Cannot recover if gmane goes offline

Cannot recover if gmane goes offline

[Tassilo Horn]

Hi all,

Lately, my gmane server frequently goes offline, or better, the gmane
nntp select method is shown as (offline) in the server buffer.  I have
no clue why it's always Gmane while other nntp and IMAP servers don't
show that issue.

  (add-to-list 'gnus-secondary-select-methods
               '(nntp "Gmane"
                      (nntp-address "news.gmane.org")))

Anyway, when I fetch new news then, all my gmane/gwene groups show up
with thousands of unread messages.  Simply reopening the connection
using `O' in *Server* doesn't help; the unread (but actually read)
messages stay unread.

Restarting Gnus using `R' in *Group* or `q' followed by M-x gnus RET
doesn't help either.  So the only solution that does work is restarting
emacs completely.

That's quite some pain, especially if you have open IRC buffers with
active discussions and stuff.  Sure, I could fire up another emacs
instance for Gnus, but I like having all in one, so that I can easily
access all buffers.

Does anyone face the same troubles?

Bye,
Tassilo
-- 
Sent from my Emacs

Re: Cannot recover if gmane goes offline

[Ted Zlatanov]

On Fri, 10 Jun 2011 10:34:54 +0200 Tassilo Horn <tassilo@member.fsf.org> wrote: 

TH> Lately, my gmane server frequently goes offline, or better, the gmane
TH> nntp select method is shown as (offline) in the server buffer.  I have
TH> no clue why it's always Gmane while other nntp and IMAP servers don't
TH> show that issue.

TH>   (add-to-list 'gnus-secondary-select-methods
TH>                '(nntp "Gmane"
TH>                       (nntp-address "news.gmane.org")))

TH> Anyway, when I fetch new news then, all my gmane/gwene groups show up
TH> with thousands of unread messages.  Simply reopening the connection
TH> using `O' in *Server* doesn't help; the unread (but actually read)
TH> messages stay unread.

It seems like something is screwed up in the server ranges.  I've never
experienced this, sorry.  Maybe look at the NNTP conversation to see if
anything is obviously wrong after the timeout?

It could be a firewall being funny :)

Ted

Re: Cannot recover if gmane goes offline

[Tassilo Horn]

Ted Zlatanov <tzz@lifelogs.com> writes:

Hi Ted,

> TH>   (add-to-list 'gnus-secondary-select-methods
> TH>                '(nntp "Gmane"
> TH>                       (nntp-address "news.gmane.org")))
>
> TH> Anyway, when I fetch new news then, all my gmane/gwene groups show up
> TH> with thousands of unread messages.  Simply reopening the connection
> TH> using `O' in *Server* doesn't help; the unread (but actually read)
> TH> messages stay unread.
>
> It seems like something is screwed up in the server ranges.  I've
> never experienced this, sorry.

Would that explain the problem occuring only for Gmane, too?

> Maybe look at the NNTP conversation to see if anything is obviously
> wrong after the timeout?

That's *nntp-log*, right?

Bye,
Tassilo
-- 
Sent from my Emacs

Re: Cannot recover if gmane goes offline

[Tassilo Horn]

Tassilo Horn <tassilo@member.fsf.org> writes:

Hi!

>> It seems like something is screwed up in the server ranges.  I've
>> never experienced this, sorry.
>
> Would that explain the problem occuring only for Gmane, too?
>
>> Maybe look at the NNTP conversation to see if anything is obviously
>> wrong after the timeout?
>
> That's *nntp-log*, right?

Ok, once again gmane has gone offline, and all my gmane/gwene groups
show thousands of unread messages.  The last Gmane output in *nntp-log*
is this:

