Hello,

On dc., jul. 17 2019, Bernard Rosset via Dng wrote:

> After I stumbled, by chance, on mirror 141.84.43.19 (being part
> of the
> deb.devuan.org pool) being unavailable for a couple of minutes,
> I set a
> script up checking TCP/80 availability for all of mirrors in
> said pool.
>
> The conclusion is clear:
> 1°) 141.84.43.19 is the only mirror suffering from this problem
> 2°) This problem was detected with a high frequency, despite
> relatively
> infrequent checks (1 every 5 min)
>
> For this day only so far, 2019-07-17, at 1600Z, all the failed
> occurences were happening at those times:
> 0050Z 0055Z 0100Z 0105Z 0110Z 0115Z 0120Z 0123Z 0145Z 0150Z
> 0155Z 0200Z
> 0205Z 0230Z 0240Z 0245Z 0250Z 0310Z 0410Z 0420Z 0455Z 0510Z
> 0515Z 0520Z
> 0535Z 0610Z 0615Z 0630Z 0640Z 0645Z 0715Z 0755Z 0800Z 0805Z
> 0825Z 0905Z
> 1035Z 1100Z

Very interesting, we'll raise the issue with the mirror
maintainer, thank you for going the extra mile than just shouting
"it's not working".
This report actually helps.

>
> Hence the following questions:
> a) Am I the only one observing this (ie someone else having set
> such a
> check up with a check frequency relatively close to mine,
> eliminating
> biases of my setup)?

We have somewhat thorough mirror checkers in-place that haven't
detected this, since they are thorough, they can't run too often
as they can result in huge amounts of traffic.

FWIW: I've been running "while; curl; sleep 30" as a similar test
for a good few minutes now and it's been solid.

But maybe it'll be interesting to add simple monitoring at a
connection level same way you ran your tests.
I'll try to setup a thing in the next couple days.

> b) If not and this a confirmed defect, would not it be
> reasonable to
> remove said host from the pool until the maintainer can inspect
> what is
> going on and act on it?

If we get more reports we totally will, so far everything is
"looking good" and all tests pass, but maybe there is indeed
something inherently spotty on the connection and that's what you
are seeing; we'll see if with more data or more reports or when
the maintainer takes a look something changes.
--
Evilham