Any way to find out exactly where in the dialog alpine is screwing up so
> that I can try to talk to the people who currently work on it…

Biggest weird thing that has stood out so far is that Alpine is doing some
really crazy stuff with regard to getting a list of messages fro ma forum.
Most of the clients I’ve used (all of which cache things) ask for the
latest message number, compare with the local copy, and then move on.
Alpine appears to be requesting a listing (index number) of every single
message in a forum. This does not make sense without some kind of
caching, so I think Alpine must, maybe only at runtime, be caching things.
Why list them all if you are not going to either pull them all (not
seeing that in this trace) or pull the ones that are new that you (the
client) do not know about (requiring caching at some level)?

At the end of one pull of indexes Alpine grabs the over information
(XOVER) of what appear to be the latest messages (again, how does the
client know this? must be caching):

Code:

XOVER 162610-162626


After this it grabs the headers and bodies of the same messages. Looking
a bit more, both of the big long lists of message indexes (from LISTGROUP
NNTP commnd) came from the same group (the .applications group here, which
is likely the busiest group here) and it happened twice, serially. The
first time it made the request was at 12:19:43.912495. The second time it
made the same request was at 12:19:48.544028, and that one ended at about
12:19:52.178064. In between this start/end about 900 full-size packets go
from the server to the client meaning about 1.3 MB data flowed when about
500 bytes should have sufficed (if that).

> Or, if it turns out that nntp.opensuse.org’s server is being overly picky
> with what should be a legal communication dialog (according to the so called
> {RFCs???}) could the nntp server maybe get tweaked to make it tolerate alpine?

I’m not seeing anything strange here, other than Alpine’s commands and
some dropped packets at the beginning. The dropped packets, I would bet,
are your network or ISP’s fault, but to really know we would need to see a
LAN trace from the server side, and maybe from somewhere in the middle.
For example if for some reason some device between your client application
and the NNTP application did not receive the packets properly from your
NNTP client then we’d see that in your LAN trace but not the server trace.
If a problem happened sending data back we’d see it in the server LAN
trace but not in yours. Usually having the traces on both sides taken at
the same time will find a problem pretty well, though in some cases
something in between (load balancer, for eaxmple… or a firewall or
router) will cause issues. Without access to the server I cannot get the
server-side trace; however, since none of us see the same issue as you,
it’s easy to point fingers at you with a little bit of a taunt in our
jeering. :wink:

Still, you said that you do not see this with other forums. This has also
been the claim of others in the past, though not for the reason you may
think. The Novell/NetIQ/SUSE/openSUSE forums are some of the largest
NNTP-based forums online. As a result, issues that would not plague a
site with a couple of groups with a couple of posts per day can show up
because of the size/scale of the site here. How many NNTP forums have
162,000+ posts? The applications forum does, and others here also have
high numbers. As a result, clients that do well on small sites often find
their weaknesses when hitting the bigger site here. Bigger environments
always cause bigger issues, and that’s okay and will hopefully help the
Alpine folks find a more-efficient way to do what they do. Perhaps they
have chosen to get a list of all messages for a reason, for example to
ensure that if post numbers are reused then something will not be missed.
I read the RFC once on the possibility of that and cannot remember what
it wrote, but if Alpine folks want to jump in here we can discuss is.

For now, see if Alpine can be changed to NOT list all post’s indexes with
every update of the forum because that’s not helping. Testing another
newsreader (several of us like Thunderbird, but I know people who like Pan
or Xananews… whatever) or from another location (I know, netcat is
working, but the error is at layer seven, not layer three, so netcat is
correct but maybe not showing enough if your ISP is doing any
dropping/filtering based on layer seven stuff) that may be interesting.

Good luck.

It would appear that on Jul 14, Carlos E. R. did say:

> You mean that you stop composing in Alpine, and take it over in another
> Linux? That can only happen if the postponed local folder is the same
> across all your installs.

Yeah, the "local folder is ~mail/ which is a symlink to actual mail directory
on my personal data partition.

> On 2013-07-14 18:02, JtWdyP wrote:

> > I also don’t want to deal with the {to me} horrid gui user interface I’ve found
> > with any of the gui clients I’ve ever looked at.
>
> Well, I use alpine all the time, but I find the interface of the text
> nntp clients I have tried awkward.

