OpenBSD Journal

multipath support added on -CURRENT

Contributed by jcs on from the which-way-to-go dept.

Eduardo Alvarenga writes: "Jun-ichiro itojun Hagino has commited an important addon: multipath routing.
CVSROOT:        /cvs
Module name:    src
Changes by:     itojun@cvs.openbsd.org  2004/04/24 20:48:04

Modified files:
sys/conf : files sys/kern : vfs_subr.c sys/net : pf_table.c radix.c radix.h route.c route.h rtsock.c sys/netinet : in_proto.c sys/netinet6 : in6_proto.c Added files: sys/net : radix_mpath.c radix_mpath.h Log message: radix tree with multipath support. from kame. deraadt ok user visible changes: - you can add multiple routes with same key (route add A B then route add A C) - you have to specify gateway address if there are multiple entries on the table (route delete A B, instead of route delete A) kernel change: - radix_node_head has an extra entry - rnh_deladdr takes extra argument TODO: - actually take advantage of multipath (rtalloc -> rtalloc_mpath)
It seems there will be much more awesome features in the next-next 3.6 release ;-)
"

(Comments are closed)


Comments
  1. By j knight (68.146.204.200) enabled@myrealbox.com on

    An additional commit was made a little while later:
    Date: Sun, 25 Apr 2004 14:02:40 -0600 (MDT)
    To: source-changes@cvs.openbsd.org
    From: Jun-ichiro itojun Hagino 
    Subject: CVS: cvs.openbsd.org: src
    
    CVSROOT:        /cvs
    Module name:    src
    Changes by:     itojun@cvs.openbsd.org  2004/04/25 14:02:39
    
    Modified files:
            sys/net        : radix.c
    
    Log message:
    permit multiple default route.  from kame
    

    woot! :)

    Comments
    1. By Dan (80.178.120.94) on

      How does it works? How the OS knows which default gateway is valid? IBM 2210 is usin ping to check the next hop (verify next_hop), and it's great tool or that. I guess ifstated will do the trick.

  2. By Anthony (192.208.10.217) on

    Linux is the Solaris/Windows killer. FreeBSD is the Linux killer.

    We now know our place. We are the Cisco killer.

    Comments
    1. By anonymous (203.108.15.199) on

      you've got to be kidding...

  3. By m (195.137.120.127) on

    It would be great to get some kind of heads up about the overall features planned for 3.6 - naturally they don't all make it for numerous reasons, but it would be informative to know.

    Comments
    1. By Otto (82.197.192.49) otto@drijf.net on

      We like to announce new features the moment they are done and when the release date is coming close: the song and feature list of a new release is made available some time before the actual release date.

      Following -current should get you a pretty close picture of what will be in the next release. Once in a while you'll see posts here and on misc@ to draw attention to new stuff.

      Anything more will just distract developers from the real thing: hacking. With all the distractions developers are already subjected to, I don't think we need more.

  4. By Tony S (62.84.195.38) on

    Hmmmm... BGP support Multipath support It looks like someone is interested in replacing some cisco gear.

    Comments
    1. By Brad (216.138.200.42) brad at comstyle dot com on

      There are other vendors besides Cisco and OSPF also supports multipath equal cost routing.

      Comments
      1. By Anonymous Coward (66.114.79.48) on

        OSPF is not a vendor

        Comments
        1. By Jared Yanovich (192.88.114.251) on

          Try reading it like this. "There are other vendors besides Cisco, and OSPF also supports multipath equal cost routing."

    2. By Jon (67.153.107.130) on http://slbd.sf.net

      i've been trying to replace some cisco and non-cisco load-balancing gear at my place of employment with this project:

      http://slbd.sf.net

      thus far it's working well, although some of the more advanced functionality isn't there right now.

      jon

    3. By Anonymous Coward (66.56.80.196) on

      Why not replace that cisco gear? With the recent TCP problem that was discussed at cansec west and the resulting advisories from cisco which pretty much covers their whole product offering the time couldn't be better. In particular because it was shown that OpenBSD was "less" impacted by this than other O.S. types and specific vendor O.S.'s i.e. cisco. Also take into consideration that OpenBSD is heavily used as an edge device it just makes sense to add this and other similar improvements that have been added recently. This is a great opportunity for our O.S. to shine and I'm glad the guys seem to be going for it!

  5. By Jim (198.62.124.245) on

    For those not that close to networking, what does this get you? What value does it provide? Can you help us out with a layman description?

    Comments
    1. By tedu (128.12.75.69) on

      it lets your isp replace their expensive routers with cheap openbsd boxen, so your bill will go down. :)

  6. By Eduardo Alvarenga (200.227.95.57) on

    Well, if multipath is OK, how about link aggregation?
    I think it is a MUST.
    Linux does and FreeBSD does. Is it planned?

    Comments
    1. By Brad (216.138.200.42) brad at comstyle dot com on

      I see you forgot to provide your diff to add LACP (802.3ad). The feature will be added as soon as you provide the URL to your diff.

    2. By Hroi (2001:1448:80:7d::3) on

      FreeBSD does not.

      Comments
      1. By strgout (65.70.175.249) on

        wrong ng_one2many ng_fec

        Comments
        1. By Brad (216.138.200.42) brad at comstyle dot com on

          ng_one2many is not the same thing as link aggregation proper. ng_fec is Cisco proprietary EtherChannel. Add netgraph support for LACP (802.3ad) and then I'll be interested.

  7. By Justin Lindberg (130.94.162.64) justin@srvo.net on

    I was just trying to figure out how to add another default route, and I am quite surprised to learn that I can't (hopefully just until 3.6). Even MS Windows lets you specify more than one gateway for a workstation. Multipath routing isn't just another "awesome" feature; it's an absolute necessity for a reliable internet.

    Way to go!

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]