--8<---------------cut here---------------start------------->8---
20110616T125303.731 news.gmane.org MODE READER
20110616T125303.772 news.gmane.org LIST ACTIVE gwene.de.pro-linux-news
20110616T125303.772 news.gmane.org LIST ACTIVE gwene.de.dhv.typetests
20110616T125303.772 news.gmane.org LIST ACTIVE gwene.de.dhv.safetynotes
20110616T125303.772 news.gmane.org LIST ACTIVE gwene.de.dhv.dhvnews
20110616T125303.772 news.gmane.org LIST ACTIVE gmane.emacs.tramp
20110616T125303.772 news.gmane.org LIST ACTIVE gmane.emacs.orgmode
20110616T125303.772 news.gmane.org LIST ACTIVE gmane.emacs.help
20110616T125303.772 news.gmane.org LIST ACTIVE gmane.emacs.gnus.user
20110616T125303.772 news.gmane.org LIST ACTIVE gmane.emacs.bugs
20110616T125303.773 news.gmane.org LIST ACTIVE gmane.emacs.auctex.general
20110616T125303.773 news.gmane.org LIST ACTIVE gmane.emacs.auctex.devel
--8<---------------cut here---------------end--------------->8---

That's the Gmane output produced when I reenabled the server using `O'
in the *Server* buffer.

--8<---------------cut here---------------start------------->8---
20110616T144328.172 news.gmane.org MODE READER
--8<---------------cut here---------------end--------------->8---

And that's what I get when hitting `g' in *Group*.

--8<---------------cut here---------------start------------->8---
20110616T144440.133 news.gmane.org LIST ACTIVE gwene.de.pro-linux-news
20110616T144440.133 news.gmane.org LIST ACTIVE gwene.de.dhv.typetests
20110616T144440.133 news.gmane.org LIST ACTIVE gwene.de.dhv.safetynotes
20110616T144440.133 news.gmane.org LIST ACTIVE gwene.de.dhv.dhvnews
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.tramp
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.orgmode
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.help
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.gnus.user
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.bugs
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.auctex.general
20110616T144440.133 news.gmane.org LIST ACTIVE gmane.emacs.auctex.devel
--8<---------------cut here---------------end--------------->8---

But still, all my gmane groups show all articles as unread... Hm, but
that doesn't look very helpful, does it?

Bye,
Tassilo
-- 
Sent from my Emacs

Re: Cannot recover if gmane goes offline

[Lars Magne Ingebrigtsen]

Tassilo Horn <tassilo@member.fsf.org> writes:

> Lately, my gmane server frequently goes offline, or better, the gmane
> nntp select method is shown as (offline) in the server buffer.  I have
> no clue why it's always Gmane while other nntp and IMAP servers don't
> show that issue.

[...]

> Anyway, when I fetch new news then, all my gmane/gwene groups show up
> with thousands of unread messages.

That's bizarre.  If Gnus can't get any response from the server, it
should simply list all the groups from that server with a "*" in front
of the group name.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: Cannot recover if gmane goes offline

[Tassilo Horn]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

Hi Lars,

>> Lately, my gmane server frequently goes offline, or better, the gmane
>> nntp select method is shown as (offline) in the server buffer.  I
>> have no clue why it's always Gmane while other nntp and IMAP servers
>> don't show that issue.
>
> [...]
>
>> Anyway, when I fetch new news then, all my gmane/gwene groups show up
>> with thousands of unread messages.
>
> That's bizarre.  If Gnus can't get any response from the server, it
> should simply list all the groups from that server with a "*" in front
> of the group name.

That's the theory, yes. ;-)

BTW: The problem is not Gmane specific.  At least once, the bad server
was gmail.  It's just that it happens with Gmane much more frequently.

Bye,
Tassilo
-- 
Sent from my Emacs

Re: Cannot recover if gmane goes offline

[Tassilo Horn]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

Hi Lars,

>> Lately, my gmane server frequently goes offline, or better, the gmane
>> nntp select method is shown as (offline) in the server buffer.  I have
>> no clue why it's always Gmane while other nntp and IMAP servers don't
>> show that issue.
>
> [...]
>
>> Anyway, when I fetch new news then, all my gmane/gwene groups show up
>> with thousands of unread messages.
>
> That's bizarre.  If Gnus can't get any response from the server, it
> should simply list all the groups from that server with a "*" in front
> of the group name.

This issue might have been caused by me setting gnus-propagate-marks to
t.  Since I've found out that for nnimap marks are propagated no matter
that variable, I stick with the default of nil again.  Since that, I've
lost the connection to gmane at least twice, but after reopening the
connection in *server* my read marks where all ok.

Bye,
Tassilo
-- 
Sent from my Emacs