As long as you don’t include alpine in your list of text nntp clients, I’d
likely agree.

To me checking any subscribed newsgroups are part of what I think of as my
email chores. I really like to use the same interface for read/compose
on both…

> >> Alternatively, you could use leafnode, as a local nntp proxy server. I
> >> use it.
> >
> > Now THAT would be a possibility. However…
>
> It is not a “true” news spool, it is rather a news proxy.

Not sure I understand the difference… But I see references to something
called fetchnews…

I use a script to retrieve most of my email via fetchmail using procmail as an
mda… would you describe that as a mail proxy?

> > I did look at a wiki on leafnode, it implied that I’d need to bother with
> > setting up a fully qualified domain name.
>
> Why? I don’t have one. If you say that “telcontar.valinor” is a fully
> qualified name, then so it is. :slight_smile:
>
> > it looked like too much work to get it all set up for what little use I’d
> > get from it…
> It is little work, and once done, it is blazingly fast, posts light up
> (even on alpine) in a split second. With nntp.opensuse.org I can count
> the seconds sometimes.

So you can use alpine with it… What about posting a message composed in
alpine?? Is it as simple as selecting the local news spool in the .pinerc
and then putting the appropriate newsgroup name in the message header?

> Look, this is my entire config:
>
>


> cer@Telcontar:~> cat /etc/leafnode/config | egrep -v "^:space:]]*$|^#"
> expire = 400
> server = nntp.opensuse.org
> server = nntp.novell.com
> timeout = 30
> groupexpire opensuse.org.help.howto = -1
> groupexpire opensuse.org.help.howto.submissions = -1
> groupexpire opensuse.org.* = 600
> initialfetch = 1000
> debugmode = 1
> maxage = 0
> maxbytes = 500000
> timeout_long = 60
> allow_8bit_headers = 1
> cer@Telcontar:~>
> 

>
> Do you really think /that/ is difficult? :slight_smile:

Not exactly… But figuring out what to put there might be.

But using your file as a guide, it looks like the one config can manage news
retrieval & expiration of newsgroups on more than one server?

IE it could take care of BOTH nntp.opensuse.org AND news.gmane.org for me?

I notice you define groupexpire details for only 3 groups. Does that mean that
your leafnode doesn’t fetch articles from:

opensuse.org.help.hardware.64-bit

for example??

I subscribe to only a few groups on any one news server, And wouldn’t want to
waste time or filespace on the ones I’m not currently interested in.

> The only snag is that it uses one file per post, and that’s a huge lot
> of inodes - thus I make sure that /var/spool/news is on reiserfs.

I’ve never tried reisefs. Does it take anything more than using the right
fs info in /etc/fstab to get ANY linux distro to mount, read, and write to it
properly???

Is there any reason the reisefs partition mounted on /var/spool/news couldn’t
be on an external usb drive??

>
> Then you need a cronjob to do the hard work:

I never got used to using cronjobs for anything. I never know when my computer
will be on, nor which Linux installation it will be running when it is.

But “need” is a funny word. Many Linux users seem to think fetchmail
needs to run automatically… But the very first time it ever runs on my
system when I hadn’t just explicitly called it on the command line, I’ll
be removing fetchmail from my system. I’d feel the same way about the
fetchnews command. I mean I wouldn’t want it to run while the drive with
the reisefs partition was attached to the other computer…

Can it be configured to fetch and or expire news ONLY upon command? And never
ever automatically do either? And if I didn’t use a cronjob, would there be a
way to command pending posts to upload to the news servers?


JtWdyP

On 2013-07-15 20:57, JtWdyP wrote:
>
> It would appear that on Jul 14, Carlos E. R. did say:

>>> I also don’t want to deal with the {to me} horrid gui user interface I’ve found
>>> with any of the gui clients I’ve ever looked at.
>>
>> Well, I use alpine all the time, but I find the interface of the text
>> nntp clients I have tried awkward.
>
> As long as you don’t include alpine in your list of text nntp clients, I’d
> likely agree.

I do :stuck_out_tongue:

I could use the UI of alpine in news just fine - but their refusal to
cache the index and status of emails is an absolute unacceptable to me.
Yes, I did ask the devs about this.

