> On Tue, Jun 16, 2015 at 1:42 AM, C Anthony Risinger <[email protected]> > wrote: > >> Hello, >> >> Does/can the websocket impl send a last ditch "termination frame" to the >> client after a PONG failure? Seems to be up to client libraries to >> decide >> how long the time out will be until reconnect. >> >> We are stress testing a custom websocket event server in python, and >> software bugs (greenlets running too long) appear to create situations >> where the PONG is handled too late, despite the client sending on time. >> >> The server discos but the client doesn't know that. If we think a >> connection is dead, could/should we blindly attempt sending a term frame >> (2 >> bytes?) before closing the underlying socket? >> > I am referring to this: > > https://tools.ietf.org/html/rfc6455#section-5.5.1 > > It doesn't appear to be happening on PONG failure from examining code, but > I have not yet confirmed via wire scrape/etc. > > Is anyone able to tell me if/when uWSGI produces a CLOSE frame? > > --
Never afair, when there is a socket problem uWSGI closes the connection. Probably ping/pong could make use of the close packet -- Roberto De Ioris http://unbit.com _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
