[Cerowrt-devel] plans and irc meeting monday

[Dave Taht]

I don't have much of a regular schedule these days, but it seems like
nearly every monday from 9AM to 1PM (whatever time zone I'm in,
currently PDT), I end up in #bufferbloat on irc, fixing bugs
discussing things that happened over the week, and what will happen in
the coming week.

Usually there are a ton of openwrt commits on sunday nights, so I try
to get a an integration cycle with cerowrt and build done by monday
morning as well (I rarely succeed before the afternoon).

This past week

- Kept up with openwrt. I'm not happy that IRQF_SAMPLE_RANDOM was just
removed, it was helpful. 3.4 and later has some nice improvements as
to randomness but...
+ I worked on acknowledging all (27!) financial donors/subscribers
thus far to the project in the cerowrt credits file, as well as
acknowledging more of the corps/institutions that have donated
resources and people to the effort. Draft of that is up at:

http://cero2.bufferbloat.net/cerowrt/credits.html

(if I missed you let me know privately. Several people have not
re-donated/re-subscribed after the amazon fiasco, also. Still credited
as I appreciate the attempt even more than the money!)

+ Pushed out the current ubnt-3.3 code I plan to test tomorrow to
github and binaries to:

http://snapon.lab.bufferbloat.net/~cero1/ubnt/3.3/3.3.8-27/

- Didn't make any progress on the outstanding buggy packages (help?
anybody that can build openwrt for anything can fix miniupnpd and
minidlna. I pushed the latest miniupnpd into ceropackages, and it
breaks in easy-to-fix looking new ways)

This week:

* I am trying to get *something* up more suitable for simulated and
actual long-RTT testing than snapon. I snagged an x86 box that can do
it, ran smack into a now fixed Linux 3.6 bug.
( for those interested: http://www.spinics.net/lists/netdev/msg212500.html )

IF I get it going it will become io.lab.bufferbloat.net, or something
like "test.sf.bufferbloat.net" on thursday

Similarly will try to update the bloatlab to current code on thursday

Moving up very high on my list is developing a specification for a
decent set of bloat-related tests rather than the ad-hoc stuff we do
now.

* Trying to push out stable code to the directional radios tuesday at
the latest so I can keep running tests while away
http://www.bufferbloat.net/projects/cerowrt/wiki/Yurtlab

* Might get a new hd or two for the build systems at isc

Coming up in the next 45 days:

* Getting a needed projects outline done
* KVM version? vmware?
* I will be in the Philly area Oct 12-Oct 25th. I have no idea if any
cerowrt users are in the area besides esr and bmc. I'll be heads down
establishing a new lab there. It seems likely a new cerowrt version
will pop out in the middle/tail end of that, with at minimum the 6to4
(bug #406) problem fixed.

* I MAY be near NYC the weeks following, or back in california. Or both.
* I had originally planned to be in Spain for the (recently concluded)
wirelesssummit.org conference. Aside from the ongoing cashflow issue,
I have a major scheduling conflict in that I'd like to make ietf 85 in
atlanta, and the linux wireless conference in spain, nov 4-9.
Obviously I can't make both and at the moment it's looking like I'll
make neither. I had some follow-on work in Paris in conjunction with
lincs.fr lab on the torrent-related horizontal standing queue problem
that fq_codel seems to have, too.

I'm looking forward to the end of the openwrt freeze and jumping
forward to 3.6 or later as soon as possible. That said, it remains my
intent to have a stable 3.3.8 release
for at least several months. So far reports on "sugarland" have been
almost universally positive...

-- 
Dave Täht

Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.html