> To me checking any subscribed newsgroups are part of what I think of as my
> email chores. I really like to use the same interface for read/compose
> on both…

Yes, me too, I use alpine for mail; I do want to use it for news, but it
is impossible. The idea of “deleting” posts that I have read is a no-no
(yes, that was what the alpine devs recommended to me). I want all posts
accessible so that I can go back and read the entire thread tomorrow or
next month. And I also want to keep track of what I read already.

Just because of that fault, I turned to Thunderbird for nntp, with the
consequence that I also use it for email sometimes, to the detriment of
Alpine :frowning:

(the other two text nntp readers I know of, are slrn and tin)

>> It is not a “true” news spool, it is rather a news proxy.
>
> Not sure I understand the difference… But I see references to something
> called fetchnews…
>
> I use a script to retrieve most of my email via fetchmail using procmail as an
> mda… would you describe that as a mail proxy?

Ok, let me try to describe.

Fetchnews pulls news from a real nntp server, and pushes to it new posts
to be sent. When you run it the first time, it just downloads the list
of groups.

When you try to view any group, with alpine, say, it will just display a
single faked message saying that the group is yet empty, but that it
will be retrieved shortly.

The next time that fetchnews runs, it downloads those groups that you
touched. And, conversely, if you stop reading a group for some time, it
will stop downloading it. Eventually, as groups are purged periodically
of old posts, that group will be empty again.

It is a proxy, because you can not create groups, nor have them
transferred to other servers. It is a “leaf” of a real server, a
“slave”. You choose the name :slight_smile:

Also it is a proxy because you can place it at the entry point of a
computer room, so that everybody connects to the news proxy instead of
the “real” server somewhere, reducing load on that server.

>> It is little work, and once done, it is blazingly fast, posts light up
>> (even on alpine) in a split second. With nntp.opensuse.org I can count
>> the seconds sometimes.
>
> So you can use alpine with it… What about posting a message composed in
> alpine?? Is it as simple as selecting the local news spool in the .pinerc
> and then putting the appropriate newsgroup name in the message header?

Absolutely.

>
>> Look, this is my entire config:
>>
>>


>> cer@Telcontar:~> cat /etc/leafnode/config | egrep -v "^:space:]]*$|^#"
>> expire = 400
>> server = nntp.opensuse.org
>> server = nntp.novell.com
>> timeout = 30
>> groupexpire opensuse.org.help.howto = -1
>> groupexpire opensuse.org.help.howto.submissions = -1
>> groupexpire opensuse.org.* = 600
>> initialfetch = 1000
>> debugmode = 1
>> maxage = 0
>> maxbytes = 500000
>> timeout_long = 60
>> allow_8bit_headers = 1
>> cer@Telcontar:~>
>> 

Do you really think /that/ is difficult? :slight_smile:

Not exactly… But figuring out what to put there might be.

Take the above as a guide - there is a sample config which I altered. As
you see, I have defined two nntp servers, with the idea that if one
fails the other may be active. It works.

But using your file as a guide, it looks like the one config can manage news
retrieval & expiration of newsgroups on more than one server?

Yes! If the groups match, when it tries to download from the second
server it sees that the posts have already been downloaded and skips them.

And, if the groups do not match (not the case here) then you have a
joined list of groups and it will automatically download and send to the
right one.

IE it could take care of BOTH nntp.opensuse.org AND news.gmane.org for me?

Yes, I believe it would. I have not tried news.gmane.org myself, but it
should.

I notice you define groupexpire details for only 3 groups. Does that mean that
your leafnode doesn’t fetch articles from:

opensuse.org.help.hardware.64-bit

for example??

No, it means that I have different settings for those 2. The rest use
default settings (the ‘*’). It means that I do not expire howto posts.
The rest expire in 600 days, almost two years, which is a lot :slight_smile:

I subscribe to only a few groups on any one news server, And wouldn’t want to
waste time or filespace on the ones I’m not currently interested in.

Correct, that’s automated.

Notice that if you keep subscribed to a group in thunderbird, as
thunderbird scans those for new posts, to display the number of new
posts on each, it means that those groups are downloaded for ever. With
thunderbird you have to unsubscribe from those groups you do not want,
and after a period of… I think it is 600 days (!) in my case.

