View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0002338GNUnetcore servicepublic2012-05-09 22:252012-05-10 12:59
ReporterChristian Grothoff Assigned ToChristian Grothoff Due Date
PriorityurgentSeveritymajorReproducibilityhave not tried
Status closedResolutionunable to reproduce 
Product VersionGit master 
Summary0002338: ERROR Assertion failed at gnunet-service-core_sessions.c:740.
DescriptionOriginally reported in 0002321, but that bug originally dealt with a different concern. Thus opening here as a separate issue.
TagsNo tags attached.

Activities

Christian Grothoff

2012-05-10 07:26

manager   ~0005848

On the 740 assertion: we seem to have gotten (and transmitted) a session key, but not a PONG confirming our session key from the other side; thus our state is most likely "KX_STATE_KEY_RECEIVED" (which means the session is "down"). We still decrypt, and then the session module wonders why it is getting messages from a peer that is not really "UP". Causes could be message loss (UDP?), multiple transports plus congestion, etc. Question is what to do:

1) delay processing (wait for PONG, then trigger)
2) drop all of these messages prior to PONG (typemap should be periodically refreshed anyway, just somewhat bad for performance if it hits typemap often)
3) ???

Christian Grothoff

2012-05-10 09:05

manager   ~0005852

Actually, that can't quite be as we DO delay processing (see kx->emsg_received assignment).

Issue History

Date ModifiedUsernameFieldChange
2012-05-09 22:25Christian GrothoffNew Issue
2012-05-09 22:25Christian GrothoffStatusnew => assigned
2012-05-09 22:25Christian GrothoffAssigned To => Christian Grothoff
2012-05-10 07:26Christian GrothoffNote Added: 0005848
2012-05-10 09:05Christian GrothoffNote Added: 0005852
2012-05-10 12:59Christian GrothoffStatusassigned => closed
2012-05-10 12:59Christian GrothoffResolutionopen => unable to reproduce
2012-05-10 12:59Christian GrothoffTarget Version0.9.3 =>