search

fmadio / public

fmadio issue tracking
MIT License
8 stars 3 forks source link

1sec split seq validation problems (fmad80) #190

Closed fmadio closed 6 years ago

fmadio commented 7 years ago

using fmad80 capture splits aren`t working correctly.

remote_download_split regression fails completely with a single 1x40G port. The full capture is ok, but the splits are being calculated incorrectly. Likely the chunk metadata is getting messed up somewhere e.g. during flush.

fmadio commented 7 years ago

updated the 40G packet generator for a configurable start sequence. regression with the 20G system uses this to fragment the capture so its not one continous data rate. e.g. capture .. idle .. capture .. idle but the final output is still a continus sequence

fmadio commented 7 years ago

packet generator can now set the sequence number enabling split regression to run. splits are now good.. mostly. some unrelated bugs it seems still persist

fmadio commented 6 years ago

its likely the same bug as 20v2 where the last chunk flush timestamp was set to 0. causes the last split to output invalid data

fmadio commented 6 years ago

would have showed up in the regressions. closing