Not sure of that, though. There may be a different setting for purging
mails and for fetching posts from a non-touched group. I could not
locate that setting if it exists, in a brief reading just now.

The only snag is that it uses one file per post, and that’s a huge lot
of inodes - thus I make sure that /var/spool/news is on reiserfs.

I’ve never tried reisefs. Does it take anything more than using the right
fs info in /etc/fstab to get ANY linux distro to mount, read, and write to it
properly???

Well, you need to create it first, obviously :slight_smile:

The fstab entry for me is:


> LABEL=c_home_mail   /home1     reiserfs        acl,user_xattr,barrier=flush    1 2

But it could as well be “defaults”. Also I have:


> Telcontar:~ # l /var/spool/news lrwxrwxrwx 1 root root 21 Sep 25  2010 /var/spool/news -> /home1/var_spool_news/

Note that it will work on any filesystem; but if you look at the
ext2/3/4 manual, you will see …] No, you don’t. That is, I can not
find it. Years ago there was a setting when creating a filesystem to
specify that you wanted it for ‘news’ and the filesystem would have a
lot of inodes and smaller chunks.

I would use xfs if reiser is not a possibility, and if not, ext4, but
adjusting for that usage. I would have to hink of the exact settings.

> Is there any reason the reisefs partition mounted on /var/spool/news couldn’t
> be on an external usb drive??

Only speed.

Well, do not use a flash system, either; use a magnetic disk, IMO,
because it would trash (wear) a lot. Many small writes.

>> Then you need a cronjob to do the hard work:
>
> I never got used to using cronjobs for anything. I never know when my computer
> will be on, nor which Linux installation it will be running when it is.

No problem. You just have the cronjob defined, and if the computer is
off, it does not run :slight_smile:

> But “need” is a funny word. Many Linux users seem to think fetchmail
> needs to run automatically… But the very first time it ever runs on my
> system when I hadn’t just explicitly called it on the command line, I’ll
> be removing fetchmail from my system. I’d feel the same way about the
> fetchnews command. I mean I wouldn’t want it to run while the drive with
> the reisefs partition was attached to the other computer…

Ok, you can run it manually :slight_smile:

I did have it manually for a time, till I noticed that I forgot to call
it. You see, if you do not call it, the posts you write are not sent.

> Can it be configured to fetch and or expire news ONLY upon command? And never
> ever automatically do either? And if I didn’t use a cronjob, would there be a
> way to command pending posts to upload to the news servers?

Expire is a cronjob installed from the rpm, don’t worry about it.

“fetchnews” is the command used to pull and push posts.

Try it… if you do not like it, remove it :slight_smile:

It was invented when people used modems, because it reduced real
traffic, and you could read posts without connection. But even nowdays
it has its uses. I use it because browsing the posts is very fast, I
don’t have to wait for the post to arrive here.


Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

It would appear that on Jul 16, Carlos E. R. did say:

> On 2013-07-15 20:57, JtWdyP wrote:
> >
> > As long as you don’t include alpine in your list of text nntp clients, I’d
> > likely agree.
>
> I do :stuck_out_tongue:
>
> I could use the UI of alpine in news just fine - but their refusal to
> cache the index and status of emails is an absolute unacceptable to me.
> Yes, I did ask the devs about this.
>
>
> > To me checking any subscribed newsgroups are part of what I think of as my
> > email chores. I really like to use the same interface for read/compose
> > on both…
>
> Yes, me too, I use alpine for mail; I do want to use it for news, but it
> is impossible. The idea of “deleting” posts that I have read is a no-no
> (yes, that was what the alpine devs recommended to me). I want all posts
> accessible so that I can go back and read the entire thread tomorrow or
> next month. And I also want to keep track of what I read already.

Well, I’ve always done the expunge for remote newsgroups… If I find a reason
to go back into the expunged messages for something I didn’t keep in my local
{interesting message} folder, I have a simple kludge.

