OT: Thunderbird newsreader problem

[Lobster]

Well, I feel kind of justified asking this here, because it's entirely due to the panning by the uk.d-i-y cogniscenti of Outlook Express (which never gave me any problems!) that I switched to Thunderbird a few months ago...

Does anybody know why it is that every few weeks, I get a flood of duplicate copies of random posts reappearing, often many days after the original was posted? They appear within threads at random, are flagged as 'new' messages, but they have the same timestamp as the original. A right PITA. Maybe I should just use Google Groups instead ;-)

Could this be a function of my ISP (ntl) rather than my newsreader software? I'm not sufficiently familiar with the mechanics of Usenet etc to know.

Thanks David

[Rob Morley]

ntl seem to re-index the messages every so often - same message with a different article number, so your newsreader thinks it's new.

[Lee]

Don't know whether that's the reason or not, but I get the same problem with the NTL server.

Might be worth paying the 10 Euros and getting a news.individual account, which doesn't have that problem.

Lee

[stejonda]

In message , Rob Morley writes

That wouldn't explain why the problem occurs with Thunderbird & not OE.

Is there a way of extending the time before articles are expired in Thunderbird? It may be that having expired old articles from the local database Thunderbird is then as a result fetching them from the newsserver anew.

[AlexW]

You could try comparing NTL with the free read-only server at freetext.usenetserver.com.

This might shed some light on the issue.

I use Thunderbird and have not seen this yet (on my company's NNTP server).

[Rob Morley]

A news client may use article number (which will be different on each server) or message id (which is unique to the message), also it may or may not deal with messages out of order.

[AlexW]

Also it might be worth upgrading to the latest version if you have been using it for a while (1.0 is available now).

(... I am preparing to be flamed now if I do not remember correctly ...)

IIRC NNTP uses an I have / send me protocol (RFC977) which uses a message id field in the message header to determine if the message exists locally (this is both server-server and reader-server).

This may look something like [e-mail] (the id of my last post).

You can check each of the duplicates by doing View->Message Source in Thunderbird. If they are the same then there is a client issue ... if they are different then its a server side thing. It sounds like a client thing as I can't imagine why an NNTP server would re-write these id's.

You might also want to check the in reply to field which is used for building threads up (not sure what Thunderbird will do with this).

Forums exist ar mozilla.org for Thunderbird support.

alex

[Andrew Chesters]

I've seen it too, with TBird & NTL. Most of my groups are now on a News.Individual account which seems OK. Still read a couple of .free groups on NTL but couldn't swear to whats going on over there any more.

[AlexW]

Hmm ... looking at the source code ... It looks like TB uses article ids rather than message Id's (as suggested by Rob Morely in a previous post on a different branch of the thread). Don't think this is as easily available to perform the checks described.

Not sure what Outlook uses (and I don't have the source code) so it still could be client or server I guess.

Alex.

[Mark Carver]

On a related subject, ever since my anti-virus program Avast updated itself to V4.6 it normally takes two or three attempts to post to Usenet via Thunderbird. I get 'Authentication Error 480', sending again straight away normally works, and third time is always lucky.

Of course, sod's law, every attempt in uk.test works first time :-)

I assume it's the upgrade to Avast V4.6, as this problem started straight after that. I use the same individual.net account to post when at work and that's fine, so I've discounted that as being at fault (probably ?).

[Lobster]

Yes, have been on 1.0 for some time

Yes the id's are indeed identical: Thunderbird's fault then?

David

[Rob Morley]

In that case it seems that Thunderbird uses article numbers rather than message IDs to retrieve messages - it makes more sense as they're allocated sequentially as new messages arrive at the server, while the only thing you can be certain about with a message ID is that it's unique. If you want to blame anyone blame ntl for not maintaining consistent numbering on their server.

[John Rumm]

It is the sort of behaviour I have only seen when an ISP does something with the news server (like installs a new one) that causes it to play catch up over a period of a few days and in the process re-number messages.

You can control how long TB holds onto messages (in the preferences) that it has retrieved - this may or may not make a difference to what you are seeing.

[John Rumm]

Does the AV program do any DNS caching?

I have had simmilar problems when my router decides to use one of the ISPs DNS twice for DNS relay rather than using the primary and secondary... Hence you get the occational DNS timeout that will casue a post to fail.

The fix (in my case) is to re-educate the router to use one of the other DNS. then it is happy.

[dmc]

From what I remember of OE (and it's been a while since I looked) I seem to recall that OE threads on Subject line. While this is totally broken if true it may explain why renumbering articles wouldn't affect it.

Of course, why the hell NTL renumber their articles regularly is a different question. I've run news servers for 10 years and have only had to do it a couple of times. Even then, I suspect that we could have avoided it if we had tried hard enough.

Darren

[AlexW]

Rather ironically, I just had similar thing occur this morning. And then TB asked me if I wanted to download all 242K new articles!

It seems it might be TB, inc v1.0.

I am not on NTL. My company server is NNTPcache. You can (probably) test what your server is by using:

tenlnet servername 119

and seeing what it says (type quit then press return to get out).

My workaround was to unsubscribe from the uk.d-i-y news group and re-subscribe - you obviously loose any message states as they all appear unread etc. This might take a while on dial-up as well.

HTH

Alex.

[SmileyFace]

As a few have suggested, NTL is to blame here. It happens periodically for no apparent reason. It also happens in OE, you must've been lucky before. A burst of them came through on the 8th for the first time in weeks.

Irritating, but better than it was when the server ran at about 20% retention.

HTH

[Lobster]

It says "command not found"... but when I tried telnet instead :-) I got:

200 ntl cablemodem news server (tornado v1.0.6)

Does that help or mean anything useful?

David

[AlexW]

"tornado" probably refers to Highwinds software nntp server, this has a load balancing front end apparently, which *may* mean that article ids are not (always) synchronised across physical hosts. Combine with the TB policy of using the article id for matching articles and this could be the problem.

To understand this issue better would probably require a detailed knowledge of how tornado works and how its configured in this instance (at least in terms or load balancing) ... its probably not worth the effort.

This does not explain what happened to me though as I beleive our server is not load balanced.

[Philip K]

I think that you mean Microsoft Outlook Express. Microsoft Outlook is not a newsreader.