CVS log for emstar/devel/sympathy_devel/sympathy_analyze.c
	Help

Request diff between arbitrary revisions / Display revisions graphically / Jump to LXR

Bug fixes

Fixed a bug in localization assignment!

Fixed several bugs resulting in flapping between 2 root causes (bad path to sink, and no neighbors). Also fixed duplicate printouts of node rebooting.

*** empty log message ***

Updated implementation in sympathy directory to reflect paper changes, added average-rssi

Implementation mtches paper and seems to be working

Minor fixes

Added Boilerplate

got rid of LEDs
updated sympathy code (including we can now ping nodes).
current RAM is at 3550

multihop:
  - took out the old way of determining an "old rounnd", and instead
  just use a simple range scheme. If the round number is within
  current-round-3, then it is old, otherwise it is new.
  the justification is that routing messages do not get flooded,
  so a range of 3 should be sufficient to pick out old messages.

  - i dropped the threshold for an acceptable path quality from 50 to
  20. we found that having the threshold so high prevented nodes
  that were farther from the sink (in terms of numbers of hops) from
  every joining the network.

  - reduced the receive beacon buffer from 2 to 1 - reduced RAM usage
  from ~3750 to ~3550

timesynch:
  - reduced the frequency of time-synch packets to 1/hour

  - added the ability to trigger time-synch packets from the
  command line (so even though the frequency is 1/hour, the
  sink can still trigger packets when it wants to).

  - got rid of dse_start_time file creation. instead we just redefine
  an epoch as starting from Jan 1, 2005, and assume all time-stamps
  start from then. of course, this will screw up the start-time
  for all the existing scripts. but this will hopefully be the last time.

  - i moved the sink code over to emstar versus running in EmTos.
  It seems like the initial motivation between running the sink
  timesync code in EmTos was to share code with the timesync
  code on the motes. But ultimately they
  did not end up sharing code, so there was no reason to leave it
  in emTos.
  The initial reason to move it was because I was noticing flaky
  behavior, and lew suggested it may be because he didnt necessarily
  trust the timers in EmTos.

  - changed the code on the mote and the sink to not use the
  hardware clock. it was not very precise, and did not scale
  deterministically to different speeds (i.e. we wanted to make
  the clock 10 ticks/second instead of 100 ticks / second in
  order to get rid of dse_start_time).

Lots of minor fixes to get ready for deployment

Changed struct names

Next try to make this work.

*** empty log message ***

Fixes for James Reserve deployment, upgraded Sympathy.

Code used for final Sympathy camera ready for Sensys!

msg

Made some bug fixes:
- how sympathy processes dse queries
- memory leak in using misc_parse_init (i never called
misc_parse_cleanup).
- added counting of CRC packets.

x

Fixed up some printing stuff.

Fixed a bug in sympathy - where it wasn't storing correct link
information.

Many very key fixes!! now dse instrumentation working properly.

Fixed compiler errors (when doing make ARCH=mica2),
and many warnings.

Factored out routing-agnostic code from routing specific code into a library
        for sympathy.

Added Surge to our tos-contrib repo as well.

Added instrumentation for mintRoute. This involved:
   - separating routing-specific code out from sympathy on the sink side
   - getting MintRoute to work (and writing wrapper code for it)
   - instrumenting MintRoute code on the mote side.

Added sink-side wrapper code for surge.

 Changed interface for reporting metrics - now we use a Smetrics_t struct
         instead of just randombly packing data into the packet.

Rewrote code
separated routing layer specific code out.

Fixes to sympathy's root failure diagnosis.

Added wrapper for surge/mintRoute.

Added a change for dse instrumentation.

Too many changes to enumerate. Many important bug fixes ins ympathy.

Separated the .h file in tos-contrib/sympathy.

Major fix for dse instrumentation with sympathy (packed couple
of structs to get byte alignment right!).

Made many minor fixes to sympathy

Changed interface slightly that exposes statistics to remove a
redundant argument (type) that tha user was setting but didn't need
to.

Small additions for implementing with bayes classification tool (all changes are surrounded by #ifdef USE_BAYES)

1) Added a new component to handle mailing updates. This component
exports a status-device: writing: "device=<path-to-device>:email=<email-address>will result in an email to the email-address every time there is an update
to the provided status-device. This does not work yet because we have not
determined which mail program to use (suggestions?)

2) Added couple extra fields to sympathy-statistics so we can trace
flow of traffic through the network, and hopefully determine where packets
are dropped. This involved adding more instrumentation to TreeDispatch.

3) Fixed a bug with incrementing counters (wasn't handling rollovers
correctly)

4) Fixed a compile error in EmLinkServerM.nc, TimeSynchM.nc

5) Fixed a bug in MultihopM.nc - where the packet type was not getting
set correctly for link packets so DSE was not receiving packets.

Fixes for ping-about, and now no-route-to-sink failure detection is working.

Fixed a bug (mess up emview),
and other minor changes...

Added flags (commented out) for 900MHz mica2.

Added support for emview in sympathy.
Fixed a couple minor bugs in sympathy.

Working with emview now, using the sink as a gateway!
Now the sink creates fake "devices" when it gets metrics from
the nodes and pushes the information to emview. Which
then displays the info!

Prints out the number of periods a failure has continued for.

Added files that should produce generic stats (run/sim files).
Changes to get rid of iter_fai.

Took out concept of fail-iteration.

Fixed it!

Now have a semi working version of printing out generic stats.
at least one bug to work out...

Fixed some printing stuff... added another failure.

*** empty log message ***

Fixed more bugs (critical functional bugs), and changed usage so that
now sympathy_sink is a macro that takes 2 arguments - in order to specify
the number of failure iterations and epoch length.

Fixed memory consumption problem - wasnt a memory leak after all.

Fixed various bugs (memroy leaks, sink can be any id...)

FIxed bug preventing sympathy-sink from aggregating.

Should be working with exported interfaces.

Fixed a bug!!

Fixed congestion issues.

running version.

Good version!

Moved multihop constants from EmStar_Stirngs. Was causing problems
from emstar-code that needed those multihop-constants - because
including EmStar_Strings causes buf*() funcs to be nops.

*** empty log message ***

*** empty log message ***

Working well.
Running overnight.

Working version - node reboot/congestion detected,
node dieing not because packet-dropping is not working yet.

Working code as of tonight.

running tests - seems to be ok

This is a working version - on the ceiling, with fault injection!

*** empty log message ***

Working with fault injection

Working ok!

Kinda working.

working mostly

*** empty log message ***

*** empty log message ***

New / modular sympathy