Being that my user is in the wheel group I can use my secondary user account
via:

Code:

JtWdyP -> /home/jtwdyp
> su - mvjap3
Password:
-su: eval: line 15: syntax error near unexpected token newline' -su: eval: line 15: Usage: lesspipe <file>’
altM3 -> /home/mvjap3
> alpine

And as long as mvjap3 doesn’t expunge newsgroups…

But I doubt you’d find that kludge acceptable…

> >> It is not a “true” news spool, it is rather a news proxy.
> >
> > Not sure I understand the difference…
>
> Ok, let me try to describe.
>
> Fetchnews pulls news from a real nntp server, and pushes to it new posts
> to be sent. When you run it the first time, it just downloads the list
> of groups.
>
> When you try to view any group, with alpine, say, it will just display a
> single faked message saying that the group is yet empty, but that it
> will be retrieved shortly.
>
> The next time that fetchnews runs, it downloads those groups that you
> touched. And, conversely, if you stop reading a group for some time, it
> will stop downloading it. Eventually, as groups are purged periodically
> of old posts, that group will be empty again.

OK So once it has the list of groups, that list would, I think, be
displayed in alpine if I go to the newsgroup collection screen and type
“a” {add group} then “^T” {To All Grps}. Presumably just displaying the
groupnames wouldn’t be enough for fetchnews to think they were touched?
And quite possibly, just subscribing might not do it. But it sounds like
attempting to display the index of just that one “faked message” likely would.

> >> It is little work, and once done, it is blazingly fast, posts light up
> >> (even on alpine) in a split second. With nntp.opensuse.org I can count
> >> the seconds sometimes.
> >
> > So you can use alpine with it… What about posting a message composed in
> > alpine?? Is it as simple as selecting the local news spool in the .pinerc
> > and then putting the appropriate newsgroup name in the message header?
>
> Absolutely.

> >> The only snag is that it uses one file per post, and that’s a huge lot
> >> of inodes - thus I make sure that /var/spool/news is on reiserfs.
> >
> > I’ve never tried reisefs. Does it take anything more than using the right
> > fs info in /etc/fstab to get ANY linux distro to mount, read, and write to it
> > properly???
>
> Well, you need to create it first, obviously :slight_smile:

Yeah… But having created it on say opensuse, I’d likely only need to mount
it on sabayon or mageia etc… to use it. Well that and install an configure
the leafnode software to actually “use” it. :wink:

> > Is there any reason the reisefs partition mounted on /var/spool/news couldn’t
> > be on an external usb drive??
>
> Only speed.
>
> Well, do not use a flash system, either; use a magnetic disk, IMO,
> because it would trash (wear) a lot. Many small writes.

Yeah, I’d like to think I’d have thought of that… Moot point though, I’ve
already got an magnetic disk usb drive with an empty partition with nearly half
a terabyte of space… I’d be thinking of using that…

> >> Then you need a cronjob to do the hard work:
> >
> > {snip} fetchnews command. I mean I wouldn’t want it to run while the
> > drive with the reisefs partition was attached to the other computer…
>
> Ok, you can run it manually :slight_smile:
>
> I did have it manually for a time, till I noticed that I forgot to call
> it. You see, if you do not call it, the posts you write are not sent.
>
> Expire is a cronjob installed from the rpm, don’t worry about it.
>
> “fetchnews” is the command used to pull and push posts.

OK I think I got that. Only issue might be that I never know which of my linux
I’ll be running when I get around to “doing” my email/newsgroup “chores”
Would there likely be any conflict to having an identical set up on each of my
linux… So that the expire cronjob configured to whichever Linux I’ve booted
would be running (hopefully not conflicting with the cronjob on the Linux I
was using the day before…) And the fetchnews command would be called
manually when I know the reisefs partition is mounted…

I wonder how the Expire cronjob would react to “/var/spool/news” being empty
due to the reisefs partition NOT being mounted…

> Try it… if you do not like it, remove it :slight_smile:

It will take some time… But since I doubt that a quick fix for the
access issues I’m having with alpine on this nntp server, will be forthcoming
from the alpine devs, I will most likely give it a try on at least one of my
linux…

Thanks!


| — ___
| <0> <->
| ^ JtWdyP
| ~___/~

On 2013-07-16 21:04, JtWdyP wrote:

>> Yes, me too, I use alpine for mail; I do want to use it for news, but it
>> is impossible. The idea of “deleting” posts that I have read is a no-no
>> (yes, that was what the alpine devs recommended to me). I want all posts
>> accessible so that I can go back and read the entire thread tomorrow or
>> next month. And I also want to keep track of what I read already.
>
> Well, I’ve always done the expunge for remote newsgroups… If I find a reason
> to go back into the expunged messages for something I didn’t keep in my local
> {interesting message} folder, I have a simple kludge.
>
> Being that my user is in the wheel group I can use my secondary user account
> via:

