diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-08 20:37:50 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-08 20:37:50 +0000 |
commit | c1f743ab2e4a7046d5500875a47d1f62c8624603 (patch) | |
tree | 709946d52f5f3bbaeb38be9e3f1d56d11f058237 /doc/config-debugging.rst | |
parent | Initial commit. (diff) | |
download | knot-resolver-c1f743ab2e4a7046d5500875a47d1f62c8624603.tar.xz knot-resolver-c1f743ab2e4a7046d5500875a47d1f62c8624603.zip |
Adding upstream version 5.7.1.upstream/5.7.1
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
-rw-r--r-- | doc/config-debugging.rst | 35 |
1 files changed, 35 insertions, 0 deletions
diff --git a/doc/config-debugging.rst b/doc/config-debugging.rst new file mode 100644 index 0000000..520c2db --- /dev/null +++ b/doc/config-debugging.rst @@ -0,0 +1,35 @@ +.. SPDX-License-Identifier: GPL-3.0-or-later + +Debugging options +================= + +In case the resolver crashes, it is often helpful to collect a coredump from +the crashed process. Configuring the system to collect coredump from crashed +process is out of the scope of this documentation, but some tips can be found +`here <https://lists.nic.cz/hyperkitty/list/knot-resolver-users@lists.nic.cz/message/GUHW4JSDXZ6SZUAYYQ3U2WWOZEIVVF2S/>`_. + +Kresd uses its own mechanism for assertions. They are checks that should always +pass and indicate some weird or unexpected state if they don't. In such cases, +they show up in the log as errors. By default, the process recovers from those +states if possible, but the behaviour can be changed with the following options +to aid further debugging. + +.. envvar:: debugging.assertion_abort = false|true + + :return: boolean (default: false in meson's release mode, true otherwise) + + Allow the process to be aborted in case it encounters a failed assertion. + (Some critical conditions always lead to abortion, regardless of settings.) + +.. envvar:: debugging.assertion_fork = milliseconds + + :return: int (default: 5 minutes in meson's release mode, 0 otherwise) + + If a process should be aborted, it can be done in two ways. When this is + set to nonzero (default), a child is forked and aborted to obtain a coredump, + while the parent process recovers and keeps running. This can be useful to + debug a rare issue that occurs in production, since it doesn't affect the + main process. + + As the dumping can be costly, the value is a lower bound on delay between + consecutive coredumps of each process. It is randomized by +-25% each time. |