New server problem

[Nick Roberts]

While I'm delighted to see that usenet is back, there seems to be a
problem with the server.

Every time I've connected the last 2 or 3 days, NewsHound has hung for
about 30 seconds waiting for a response from the server between getting
the list of new articles for each newsgroup and the articles starting
to download. This means that a typical download takes 2 - 3 minutes
longer than it should (4-6 newsgroups with a 30 second delay on each).

I wouldn't mind, but actually downloading the articles typically takes
less than 5-10 seconds per newsgroup, so this delay is imposing a
300-600% overhead.

[J Peachey]

Post by Nick Roberts
While I'm delighted to see that usenet is back, there seems to be a
problem with the server.
Every time I've connected the last 2 or 3 days, NewsHound has hung for
about 30 seconds waiting for a response from the server between getting
the list of new articles for each newsgroup and the articles starting
to download. This means that a typical download takes 2 - 3 minutes
longer than it should (4-6 newsgroups with a 30 second delay on each).
I wouldn't mind, but actually downloading the articles typically takes
less than 5-10 seconds per newsgroup, so this delay is imposing a
300-600% overhead.

Yes I've noticed this as well.
Heres the latest log from newshound.

18 Aug 14:39:59 050 Newsfetch complete. Received 23 articles, 57093
bytes (57093 cps) in -34 seconds.
18 Aug 14:39:59 050 Warning - duration of fetch was negative
(-1069481984)
18 Aug 14:46:40 100 Articles 150443 to 150452 have been cancelled in
group comp.sys.acorn.apps (we last saw 150452, server has 110591 to
150443
18 Aug 14:46:40 100 Articles 298292 to 298294 have been cancelled in
group comp.sys.acorn.misc (we last saw 298294, server has 234632 to
298292
18 Aug 14:46:40 100 Articles 51617 to 51619 have been cancelled in
group comp.sys.acorn.games (we last saw 51619, server has 47704 to
51617
18 Aug 14:46:40 100 Articles 84730 to 84732 have been cancelled in
group comp.sys.acorn.programmer (we last saw 84732, server has 69826
to 84730
18 Aug 15:45:44 050 Newsfetch complete. Received 1 articles, 10084
bytes (56 cps) in 180 seconds.
18 Aug 21:04:48 050 Newsfetch complete. Received 12 articles, 32888
bytes (276 cps) in 119 seconds.

Hope it doesn't mean the server is busy sending dvds out again.

John

[Paul Vigay]

Post by J Peachey
Hope it doesn't mean the server is busy sending dvds out again.

I think it's caused by everyone's article pointers being invalid. The last
time Newshound successfully fetched news was probably before it all went
pear-shaped. The news server was then down for 15 days or so.

When it came up, all the articles that Newshound was expecting to find have
probably expired on the server, I think Giganews only keep articles for 10
days or so.

Therefore, I suspect that Newshound is stalling having to search through
all news articles for the next article ID, which by now will be invalid.

On my newshound here I'd actually unsubscribed from all newsgroups, then
when the server came back online, I resubscribed to them all, and other
than downloading the last 50 articles, Newshound has been working fine from
here.

I don' really want to purge the cache, because it means the Orpheus server
will then have to re-download everything from Giganews, adding to our quota
again.

Have you tried unsubscribing from newsgroups (actually unsubscribing rather
than getting Pluto to temporarily disable them) and then resubscribing.
That /may/ sort the invalid article IDs on your machine out.

[Steve Clark]

In article <4f147f8625invalid-email-***@invalid-domain.co.uk>,
Paul Vigay <invalid-email-***@invalid-domain.co.uk> wrote:

[Snip]

Post by Paul Vigay
Have you tried unsubscribing from newsgroups (actually unsubscribing
rather than getting Pluto to temporarily disable them) and then
resubscribing. That /may/ sort the invalid article IDs on your machine
out.

er. How do you unsubscribe from a newsgroup?

Do you mean delete it from the Pluto list, run Newshound and then re-enter
it? or do you mean wipe the Newshound config file and then do a download
of the last e.g. 1 articles from the Pluto list?

[Paul Vigay]

Post by Steve Clark
Do you mean delete it from the Pluto list, run Newshound and then
re-enter it? or do you mean wipe the Newshound config file and then do a
download of the last e.g. 1 articles from the Pluto list?

Normally you can just open the newsgroups list in Pluto. Select the
required newsgroup(s) and then click on the dustbin (as opposed to the 'X')
which will remove them from the list.

I think then, simply clicking on the '+' icon to re-add it is enough. You
shouldn't need to do anything to newshound at all.

HAVING SAID THAT..... are people still getting timeouts? I've cleaned the
cache on the server and extended the local cache by a factor of 10, and
it's certainly sped everything up from this end, so has anyone else noticed
any improvements since about midnight?

[Dr Peter Young]

On 19 Aug 2007 Paul Vigay
<invalid-email-***@invalid-domain.co.uk> wrote:

[snip]

Post by Paul Vigay
HAVING SAID THAT..... are people still getting timeouts? I've cleaned the
cache on the server and extended the local cache by a factor of 10, and
it's certainly sped everything up from this end, so has anyone else noticed
any improvements since about midnight?

Still slow with frequent stops here, but no more timeouts yet. Thanks
for the help over this, and I do now think it's not my setup! (see the
thread on orpheus.acorn.misc).

With best wishes,

Peter.