Ugh. I just push the {up} key in the keyboard :-p

> And as long as mvjap3 doesn’t expunge newsgroups…
>
> But I doubt you’d find that kludge acceptable…

Nope! :slight_smile:

>> The next time that fetchnews runs, it downloads those groups that you
>> touched. And, conversely, if you stop reading a group for some time, it
>> will stop downloading it. Eventually, as groups are purged periodically
>> of old posts, that group will be empty again.
>
> OK So once it has the list of groups, that list would, I think, be
> displayed in alpine if I go to the newsgroup collection screen and type
> “a” {add group} then “^T” {To All Grps}. Presumably just displaying the
> groupnames wouldn’t be enough for fetchnews to think they were touched?
> And quite possibly, just subscribing might not do it. But it sounds like
> attempting to display the index of just that one “faked message” likely would.

Yep.

Although I had that running in Alpine, now it doesn’t work, and I don’t
remember if I deactivated it. I can not get any groups.

I have a collection defined thus:


Nickname  : Local News
Server    : telcontar.valinor
Path      : #news./
View      :

and the config is:


NNTP Server (for news)  = telcontar.valinor

That, or very similar, worked once, but not today. No idea why. On the
other hand, slrn, which I have not touched in months or years, continues
working.

>>> I’ve never tried reisefs. Does it take anything more than using the right
>>> fs info in /etc/fstab to get ANY linux distro to mount, read, and write to it
>>> properly???
>>
>> Well, you need to create it first, obviously :slight_smile:
>
> Yeah… But having created it on say opensuse, I’d likely only need to mount
> it on sabayon or mageia etc… to use it. Well that and install an configure
> the leafnode software to actually “use” it. :wink:

Yep.

Thinking… :-?

openSUSE was the first one to support reiserfs, even before it was
mainlined in the kernel. And it keeps suporting it, but nobody knows for
how long. I do not know if other distributions support it as well.

>> Well, do not use a flash system, either; use a magnetic disk, IMO,
>> because it would trash (wear) a lot. Many small writes.
>
> Yeah, I’d like to think I’d have thought of that… Moot point though, I’ve
> already got an magnetic disk usb drive with an empty partition with nearly half
> a terabyte of space… I’d be thinking of using that…

You don’t need a lot of space. I have 562579 files there, but it uses
just 1200M. The partition is 50G, used 32G (the rest is email).

You could use internal space if you can create a large file anywhere,
loop it, and create a filesystem in the loop.

>> “fetchnews” is the command used to pull and push posts.
>
> OK I think I got that. Only issue might be that I never know which of my linux
> I’ll be running when I get around to “doing” my email/newsgroup “chores”
> Would there likely be any conflict to having an identical set up on each of my
> linux…

Dunno… I never did that.

> So that the expire cronjob configured to whichever Linux I’ve booted
> would be running (hopefully not conflicting with the cronjob on the Linux I
> was using the day before…) And the fetchnews command would be called
> manually when I know the reisefs partition is mounted…

I think it would just abort. I had the partition umounted yesterday to
run an fsck on it, and I didn’t bother to stop cronjobs.

However… my “/var/spool/news” is a symlink to
“/home1/var_spool_news/”; if something wants to work in
“/var/spool/news” when “/home1/” is not mounted it gets an error.

But, if “/var/spool/news” is a mount point, and the partition is not
mounted, the mount point is a directory… there would be no error
trying to use it. If you expect a directory or file inside it would
error out, but it might try to create the structures… in the wrong
partition.

>
> I wonder how the Expire cronjob would react to “/var/spool/news” being empty
> due to the reisefs partition NOT being mounted…

Expire would be fine, its job is to remove old messages, which would not
exist. But leafnode would perhaps try to create the structure. I’m unsure.

>> Try it… if you do not like it, remove it :slight_smile:
>
> It will take some time… But since I doubt that a quick fix for the
> access issues I’m having with alpine on this nntp server, will be forthcoming
> from the alpine devs, I will most likely give it a try on at least one of my
> linux…
>
> Thanks!

