Permalink
Commits on Sep 18, 2012
  1. debloat: switch to nfq_codel as the default

    Dave Taht committed Sep 18, 2012
  2. debloat: add efq, nfq, ns2 codel models

    Dave Taht committed Sep 18, 2012
Commits on Sep 16, 2012
  1. bind: update to v9.9.1-P3 (CVE-2012-4244)

    Dave Taht committed Sep 16, 2012
  2. quagga: update to head

    mildly better filtering options
    Dave Taht committed Sep 16, 2012
Commits on Aug 26, 2012
  1. debloat: some hacks to reduce memory pressure somewhat

    Totally not the right thing(s), but more work needs to happen
    elsewhere.
    Dave Taht committed Aug 26, 2012
Commits on Aug 23, 2012
Commits on Aug 12, 2012
  1. try 10ms target for VO

    Dave Taht committed Aug 12, 2012
  2. Use a 256 quantum for everything but VO

    Dave Taht committed Aug 12, 2012
Commits on Aug 9, 2012
  1. quagga: update to head

    some security related changes.
    Dave Taht committed Aug 9, 2012
Commits on Aug 1, 2012
  1. Quiet ledbat some more

    Dave Taht committed Aug 1, 2012
  2. target restored to 5ms

    fiddling with target accomplishes very little on the current
    codel implementation, and in fact, hurts.
    Dave Taht committed Aug 1, 2012
Commits on Jul 29, 2012
  1. quagga: enable diversity routing by default

    Dave Taht committed Jul 29, 2012
Commits on Jul 27, 2012
  1. CVE: bind-9.9.1-P2 update

    Dave Taht committed Jul 27, 2012
  2. babelm: update to main branch

    smoothed metrics are now there
    Dave Taht committed Jul 27, 2012
Commits on Jul 22, 2012
  1. quagga: update for babeld stats counters

    Dave Taht committed Jul 22, 2012
Commits on Jul 12, 2012
  1. quagga: update to support diversity routing

    Dave Taht committed Jul 12, 2012
  2. debloat: increase VI,BE queue sizes to 1200

    600 proved to induce drop tail behavior at high rates.
    Arbitrarily doubling.
    Dave Taht committed Jul 12, 2012
Commits on Jul 11, 2012
  1. netperf: update to head

    Dave Täht committed Jul 11, 2012
  2. quagga: add babelz variant

    Dave Taht committed Jul 11, 2012
Commits on Jul 10, 2012
  1. debloat: Pays to update the release

    Grumble.
    Dave Taht committed Jul 10, 2012
  2. debloat: smaller quantums seem to be better

    despite intuition saying we wanted packet aggregation to work better,
    smaller quantums seem to work better.
    Dave Taht committed Jul 10, 2012
  3. quagga: 0.99.17.11

    Dave Taht committed Jul 10, 2012
Commits on Jul 8, 2012
  1. quagga: add config notes for redistribute kernel and rxcost

    I use the rxcost parameter to give a mild bias to the 5.x
    ghz radios. In the test deployment I found it necessary to bump up
    rxcost in proportion to the radio's quality and intent.
    
    so I ended up with rxcost of 1024 and 1023 for wndr3800s
    rxcost of 768 for picostation 2HPs (used by users)
    rxcost 256 for the directional nanostation M5 backbone
    
    This gave the m5 radios preference and only in case of failure
    do things fall back to the 2HPs and wndr3800s. Otherwise
    the m5s would prefer a direct connection to a lousy radio
    vs a 2 hop or more route that used good radios and ethernet.
    
    I'm not saying this is correct, but it worked.
    
    redistribute kernel works correctly for dhcpv4 inserted default
    routes. It does not appear to work correctly for RA inserted
    ipv6 routes, nor has it been tested for other forms of ipv6 routes
    as yet. WIP.
    Dave Taht committed Jul 8, 2012
  2. Buildable babelm package

    Dave Taht committed Jul 8, 2012
  3. quagga: update to head

    Dave Täht committed Jul 8, 2012
  4. Arbitrarily enable ECN on babel packets

    While arguably the wrong thing (babel should do something sane
    upon receipt of a ECT(3)) I am trying to analyze the behavior
    of the multicast queue which is buried deep in the stack,
    far below where it can be seen easily.
    
    It is my intent to monitor codel behavior by (in part) observing
    ECT(0) being asserted on babel packets via an ip6tables rule.
    
    In fact I intended to track all sorts of ECN behavior via iptables
    rules in the upcoming test series.
    Dave Täht committed Jul 8, 2012
  5. Added babelm branch which contains a new smoothed metric algo

    I note that this experimental package also contains a patch
    to mark babel packets as ECN capable, when in reality they
    are not, at present.
    
    In the words of Juliusz:
    
    "that's a horrible kludge.  You're applying AQM to locally-originated data in
    order to compensate for the lack of backpressure, and then falsely
    asserting ECN in order to bypass the AQM policy."
    
    As for the first statement, queues are necessary at multiple layers
    in the stack, so I can think of no effective means of supplying
    backpressure at insertion time that will work. I can think of means of
    supplying congestion related drop indications to babel, but that is as
    yet, unimplemented.
    
    As to the second objection, yes this is bypassing AQM policy, however,
    at some point, using ECN marking in babel could be used for something.
    
    The udp markings can be easily obtained with pktinfo in userspace,
    and I do need to produce a patch for that.
    
    In the interim, the problem I am trying to solve is that when a rate
    change happens in wifi, it is abrupt and can take a long time (with
    the fq_codel aqm rapidly ramping up packet drop) to drop the backlog back to
    a normal level for the new rate. IMHO it is better to have a radio
    continue to function rather that potentially stop entirely because it
    dropped a bunch of babel packets while seeking a new equilibrium.
    Dave Taht committed Jul 8, 2012
Commits on Jul 6, 2012
  1. debloat: More lightly cuddle everything

    14ms was a good goal. 27ms is saner
    Dave Taht committed Jul 6, 2012
  2. quagga: various updates

    Dave Taht committed Jul 6, 2012
Commits on Jul 4, 2012
  1. update to use fq_codel_ll

    This cuts the wifi driver queues down to a bare minimum
    Dave Taht committed Jul 4, 2012
Commits on Jun 30, 2012
  1. quagga: Add rxcost to babeld command set

    One way to hint to old babeld to prefer one radio over an another
    in case of tie, is to set rxcost, which this set of commits
    implements.
    Dave Taht committed Jun 30, 2012
Commits on Jun 29, 2012
  1. quagga: add support for digest length

    This backward incompatible change makes forward compatible
    changes more feasible for authenticated babel.
    Dave Taht committed Jun 29, 2012
Commits on Jun 28, 2012
  1. quagga: Better integrated ND-RA support

    supports the concept of "connected" better now.
    Dave Taht committed Jun 28, 2012