On 18/11/2021 09:18, ma...@apache.org wrote:
This is an automated email from the ASF dual-hosted git repository.
markt pushed a commit to branch 9.0.x
in repository https://gitbox.apache.org/repos/asf/tomcat.git
The following commit(s) were added to refs/heads/9.0.x by this push:
new bbbb7a8 Protect against a known OS bug
bbbb7a8 is described below
commit bbbb7a8c4315bbd5f6b3d620f748cf30a41ba2e9
Author: Mark Thomas <ma...@apache.org>
AuthorDate: Wed Nov 17 18:48:33 2021 +0000
Protect against a known OS bug
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1924298
Hi all,
Back-porting this to 8.5.x is complicated by 8.5.x having the option for
multiple acceptor threads. I can see the following options.
1. Don't back-port this protection to 8.5.x.
===
8.5.x has survived this long without the protection.
There aren't any outstanding 8.5.x bug reports that could be
attributable to this issue.
Having the protection (and log message) in the other branches should
be enough to raise visibility of the bug and get it fixed.
There may be uses that are experiencing this bug on rare occasions
without realising it.
2. Limit 8.5.x to a single acceptor thread and back-port the 9.0.x fix
===
Multiple acceptor threads have been shown to offer little/no
performance benefit.
Making the acceptorThreadCount setter a NO-OP and hard-coding the
getter to return 1 would be relatively simple.
This feels like a fairly big change this late in the 8.5.x series.
3. Introduce a synchronization block around the accept() call and add
the duplicate check inside the synchronization block.
===
There may be a performance impact for users who have configured
multiple acceptor threads.
My current preference order is 2 first, closely followed by 1 with 3
further behind.
Thoughts?
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