Welcome.

Leafnode, once you try it, is useful.

You might want to look at slrn and tin, both text mode nntp readers.
However, tin is difficult to find; I got “tin-unstable” from the
“server:mail” repo, version 2.1. There is 2.0.1 on a home repo.


Cheers / Saludos,

Carlos E. R.
(from 12.3 x86_64 “Dartmouth” at Telcontar)

It would appear that on Jul 17, Carlos E. R. did say:

> On 2013-07-16 21:04, JtWdyP wrote:
> > Being that my user is in the wheel group I can use my secondary user account
> > via:
>
> Ugh. I just push the {up} key in the keyboard :-p

I’m thinking this {up} key is a gnome thing?

If my use of the {su - mvjap3} was often enough to warrant it, I would most
likely assign a command to some keybinding in E17… {I may have told you
that I ran away from kde when most distros began defaulting to kde4, I did
however take with me a fondness for a few kde apps such as konsole… Thus
the command string that I’d likely put on <ctrl>+<alt>+a is:
Code:

konsole --profile RedGray -e su “-c alpine” - mvjap3

The RedGray profile insuring the screen colors were different enough from
my normal alpine within konsole session as to preclude getting them mixed up.

> > And as long as mvjap3 doesn’t expunge newsgroups…
> >
> > But I doubt you’d find that kludge acceptable…
>
> Nope! :slight_smile:
>
> > OK So once it has the list of groups, that list would, I think, be
> > displayed in alpine if I go to the newsgroup collection screen and type
> > “a” {add group} then “^T” {To All Grps}. Presumably just displaying the
> > groupnames wouldn’t be enough for fetchnews to think they were touched?
> > And quite possibly, just subscribing might not do it. But it sounds like
> > attempting to display the index of just that one “faked message” likely would.
>
> Yep.
>
> Although I had that running in Alpine, now it doesn’t work, and I don’t
> remember if I deactivated it. I can not get any groups.

Since leafnode IS your local news spool, I would think that defining { & selecting ;-p }
the news server name in the alpine config of “[127.0.0.1]” or in your case “telcontar.valinor”
{{Though, for me the “[127.0.0.1]” form would be more useful, as I sometimes
alter my .pinerc and the copy the new version to my other Linux… And each
installed Linux has a different hostname.}}

And then going into the news collection screen, and typing an “a” then “^T” should
get you a list of possible groups… But you would need to “subscribe” to one
before alpine would display the group itself in it’s news collection screen…

If/when I get this working I’ll let you know if that was all there was to it…

>
> I have a collection defined thus:
>
>


> Nickname  : Local News
> Server    : telcontar.valinor
> Path      : #news./
> View      :
> 

and the config is:


> NNTP Server (for news)  = telcontar.valinor
> 

>
>
> That, or very similar, worked once, but not today. No idea why. On the
> other hand, slrn, which I have not touched in months or years, continues
> working.

Code:

if “telcontar.valinor” = “$(cat /etc/hostname)” ] ; then echo “That shoulda worked!” ;fi

>
>
> >>> I’ve never tried reisefs. Does it take anything more than using the right
> >>> fs info in /etc/fstab to get ANY linux distro to mount, read, and write to it
> >>> properly???
> >>
> >> Well, you need to create it first, obviously :slight_smile:
> >
> > Yeah… But having created it on say opensuse, I’d likely only need to mount
> > it on sabayon or mageia etc… to use it. Well that and install an configure
> > the leafnode software to actually “use” it. :wink:
>
> Yep.
>
> Thinking… :-?
>
> openSUSE was the first one to support reiserfs, even before it was
> mainlined in the kernel. And it keeps suporting it, but nobody knows for
> how long. I do not know if other distributions support it as well.

