MTU problem, probably server related?

I have done a search and quick results scan, but may have missed
something relevant.

Using 13.1/13.2 (I’m in the process of configuring 13.2) to access
ancestry.co.uk with the router set to MTU 1492 I get no response after
entering the login details. If I use Vbox Win 7 then I get a relatively
slow response; using Palemoon 64bit in both, though Firefox and
Konqueror act similarly.

Depending on the ISP gateway used (they change each time I change MTU),
I sometimes get “pmtu 1476” against one of the “unknown.prolexic.com
entries from “tracepath secure.ancestry.co.uk” (the sub-domain accessed
to confirm login).

Changing the MTU to 1476 in the router results in fast access to most of
the ancestry facilities, but images of pages return the error message
“Oops, we’ve hit a snag. We were unable to retrieve the image. Please
try again.” - again Win7 is unaffected as is 13.2.

The “simple” solution is to finish setting up 13.2, but I’m not convince
it will be a long term solution, as this has only happened in the last 6
months (about the time Ancestry updated their systems).

Ancestry appears to be the only site affected, but it is not clear if it
is a problem at their end or the route (via “unknown.prolexic.com”), as
other US sites use different routes, or with the openSUSE configuration.

Any suggestions for further test or explanations welcomed, including
“have a go at Ancestry” - though they will likely respond “we don’t
support Linux”.


PeeGee

MSI m/b 870-C45, AMD Athlon II X3 445, 8GB, openSUSE 13.2/13.1 KDE
x86_64 dual boot + Win7 Premium 64bit in VBox
Asus m/b M2NPV-VM, AMD 64 X2 3800+, 4GB, openSUSE 13.2 KDE x86_64/Win7
Premium 64bit dual boot

Why are you messing around with MTU?
Unless you’re using very ancient equipment (or newer equipment using ancient parts), all modern devices have standardized on an MTU around 1600 (I don’t remember the exact number off the top of my head).

Either your equipment (or your ISP) would have to be using equipment more than 18 years old (approx) to run into non-standard MTU.

Maybe if you were a big, trunk routing company you might have a different reason to look at MTU but I don’t think that’s your scenario.

My advice:
Reset to defaults and leave alone. If you’re experiencing some kind of throughput issue, look elsewhere for a cause.

TSU

On 15/08/15 23:26, tsu2 wrote:
>
> Why are you messing around with MTU?

Though a year or so old, nearly all articles via a search suggest MTU
problems for slow page access (due to mis-configured servers).

> Unless you’re using very ancient equipment (or newer equipment using
> ancient parts), all modern devices have standardized on an MTU around
> 1600 (I don’t remember the exact number off the top of my head).
>
> Either your equipment (or your ISP) would have to be using equipment
> more than 18 years old (approx) to run into non-standard MTU.
>
> Maybe if you were a big, trunk routing company you might have a
> different reason to look at MTU but I don’t think that’s your scenario.
>

I tried a MTU of 1476 because of the prolexic response - I now see they
provide “anti” DDoS services, presumably used by Ancestry and, possibly
significant, the connection stalls at the login stage.

From Wikipedia: “Prolexic operates a DDoS mitigation platform and a
global network of traffic scrubbing centers”.

> My advice:
> Reset to defaults and leave alone. If you’re experiencing some kind of
> throughput issue, look elsewhere for a cause.
>
> TSU
>
>

I have returned to the default settings.

Thanks for your response. I’m awaiting a response (but not holding my
breath) from Ancestry.


PeeGee

MSI m/b 870-C45, AMD Athlon II X3 445, 8GB, openSUSE 13.2/13.1 KDE
x86_64 dual boot + Win7 Premium 64bit in VBox
Asus m/b M2NPV-VM, AMD 64 X2 3800+, 4GB, openSUSE 13.2 KDE x86_64/Win7
Premium 64bit dual boot

On 16/08/15 09:58, PeeGee wrote:
> On 15/08/15 23:26, tsu2 wrote:
>>
>> Why are you messing around with MTU?
>
> Though a year or so old, nearly all articles via a search suggest MTU
> problems for slow page access (due to mis-configured servers).
>
>> Unless you’re using very ancient equipment (or newer equipment using
>> ancient parts), all modern devices have standardized on an MTU around
>> 1600 (I don’t remember the exact number off the top of my head).
>>
>> Either your equipment (or your ISP) would have to be using equipment
>> more than 18 years old (approx) to run into non-standard MTU.
>>
>> Maybe if you were a big, trunk routing company you might have a
>> different reason to look at MTU but I don’t think that’s your scenario.
>>
>
> I tried a MTU of 1476 because of the prolexic response - I now see they
> provide “anti” DDoS services, presumably used by Ancestry and, possibly
> significant, the connection stalls at the login stage.
>
> From Wikipedia: “Prolexic operates a DDoS mitigation platform and a
> global network of traffic scrubbing centers”.
>
>> My advice:
>> Reset to defaults and leave alone. If you’re experiencing some kind of
>> throughput issue, look elsewhere for a cause.
>>
>> TSU
>>
>>
>
> I have returned to the default settings.
>
> Thanks for your response. I’m awaiting a response (but not holding my
> breath) from Ancestry.
>

Further experimentation showed the problem to be with my router (tp-link
wr1043ndv1), but only in PPPoE mode. Switched to dynamic IP and inserted
another router to make the PPPoE connection (it was easier than
replacing the router and I don’t lose the “benefits”!) and Ancestry was
accessible and fast!


PeeGee

MSI m/b 870-C45, AMD Athlon II X3 445, 8GB, openSUSE 13.2/13.1 KDE
x86_64 dual boot + Win7 Premium 64bit in VBox
Asus m/b M2NPV-VM, AMD 64 X2 3800+, 4GB, openSUSE 13.2 KDE x86_64/Win7
Premium 64bit dual boot