#3341 NORM Untriag: Failure to transfer large files over mesh interface while associated

[Zarro Boogs per Child]

#3341: Failure to transfer large files over mesh interface while associated
-----------------------+----------------------------------------------------
  Reporter:  jcardona  |       Owner:  jcardona                         
      Type:  defect    |      Status:  new                              
  Priority:  normal    |   Milestone:  Untriaged                        
 Component:  wireless  |     Version:  Development build as of this date
Resolution:            |    Keywords:                                   
  Verified:  0         |  
-----------------------+----------------------------------------------------

Comment(by jcardona):

 Further tests:

 Left both xo's associated to the AP and not sending or receiving network
 traffic.
 After 1 hour both xo's were still associated.

 Rerun the sftp tests, and this time around they would not fail  (could the
 previous association losses be caused by unfinished processes left pending
 by NM?):

 {{{
 bash-3.2#  while true; do sftp 169.254.7.208:bf10M; done
 Connecting to 169.254.7.208...
 Fetching /root/bf10M to bf10M
 /root/bf10M                                   100%   52MB   1.0MB/s
 00:52
 Connecting to 169.254.7.208...
 Fetching /root/bf10M to bf10M
 /root/bf10M                                   100%   52MB   1.0MB/s
 00:50
 Connecting to 169.254.7.208...
 Fetching /root/bf10M to bf10M
 /root/bf10M                                   100%   52MB   1.1MB/s
 00:48
 Connecting to 169.254.7.208...
 Fetching /root/bf10M to bf10M
 /root/bf10M                                   100%   52MB 960.0KB/s
 00:55
 }}}

 While running the tests, started a scanning on one of the nodes:

 {{{
 # while true; do iwlist eth0 scan ; done
 }}}

 that stalled the ftp transfer:

 {{{
 Fetching /root/bf10M to bf10M
 /root/bf10M                                     5% 3104KB  62.3KB/s -
 stalled -
 /root/bf10M                                     5% 3104KB  50.5KB/s -
 stalled -
 /root/bf10M                                     5% 3104KB   0.0KB/s -
 stalled -
 /root/bf10M                                     5% 3104KB   0.0KB/s -
 stalled -
 /root/bf10M                                     5% 3104KB   0.0KB/s -
 stalled -Killed by .
 }}}

 Stopping the scans did not resume the transfer, but restarting sftp did
 work.  I don't think sftp can recover from broken TCP connections.

 The same results were observed with {{{scp}}}.

 {{{
 #  while true; do scp 169.254.7.208:bf10M . ; done
 bf10M                                          17% 9436KB 598.1KB/s
 01:12 ETA
 bf10M                                          30%   16MB 730.2KB/s
 00:50 ETA
 [scanning loop started here]
 bf10M                                          33%   17MB 531.5KB/s -
 stalled -
 bf10M                                          33%   17MB  16.4KB/s -
 stalled -
 bf10M                                          33%   17MB   8.7KB/s -
 stalled -Killed by .
 }}}

 ... but rsync is robust enough to handle the loss of TCP connection caused
 by the scans:

 {{{
 # while true; do rsync --progress --compress-level=0169.254.7.208:bf10M ;
 done
     37752520  69%    5.91MB/s    0:00:02
     54067200 100%    5.84MB/s    0:00:08 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  5227.94 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
 [scan loop started here]
     54067200 100%    2.38MB/s    0:00:21 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  2104.75 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     32517896  60%    2.11MB/s    0:00:09
     54067200 100%    2.22MB/s    0:00:23 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  2220.08 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     54067200 100%    2.60MB/s    0:00:19 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  2656.82 bytes/sec
 total size is 54067200  speedup is 667.22
 [scan loop stopped here]
 bf10M
     54067200 100%    5.91MB/s    0:00:08 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  5588.48 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     54067200 100%    5.91MB/s    0:00:08 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  5588.48 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     54067200 100%    5.91MB/s    0:00:08 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  5227.94 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     54067200 100%    5.91MB/s    0:00:08 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  5588.48 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     14630480  27%    3.62MB/s    0:00:10
 [scan loop started here]
     54067200 100%    1.14MB/s    0:00:45 (xfer#1, to-check=0/1)

 sent 51527 bytes  received 29506 bytes  1573.46 bytes/sec
 total size is 54067200  speedup is 667.22
 bf10M
     54067200 100%    2.02MB/s    0:00:25 (xfer#1, to-check=0/1)

 }}}

-- 
Ticket URL: <https://dev.laptop.org/ticket/3341#comment:2>
One Laptop Per Child <https://dev.laptop.org>
OLPC bug tracking system