[Dr Peter Young]

Post by Dr Peter Young
On 19 Aug 2007 Paul Vigay
[snip]

Post by Paul Vigay
HAVING SAID THAT..... are people still getting timeouts? I've cleaned the
cache on the server and extended the local cache by a factor of 10, and
it's certainly sped everything up from this end, so has anyone else noticed
any improvements since about midnight?

Still slow with frequent stops here, but no more timeouts yet. Thanks
for the help over this, and I do now think it's not my setup! (see the
thread on orpheus.acorn.misc).

I wrote too soon; still getting timeouts (set to 6 minutes in
NewsHound) :-(

With best wishes,

Peter.

[John Nolan]

Post by Dr Peter Young
Still slow with frequent stops here, but no more timeouts yet. Thanks
for the help over this, and I do now think it's not my setup! (see the
thread on orpheus.acorn.misc).

I thought it was I who was at fault, too. I felt doubly banished to outer
space, because
1. A check I did showed that I had not received News between the dates of
about 22July - 15August, or later.
The situation had become so grim that I had stopped checking to see if
News had come in, thinking it was my problem. By chance, I noticed that the
no. of articles in zfc was 1 whereas before it had been 0. That alerted me
to the fact that something was working again :))
2. I have an ongoing email problem, detailed in another thread. Since Puul
hadn't replied I thought that my webmail wasn't working either, in the
outward direction.
Now I understand that PV had probably been getting his priorities right
by trying to resolve this ongoing News problem.

John N.

[Tennant Stuart]

Post by Paul Vigay
HAVING SAID THAT..... are people still getting timeouts? I've cleaned the
cache on the server and extended the local cache by a factor of 10, and
it's certainly sped everything up from this end, so has anyone else
noticed any improvements since about midnight?

I got nothing but timeouts until two nights ago, when a huge backlog
of news succesfully arrived. But last night, and again at 6pm this
evneing, I got timeouts again. I don't think it can be anything wrong
at this end if a) others are affected b) it worked for me the once.


Tennant Stuart

[Paul Vigay]

I got nothing but timeouts until two nights ago, when a huge backlog of
news succesfully arrived. But last night, and again at 6pm this evneing,
I got timeouts again. I don't think it can be anything wrong at this end
if a) others are affected b) it worked for me the once.

There's definitely nothing wrong with the server, as I've checked it
numerous times over the past few days, and everything is working as it
should.

The only thing I can see is that it's swamped by XOVER requests, which I
presume are coming from 100s of people all trying to start re-downloading
news.

I'll restrict the number of simultaneous connections to the news server,
which may prevent too many people logging in at the same time, in the hopes
this may stop the server slowing down - even though it still has a load of
under 2.

Paul

[Mr John FO Evans]

Post by Paul Vigay
The only thing I can see is that it's swamped by XOVER requests, which I
presume are coming from 100s of people all trying to start re-downloading
news.
I'll restrict the number of simultaneous connections to the news server,
which may prevent too many people logging in at the same time, in the hopes
this may stop the server slowing down - even though it still has a load of
under 2.

I can confirm the very slow starts and downloads (1cps sometimes).
No better today WED 10.30am

John

[Tennant Stuart]

Post by Paul Vigay

I got nothing but timeouts until two nights ago, when a huge backlog of
news succesfully arrived. But last night, and again at 6pm this evneing,
I got timeouts again. I don't think it can be anything wrong at this end
if a) others are affected b) it worked for me the once.

There's definitely nothing wrong with the server, as I've checked it
numerous times over the past few days, and everything is working as it
should.
The only thing I can see is that it's swamped by XOVER requests, which I
presume are coming from 100s of people all trying to start re-downloading
news.
I'll restrict the number of simultaneous connections to the news server,
which may prevent too many people logging in at the same time, in the
hopes this may stop the server slowing down - even though it still has a
load of under 2.

Can't you just ban everybody not using Voyager, which works occasionally?


Tennant

[Nick Roberts]

Post by Paul Vigay

Post by J Peachey
Hope it doesn't mean the server is busy sending dvds out again.

I think it's caused by everyone's article pointers being invalid. The
last time Newshound successfully fetched news was probably before it
all went pear-shaped. The news server was then down for 15 days or
so.
When it came up, all the articles that Newshound was expecting to
find have probably expired on the server, I think Giganews only keep
articles for 10 days or so.

I'd be very surprised if text newsgroups died so quickly, although
binaries may do so.

Post by Paul Vigay
Therefore, I suspect that Newshound is stalling having to search
through all news articles for the next article ID, which by now will
be invalid.

But surely this would only apply the _first_ time NewsHound grabbed
stuff? Thereafter, it should sort itself out and there would be no
stall.

That said - it seems to be a lot better now.

[Dr Peter Young]

On 19 Aug 2007 Nick Roberts <***@orpheusinternet.co.uk> wrote:

[snip]

Post by Nick Roberts
That said - it seems to be a lot better now.

Better here, but not a lot better. No timeouts since this morning, but
still stop and start, with a lot of stop. Still, I have cunning plan,
maybe, based on something Paul suggested ...

With best wishes,

Peter.

[Nick Roberts]

Post by Dr Peter Young
[snip]

Post by Nick Roberts
That said - it seems to be a lot better now.

Better here, but not a lot better. No timeouts since this morning, but
still stop and start, with a lot of stop. Still, I have cunning plan,
maybe, based on something Paul suggested ...

You're quite right (unfortunately). Back to 30+ sec pauses for each
group.

[Dr J E Hurley]

Post by Nick Roberts
You're quite right (unfortunately). Back to 30+ sec pauses for each
group.

Do not seem to be experiencing any of these problems using dial up and dear
old Voyager .... perhaps because it is so slow anyway ;)