• [gentoo-dev] Fwd: open pull requests [was on netifrc]

    From Jaco Kroon@21:1/5 to All on Wed Aug 28 08:50:01 2024
    This is a multi-part message in MIME format.
    Hi All,

    I've got two PRs open on netifrc project, one of which is becoming more
    and more of a pain for us, I'd prefer to push it upstream into netifrc
    since more people can benefit from it than just us, but I'm struggling
    to get that done so may have to revert to pushing it into our local repositories and projects instead.

    Hoping for some advise on how to proceed.

    Kind regards,
    Jaco



    -------- Forwarded Message --------
    Subject: open pull requests
    Date: Thu, 22 Aug 2024 09:06:58 +0200
    From: Jaco Kroon <jaco@uls.co.za>
    Organization: Ultimate Linux Solutions (Pty) Ltd
    To: netifrc@gentoo.org



    Hi,

    Would someone please be able to look at the open pull requests for netifrc?

    I'm in more and more need of https://github.com/gentoo/netifrc/pull/56
    in particular, the other PR I'm much less worried about, and was a
    "simple" drive-by based on the suggestion from the referenced bug.

    I've deployed same to one or two servers already, but we really want it
    more widespread.  Due to the way ipv6 routing works configuring that by
    hand is a massive PITA, thus the only effective solutions (which is
    really much more convenient than IPv4) is to:

    1.  Rely on DHCPv6 (requires a dhcp client to be running, something I
    prefer to avoid in DC);
    2.  Rely on RAs from the gateways (yea, plural), the host can then self-configure using EUI64 or the iptoken mechanism, MAC addresses can
    (and does) change, tokens can be configured which gives predictability.

    Combined with other changes since last release may warrant a new point
    release as well (wireguard improvements, as well as non-device routes - something which we've done in a custom netifrc script for quite some
    time already, including routing rules, and specific to net.lo, so we
    loaded these as unreachable_routes=(...) and prohibit_routes=(...), and route[46]_rules=(...) in conf.d/net, happy to share these, they'll
    conflict a bit with [1] as in it's providing multiple mechanisms to get
    the same job done.  I think it would be good for Gentoo to standardise
    the mechanisms.  I do like the idea of just adding non-device routes to routes_lo=, that said, I've often wondered about just externalising
    non-devices routes and routing rules out of netifrc handling completely
    into it's own init script which does depend() { before net.lo; }.  This
    is a separate discussion though.

    Kind regards,
    Jaco

    1. https://github.com/gentoo/netifrc/commit/7c6a8de0c521ea474bccb0dbda4338ff293cdfc6

    <!DOCTYPE html>
    <html>
    <head>

    <meta http-equiv="content-type" content="text/html; charset=UTF-8">
    </head>
    <body>
    <p>Hi All,</p>
    <p>I've got two PRs open on netifrc project, one of which is
    becoming more and more of a pain for us, I'd prefer to push it
    upstream into netifrc since more people can benefit from it than
    just us, but I'm struggling to get that done so may have to revert
    to pushing it into our local repositories and projects instead.</p>
    <p>Hoping for some advise on how to proceed.<br>
    </p>
    <p>Kind regards,<br>
    Jaco<br>
    </p>
    <div class="moz-forward-container"><br>
    <br>
    -------- Forwarded Message --------
    <table class="moz-email-headers-table" cellspacing="0"
    cellpadding="0" border="0">
    <tbody>
    <tr>
    <th valign="BASELINE" nowrap="nowrap" align="RIGHT">Subject:
    </th>
    <td>open pull requests</td>
    </tr>
    <tr>
    <th valign="BASELINE" nowrap="nowrap" align="RIGHT">Date: </th>
    <td>Thu, 22 Aug 2024 09:06:58 +0200</td>
    </tr>
    <tr>
    <th valign="BASELINE" nowrap="nowrap" align="RIGHT">From: </th>
    <td>Jaco Kroon <a class="moz-txt-link-rfc2396E" href="mailto:jaco@uls.co.za">&lt;jaco@uls.co.za&gt;</a></td>
    </tr>
    <tr>
    <th valign="BASELINE" nowrap="nowrap" align="RIGHT">Organization:
    </th>
    <td>Ultimate Linux Solutions (Pty) Ltd</td>
    </tr>
    <tr>
    <th valign="BASELINE" nowrap="nowrap" align="RIGHT">To: </th>
    <td><a class="moz-txt-link-abbreviated" href="mailto:netifrc@gentoo.org">netifrc@gentoo.org</a></td>
    </tr>
    </tbody>
    </table>
    <br>
    <br>
    Hi,<br>
    <br>
    Would someone please be able to look at the open pull requests for
    netifrc?<br>
    <br>
    I'm in more and more need of
    <a class="moz-txt-link-freetext" href="https://github.com/gentoo/netifrc/pull/56">https://github.com/gentoo/netifrc/pull/56</a> in particular, the other
    PR I'm much less worried about, and was a "simple" drive-by based
    on the suggestion from the referenced bug.<br>
    <br>
    I've deployed same to one or two servers already, but we really
    want it more widespread.  Due to the way ipv6 routing works
    configuring that by hand is a massive PITA, thus the only
    effective solutions (which is really much more convenient than
    IPv4) is to:<br>
    <br>
    1.  Rely on DHCPv6 (requires a dhcp client to be running,
    something I prefer to avoid in DC);<br>
    2.  Rely on RAs from the gateways (yea, plural), the host can then
    self-configure using EUI64 or the iptoken mechanism, MAC addresses
    can (and does) change, tokens can be configured which gives
    predictability.<br>
    <br>
    Combined with other changes since last release may warrant a new
    point release as well (wireguard improvements, as well as
    non-device routes - something which we've done in a custom netifrc
    script for quite some time already, including routing rules, and
    specific to net.lo, so we loaded these as unreachable_routes=(...)
    and prohibit_routes=(...), and route[46]_rules=(...) in
    conf.d/net, happy to share these, they'll conflict a bit with [1]
    as in it's providing multiple mechanisms to get the same job
    done.  I think it would be good for Gentoo to standardise the
    mechanisms.  I do like the idea of just adding non-device routes
    to routes_lo=, that said, I've often wondered about just
    externalising non-devices routes and routing rules out of netifrc
    handling completely into it's own init script which does depend()
    { before net.lo; }.  This is a separate discussion though.<br>
    <br>
    Kind regards,<br>
    Jaco<br>
    <br>
    1.
    <a class="moz-txt-link-freetext" href="https://github.com/gentoo/netifrc/commit/7c6a8de0c521ea474bccb0dbda4338ff293cdfc6">https://github.com/gentoo/netifrc/commit/7c6a8de0c521ea474bccb0dbda4338ff293cdfc6</a><br>
    <br>
    </div>
    </body>
    </html>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)