Date: Fri, 2 Oct 2020 17:59:41 +0200 From: Gert Doering Message-ID: <20201002155941.GF976@...1296...> References: <63aacbfb-20a0-de78-fccb-bbf8a2ad4189@...2318...> MIME-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="lYVfafuUkPqz/tKz" Content-Disposition: inline In-Reply-To: <63aacbfb-20a0-de78-fccb-bbf8a2ad4189@...2318...> Subject: Re: [Openvpn-devel] with IPv6 VPN connection, IPv4 traffic does not get router over VPN List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: =?iso-8859-1?Q?Fran=E7ois?= Kooman Cc: "openvpn-devel@lists.sourceforge.net" --lYVfafuUkPqz/tKz Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Hi, On Fri, Oct 02, 2020 at 03:26:29PM +0200, Fran=E7ois Kooman wrote: > After testing connecting over native IPv6 to the VPN server, it turns=20 > out the IPv4 traffic is not routed over the VPN. This worked in older=20 > versions of OpenVPN (2.4.x) but no longer in OpenVPN 2.5rc2. I am=20 > testing with Windows 8.1, but the same was reported on Windows 10. I can reproduce this on FreeBSD. 2020-10-02 17:53:07 NOTE: unable to redirect IPv4 default gateway -- Cannot= obtain current remote host address Apologies for not *testing* this case - I need to improve my client side test rig to better cover the 4 possible "default-gateway" combinations (v4/v6, default-gateway for ipv4/ipv6).=20=20 As an explanation why this was overlooked - *these* client tests run on my desktop machine and basement server, and they run without redirecting gateway to avoid breaking my "normal work" networking.=20=20 The buildslaves run on VMs and have little other network activity, so=20 could very well test this (AND SHOULD), but their config is basically=20 a "slimmed down to the most important system-specific variants" of=20 the client tests on my desktop - we run less tests there because=20 everything takes ages, and the focus is on "this was broken on Solaris at some point, so have test for it" (etc.) - but here, *more* tests, at least on some of the buildslaves, would certainly be better (or twisting some of the existing variants a bit). Now, the server side tests also run between VMs, and the client side for those has about 3 times the number of tests my workstation has -=20 but these are all focused on torturing server code paths (like, "if I do this setenv, the client-connect plugin on the server will stall for 20 seconds, and then FAIL - is this seen on the client?")... Yeah, I'm annoyed at myself, and very happy that Francois found this :-) gert --=20 "If was one thing all people took for granted, was conviction that if you=20 feed honest figures into a computer, honest figures come out. Never doubte= d=20 it myself till I met a computer with a sense of humor." Robert A. Heinlein, The Moon is a Harsh Mistre= ss Gert Doering - Munich, Germany gert@...1296... --lYVfafuUkPqz/tKz Content-Type: application/pgp-signature; name="signature.asc" -----BEGIN PGP SIGNATURE----- iQGcBAEBAgAGBQJfd05sAAoJEB2Cnv7KVigSYRwL/j0wMQ6d1kiSn9ad7j48YarT R98VDsm45dBSRp6v7hGrLc8teAFIC00d9cjWFeAGCR6ve8YQyatqMIQ+XEsaBEdO mkcTlM2XWcJEJC8lsy01Nrr3a3N9EhCp+u0zC+j9Y3dmMrVNRHor/T6IUogZi5mh R2gyPZswqRBPZsSRLw1ixV1nUb4kFf6XozeRTCrAAcq06CPhD3zHsLiXvb3QcxR7 Hl7+zCM1doC7rSlefrzS3cUPZZRT3JaVuEiysik6Te94G6hSVVF3Q1/+PqcJLDew J3AT4ZoWQckrNck2+xa7elJQQ2v5D8mrX1XweEsoIdgj6oAdHj96Wlc/87zgSOAs 7OxU4/P/HQIHG5Q1bUCyjh4ld8vODSc8bcAJROtqCySZRCDLVRWxdqoNfKXwwBy9 iXmrxh59T6sFhC7rZqNLmZ2AJH+IKgUFftpZHchDBOBdwtBlTSkCNpMIrqWfjvvN WH8cQ3AvqWebEMFb9QYMHOWAFYH+NZiFEzFUh/ICgA== =j0q0 -----END PGP SIGNATURE----- --lYVfafuUkPqz/tKz--