Treat StatelessReset error differently
Categories
(Core :: Networking: HTTP, task, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox88 | --- | fixed |
People
(Reporter: dragana, Assigned: kershaw)
References
Details
(Whiteboard: [necko-triaged])
Attachments
(1 file)
Reporter | ||
Comment 1•3 years ago
|
||
We treat all error from neqo as a protocol error and we exclude domains from HTTP3 if we received such an error. This is done here.
There are some error that should be treated as netwok_interrupt or reset and also we should not exclude the domains from HTTP3. This errors are:
StatelessReset(from neqo (mapping is here)
and NO_ERROR errors (from neqo or from a peer app or transport error: CloseError::TransportError with code 0, CloseError::PeerError witth code 0, CloseError::PeerAppError with code 0x100, and CloseError::AppError with code 0x100).
Reporter | ||
Comment 2•3 years ago
|
||
Error mappings forn neqo to neqo_glue are here and here
this is an example how these error codes are read in necko (c++) : https://searchfox.org/mozilla-central/rev/63fcc3f1a2cc73488d8986f4cf91fce2cd4b7564/netwerk/protocol/http/Http3Session.cpp#1676-1679
Reporter | ||
Comment 3•3 years ago
|
||
We should use NET_ERROR_RESET so that transaction can be restarted if possible.
Assignee | ||
Comment 4•3 years ago
|
||
Updated•3 years ago
|
Pushed by kjang@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/f6eed285f801 Make it possible to restart the transaction when having a StatelessReset error r=necko-reviewers,dragana
Comment 6•3 years ago
|
||
bugherder |
Description
•