Allen said:
I get this occasionally when I try to read posts. Why would posts be
no longer available? I tried to find this one in Google but still
nothing.
Al.
----------------------------------------------------------------
Subject: Re: Changing MB and CPU W7
From: Allen Drake <
[email protected]>
Date: Sun, 08 Jan 2012 05:53:59 -0500
Lines: 117
[Sorry. This message is no longer available.]
Message-ID: <
[email protected]>
nntp://news60.forteinc.com/alt.windows7.general/65482
----------------------------------------------------------------
Not a Windows 7 issue. To discuss news servers, visit the
alt.free.newsservers group (even if using a paid one since the community
there is more familiar than this one about operation of NNTP servers).
There are 2 databases, not 1, that have to get synchronized to each
other: overview database (headers) and articles database (bodies). Then
there are spooled articles that are lost if the server crashes without
flushing (updating the databases with the) spooled articles. Could be
the overview and articles databases are out of sync. Users (the ones
who submitted the posts) can cancel their own posts (if cancels are
honored by the server). You would have to ask your NNTP provider is
they honor cancels direct from their users or when syncing to their
peered NNTP servers. A cancel requests removal of the article. If the
article was posted with your same NNTP server, you can look in the
control.cancel newsgroup to see if the poster requested a cancel.
Syncing of the overview database is not immediate. Instead the sync
gets scheduled so you'll have to ask your NNTP server admin how often
they sync their overview and articles databases. Some only do it once
per day. Cancels are not the only cause for the overview and articles
databases to get out of sync. Spooled articles will get lost if the
NNTP server crashes; see
http://www.eyrie.org/~eagle/faqs/inn.html,
section 4.4 as an example. I've seen some NNTP servers undergo
maintenance or upgrades and it takes a day or two before the full sync
and usually results from user complaints initiating the NNTP admin to
perform a pre-scheduled sync. If the Usenet provider is running an NNTP
server farm (especially someone like EasyNews who is a 3rd tier provider
contracting with Highwinds for NNTP service), there can be a delay
between [incremental] syncing of the NNTP servers.
All NNTP servers have a fixed amount of storage space. Despite the ever
increasing size of hard disks, any implementation using them will have a
fixed maximum storage capability. Plus there's the problem of bandwidth
to supply all those articles. Giganews has a retention of somewhere
around 5 years (for text groups; I don't know what it is for binary
groups but expect it to be far shorter). Google Groups can only go back
as far as Deja News archives went back and, I think, that was back to
1982 (although Google has become pretty bad in the last 3 years so
expect to discover a lot of missing articles, or articles so impossible
to find even if they are recorded that you won't find them using
Google's inadequate search algorithms). EasyNews, your Usenet provider,
probably has whatever Highwinds has for retention of articles. When an
article gets expired, you cannot retrieve it anymore. However, as noted
above, it can be several hours before the overview database gets
synchronized to the articles database, so your newsreader can find the
headers (from the overview database) for the expired article but cannot
retrieve the body (from the articles database). From a quick glance at
the EasyNews web site, retention is 1200 days (212 days if you use their
webnews-for-boobs interface). I think that's malarky. Highwinds, the
Usenet backbone provider they use and who is really providing the Usenet
service, has a retention of 1080+ days but only the last 500 days are
available in the header index (overview database). That means your
newsreader can only get the last 500 days of headers and not the 1080+
days that are retained on the server.
How often is "occasionally"? Does the error disappear after a day?
You'll have to ask your Usenet provider how often they sync their
overview and articles databases. It's a batch update to get the
overview database in sync to the articles database after cancels have
changed the articles database. Because of the overhead involved and
impact to the server, and because most NNTP servers don't honor cancels,
the batch op doesn't get run immediately on every articles database
update or even every hour but something like 4 or 24 hours. It's all up
to the NNTP admin how update they run a batch update to ensure the
overview database (just the headers for fast access by the majority of
accesses) is synchronized to the articles database. I don't remember
which but when I asked one NNTP admin about this he said that the batch
sync gets run once per day. EasyNews has no control over when the sync
occurs because they are a 3rd tier Usenet provider contracting with
Highwinds for the Usenet backbone service.
One possible reason why the Google Groups search failed is if the poster
used the "X-No-Archive: Yes" header (or put it in the first line of
their body), the article gets deleted after an indeterminate amount of
time *if* the NNTP server honors the request. Google expires articles
with this header that are over 6 days old. Other NNTP servers that may
honor this header (they are rare) can specify any expiration period they
want. Since this is rude behavior by the poster to punch holes in a
thread, my expiration is 0 days (i.e., immediate deletion). Hey, the
poster declared their message had no value for it to stick around but
they don't get to configure anyone's NNTP server or client as to the
expiration period. They want others to honor their cancel request to
delete their self-proclaimed worthless article so I oblige and expire
immediately. The header only requests that their article be removed.
It doesn't state when their article gets removed.
It's also possible your NNTP client's message store is corrupted. You
would have to reset the newsgroup or unsubscribe, clear any remnant, and
resubscribe to recollect all the article headers. How the reset or
fresh subscribe gets done depends on the NNTP client. However, the
message you described is probably issued by the NNTP server, not by your
NNTP client although a log might show from where the error originated.
Folks over in the Forte newsgroup might better know how that NNTP client
works (alt.usenet.offline-reader.forte-agent), like using a debug log,
since this is not an issue with any version of Windows.
You won't find a search result hunting for articles in Google Groups for
this newsgroup. Google doesn't index the alt.windows7.general group.
See the result from
http://groups.google.com/group/alt.windows7.general
or go to
http://groups.google.com/ and enter "alt.windows7.general" in
the "Search for a group" textbox. If the article was cross-posted to
another group then you can find the article (in another newsgroup) but
not if it was posted only to this newsgroup.
