On 05/03/2021 18:58, Gert Doering wrote:
Hi,
On Fri, Mar 05, 2021 at 06:52:39PM +0000, tincanteksup wrote:
I do not require help with my network, I only want to know what
the current situation is regarding --fragment. Is it no longer
supported for Windows ?
--fragment is totally OS-independent, so "supported everywhere".
So, what is the problem with --fragment in Windows ?
The error:
FRAG_IN error flags=0xfa2a187b: FRAG_TEST not implemented
???
Well, I can grep in the source code as well as you can...
I can grep it but not grok it ..
The code says
else if (frag_type == FRAG_TEST)
{
FRAG_ERR("FRAG_TEST not implemented");
}
(unconditionally, so not "not supported on Windows", but "nowhere)
um .. so, not supported at this time ?
so this seems to be something half-implemented to eventually use this
for "can you confirm that it works?" or something.
My guess on why this happens is about as good as yours - I would guess
that the other end is creating a malformed packet, which can not be
properly decoded, so the receiver thinks "it is a FRAG_TEST packet!"
and then complains that it doesn't know what to do with it.
The server has errors about comp-lzo:
Bad compression stub decompression header byte: 0
This error occurs precisely as the packet from the client using
--fragment arrives.
I notice the comment on trac:
Comment (by plaisthos):
can you test if ncp-disable on both sides makes a difference in behaviour?
This might a side effect of our sub optimal frame recalculation on NCP.
I remember something about ciphers and frame calculations being
flaky-ish.
I'll experiment some more and see what's what ..
Why it is not possible to decode? Hard to say. Maybe something external
is breaking packets (and this is why you have problems without --fragment
in the first place). I can only guess.
There are definitely network problems, most likely MTU.
Following the manual to use --fragment & --mssfix is not
currently an option due to this error.
It appears that I will have to implement a full MTU based
solution.
Another possible solution is to roll back the versions of
OpenVPN in use.
Things look a little dicey for 2.5 when taken out of its comfort
zone..
Thanks anyway..
_______________________________________________
Openvpn-users mailing list
Openvpn-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openvpn-users
