https://bugs.exim.org/show_bug.cgi?id=2027
Jeremy Harris <jgh146exb@???> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
--- Comment #1 from Jeremy Harris <jgh146exb@???> ---
OK, it breaks the auxiliary connections such as the rspamd one, but not
transport connections, and only if you run a binary built on a modern system
(having TCP_FASTOPEN defined in standard headers) on an old kernel (not having
any TFO support).
Plan is to spot the setsocket() fail in the daemon startup and remember for
child processes. Independent processes (eg. command-line mail sending,
and socket-activated reception) will have to default to off.
On balance, don't try to override a transport's configuration of
hosts_try_fastopen with this knowledge.
--
You are receiving this mail because:
You are on the CC list for the bug.
