In message <u3bpk8.p7k.1@ID-201911.user.individual.net> at Mon, 8 May
2023 19:29:55, Frank Slootweg <this@ddress.is.invalid> writes
>J. P. Gilliver <G6JPG@255soft.uk> wrote:
>> In message <u3ahhp$3rkrc$1@dont-email.me> at Mon, 8 May 2023 06:05:45,
>> Paul <nospam@needed.invalid> writes
>> []
>> >You might be able to tell, by testing of a second
>> >computer while the first one is showing failure signs.
>>
>> That would indeed be a good way to eliminate the router. Unfortunately I
>> don't have another computer ...
>
> This remark triggered me to realize that (I think) you're currently
>being forged in the comp.mobile.android newsgroup.
>
> N.B. I think another poster (w....t) is being forged as well.
>
> Sad state of affairs.
>
>[...]

Ah, I remember last time I was on usenet (last July and before) being
told of something similar (I forget which 'groups). As there's nothing I
can think of I can do about it, I ignored it: presumably those who read
both 'groups I post in and ones I don't but the forger does will soon
learn my style, and be able to tell us apart.
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

Feudalism : It's your count that votes.

In message <u36ruj$32dck$1@dont-email.me> at Sat, 6 May 2023 20:38:42,
Paul <nospam@needed.invalid> writes
>On 5/6/2023 6:53 PM, J. P. Gilliver wrote:
>> I don't _think_ it's Chrome that's the cause - but it could be I
>>suppose, if it's got some sort of leak.
>> The time to fail this time was about (log file info) 18:46:29 to
>>23:16:34, or about 4½ hours.
[]
The fault _seems_ to have gone away as mysteriously as it came:
connected time now over 3 days 17 hours.

The only slightly unusual thing I've done is install the paid version of
EaseUS's undelete utility, which involved briefly turning off AVG so it
could register itself. (Turned it back on immediately.) Can't think why
that should have had any effect, good or bad, on a strange connection
fault.
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

You can tell that you are getting old when typing VCR into a document [results
that the spell checker] wants to change it to Vicar.
- Brian Gaff in uk.tech.broadcast, 2016-3-6

On 5/11/2023 10:13 PM, J. P. Gilliver wrote:
> In message <u36ruj$32dck$1@dont-email.me> at Sat, 6 May 2023 20:38:42, Paul <nospam@needed.invalid> writes
>> On 5/6/2023 6:53 PM, J. P. Gilliver wrote:
>>> I don't _think_ it's Chrome that's the cause - but it could be I suppose, if it's got some sort of leak.
>>>  The time to fail this time was about (log file info) 18:46:29 to 23:16:34, or about 4½ hours.
> []
> The fault _seems_ to have gone away as mysteriously as it came: connected time now over 3 days 17 hours.
>
> The only slightly unusual thing I've done is install the paid version of EaseUS's undelete utility, which involved briefly turning off AVG so it could register itself. (Turned it back on immediately.) Can't think why that should have had any effect, good or bad, on a strange connection fault.

https://serverfault.com/questions/131935/network-error-no-buffer-space-available

"This shit still happens in Windows 10 with basically just Firefox, Avast and Putty running"

They refer to the problem as "leaking network buffers".
And also, that netstat does not record anything out of the ordinary.
So the "buffer space" is not related to a currently active
connection, just some sort of connection that needs to be
harvested and returned to the pool.

Paul

In message <u3kmf8$1e5o0$1@dont-email.me> at Fri, 12 May 2023 02:31:04,
Paul <nospam@needed.invalid> writes
>On 5/11/2023 10:13 PM, J. P. Gilliver wrote:
[]
>> The fault _seems_ to have gone away as mysteriously as it came:
>>connected time now over 3 days 17 hours.
[]
>https://serverfault.com/questions/131935/network-error-no-buffer-space-a
>vailable

Thanks: interesting article. The original post says "IE won't open any
new connections", which isn't what I was experiencing - _existing_ ones
(in Chrome and other software) stopped working. But later ones I think
_did_ mention existing things stopping.
>
> "This shit still happens in Windows 10 with basically just Firefox,
>Avast and Putty running"
>
>They refer to the problem as "leaking network buffers".
>And also, that netstat does not record anything out of the ordinary.
>So the "buffer space" is not related to a currently active
>connection, just some sort of connection that needs to be
>harvested and returned to the pool.
>
> Paul
>
The several suggestions of stopping processes until the problem goes
away sounds very tedious. The suggestion (for XP, but 7 is often
similar) of adding MaxUserPort as a new DWORD to
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters,
and then setting it to the max, sounds promising: I have that "folder"
but not that DWORD, so might try that if I get the problem again. I was
a bit puzzled, since he said he was adding it as a new key, how he knew
what the default value (5000, 0x1388) and range (5000-65534) were: I
presume he got them (and the whole idea) from the MS article he cites
the URL of, which of course now doesn't lead to any such information.

Connect time now 4 days 2½ hours (-:
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

"quidquid latine dictum sit, altum viditur". ("Anything is more impressive if
you say it in Latin")

On 12/05/2023 12:37, J. P. Gilliver wrote:
>
> I presume he got them (and the whole idea) from the MS article he cites
> the URL of, which of course now doesn't lead to any such information.

Wayback Machine?

https://archive.org/web/

--

Fake news kills!

I may be contacted via the contact address given on my website:
www.macfh.co.uk

"J. P. Gilliver" <G6JPG@255soft.uk> wrote:

> W7-32; I don't know if that's relevant.
>
> This machine loses _some_ connection after some hours; a restart cures
> it. It only started happening in the last month or so, not from when I
> got it, so it's something I've done - unfortunately I don't know what.
>
> After whatever happens has happened: Many - probably most - websites
> don't work (e. g. BBC Weather, Twitter, Ancestry). Ping works (so I've
> [a] got _some_ connection [b] got DNS). yt-dlp fails (with a page full
> of error messages). BUT: YouTube and Google (the search engine)
> _continue to work_; I can view clips I haven't viewed before, and add
> comments, on YouTube, and I can do a Google search I haven't done
> before, and get pages of results (though mostly, if not entirely,
> clicking on the links in them doesn't work).
>
> Any idea what's going on, and/or where I should look? It seems a very
> odd fault!
>
> Yes, it hasn't escaped my notice that YouTube and Google (and the Chrome
> browser I'm using - latest for W7 I think) are all from the same
> company.

At the time the problem starts where you cannot reach some sites, look
in Event Viewer. Could be a TCP max connection or rate limit setting
which would generate an event error.