I just did a quick peek at the file systems of all 5 linux I’ve got on my
desktop… Only saw /sbin/mkfs.xfs twice, But all of them save Mageia had
/sbin/mkfs.reiserfs And that was quickly fixed by installing a package named
“reiserfsprogs”

> >> “fetchnews” is the command used to pull and push posts.
> >
> > OK I think I got that. Only issue might be that I never know which of my linux
> > I’ll be running when I get around to “doing” my email/newsgroup “chores”
> > Would there likely be any conflict to having an identical set up on each of my
> > linux…
>
> Dunno… I never did that.
>
> > So that the expire cronjob configured to whichever Linux I’ve booted
> > would be running (hopefully not conflicting with the cronjob on the Linux I
> > was using the day before…) And the fetchnews command would be called
> > manually when I know the reisefs partition is mounted…
>
> I think it would just abort. I had the partition umounted yesterday to
> run an fsck on it, and I didn’t bother to stop cronjobs.
>
> However… my “/var/spool/news” is a symlink to
> “/home1/var_spool_news/”; if something wants to work in
> “/var/spool/news” when “/home1/” is not mounted it gets an error.

GOOD! was already planning to test if could use symlink for that. Means
that {assuming it all works for me} When I decide to move the usb drive to
the laptop, I can be using a symlink to a directory on the reiserfs
partition instead of the whole partition. That way I could include on it, a copy
of my ~/mail/postponed-msgs folder, etc…

> But, if “/var/spool/news” is a mount point, and the partition is not
> mounted, the mount point is a directory… there would be no error
> trying to use it. If you expect a directory or file inside it would
> error out, but it might try to create the structures… in the wrong
> partition.

Yet another good reason to love symlinks…

> >
> > I wonder how the Expire cronjob would react to “/var/spool/news” being empty
> > due to the reisefs partition NOT being mounted…
>
> Expire would be fine, its job is to remove old messages, which would not
> exist. But leafnode would perhaps try to create the structure. I’m unsure.

Methinks the symlink might be a blessing here as well then.

> > Thanks!
>
> Welcome.
>
> Leafnode, once you try it, is useful.
>
>
> You might want to look at slrn and tin, both text mode nntp readers.
> However, tin is difficult to find; I got “tin-unstable” from the
> “server:mail” repo, version 2.1. There is 2.0.1 on a home repo.

Good to know in case I can’t get it working in alpine.

Thanks again!


| ~^~ ~^~
| <> <>
| ^ JtWdyP
| ___/

It would appear that on Jul 14, Carlos E. R. did say:

> On 2013-07-14 18:02, JtWdyP wrote:
> > I did look at a wiki on leafnode, it implied that I’d need to bother with
> > setting up a fully qualified domain name.
>
> Why? I don’t have one. If you say that “telcontar.valinor” is a fully
> qualified name, then so it is. :slight_smile:

I think that may depend on which version your using… AND/OR where/how you
got it…

Since I intended to use this on multiple linux, I didn’t think it mattered
which linux I tried it on first… I was running Mageia when I read this so
I checked their repo for leafnode… It took me a bit to get it configured
well enough for fetchnews to actually try to run. The error text made me think
that I only needed to get something unique enough, so I played with Mageia’s
systemd hostname utility:



I thought it might be a matter of latest/greatest versions being picky…
My antiX distro’s repos still have older versions of some software So I tried
it there next: There, I wound up getting really creative with the hostname
stuff. decided to display the modified settings via antiX’s hostname &
domainname commands before my last attempt to run fetchnews there:



For opensuse, I resorted to using the:

http://software.opensuse.org/search

for leafnode on opensuse 12.2
I think there were two versions available for “one click install”. I selected
the older one…

Finally I had a leafnode the didn’t throw an FQDN error when I tried to run
fetchnews… But after the first run, I needed to access the groups with a
news client to make them “interesting” enough for fetchnews to retrieve
anything. With Alpine, I needed to “subscribe” to a group for that. But when I
tried to list available groups in the add group dialog, alpine couldn’t find a
list of them…

I really dislike slrn’s interface, but it would let me use vim to compose…
But slrn couldn’t find any groups to open either…

Since then I needed to house/dog sit at my sisters for a few days. And her
broadband is from a different isp than mine. And at her house my laptop has no
trouble opening nntp.opensuse.org with directly with alpine. So It seams that
my access problem might be mostly an ISP thing…

At this point, I’m not thinking I can get leafnode working on enough distros
without paying somebody for an FQDN… Giving up {for now} have other fish to
fry…

But thanks for the help… At least I learned something. ;-7


| ^^^ ^^^
| <o> <o>
| ^ JtWdyP
| ___

… <sigh>