https://bz.apache.org/bugzilla/show_bug.cgi?id=60956
--- Comment #8 from Luca Toscano <toscano.l...@gmail.com> ---
Hi Frank,
everybody is a bit busy with the upcoming 2.4.26 release but this bug will be
addressed, I don't have the necessary skills but I'll find somebody soon enough
:)
I tried to check event's code and everything seems originating from the
periodical call to the following snipped of the listener (as it was stated
previously):
/* Step 2: write completion timeouts */
process_timeout_queue(write_completion_q, timeout_time,
start_lingering_close_nonblocking);
This is now done periodically by the listener, and when Timeout expires then
start_lingering_close_nonblocking ends up in blocking as you described.
Since in this case mod_ssl will have to send a close-notify to gracefully close
the TLS connection, I'd think that the listener should not have the chance to
even attempt to do any work (risking to block), but just offload it to a spare
worker.
An alternative would be to figure out if modssl_smart_shutdown could avoid to
block or just for a very brief amount of time to avoid blocking the listener.
We'll see how others thinks about it! I'll dig a bit more into this issue and
report my findings.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscr...@httpd.apache.org
For additional commands, e-mail: bugs-h...@httpd.apache.org
