Josh Soref authored 3 years ago and Tomas Krizek committed 3 years ago

Signed-off-by: Josh Soref <jsoref@users.noreply.github.com>

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

It's special: always on and not shown in log_groups() output.

It's been quite a long fight to find how to best deal with such
a special case (from user perspective; code itself is easy).

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

Let's be consistent witk kr_log_name2level()
and even generally we tend to use negative numbers for errors.

Use a single call as a setter and getter for log groups.

Refactor into more consistent lua API.

Using a single function to get/set values is more consistent with our
existing lua API rather than having two separate set and get functions.

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

It's mainly in tests.

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

- add lua set_log_target() + docs
- default to 'stderr' (manual runs)
- switch to 'syslog' in distro-preconfig.lua
- a few minor tweaks

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

Probably, it seems "more consistent".
Some defines still don't have it, but I left those.

Vladimír Čunát authored 3 years ago and Tomas Krizek committed 3 years ago

- const for names of log groups
- enum kr_log_group for a single log group
- the kr_log_groups bitmap doesn't need to be exposed or even exported
- return bool instead of int

Plural makes more sense to me here, since it can take a table with
multiple entries as input.

To ensure all log groups have been properly defined, iterate over the
entire enum. If a value is missing, assert will be triggered.

Also fixes groups with number >=32 (32bit value was used to test
whether group is set before).

The function returns a table where key is the group name and the value
is either true (when logging selected group), or nil (when group isn't
selected).

This replaces the list_log_group() which didn't really work.

Ensure both these functions return string that represents the log level.

Also avoid using printf - it'd probably break API when using map().

To (hopefully) improve readability, rename the typical macro usage of:

    if (!kr_assume(x)) y; // to
    if (kr_fails_assert(x)) y;

As a convenience, replace the assert without a return value to a more
simple version:

    (void)!kr_assume(x);  // becomes
    kr_assert(x);

- add lua options debugging.assumption_abort and
  debugging.assumption_fork to make the behaviour configurable
- select default values based on meson buildtype:
  - debug/debugoptimized (default, e.g. used in CI) - abort and don't
    fork
  - plain/release - don't abort, but fork if abort is user-configured

It's quite cheap for us, and it might help with dumb clients
overusing the lowest IP from each set.

"ctx" is the usual name elsewhere for a kr_context pointer

Štěpán Balážik authored 5 years ago and Vladimír Čunát committed 4 years ago

Design discussion: #447
Code discussion: !1030

Lukas Jezek authored 4 years ago and Tomas Krizek committed 4 years ago

This change allows map() to work with systemd integration.

As a bonus the new client implementation is based on Lua cqueues
allows caller to wrap map() in worker.corroutine() and get
asynchronous execution/avoid blocking main loop.

Currently socket communication does not employ timeouts so a hang
instance will lead to hang map() call. This does not affect query
processing _if_ map() is being run in worker.corroutine.

Fixes: #554
Fixes: #620

Vladimír Čunát authored 4 years ago and Tomas Krizek committed 4 years ago

Two styles were used: (un)defined and 0/1.  We switch to 0/1.
Advantage: it can be used also like `if (ENABLE_FOO)`
(outside preprocessor).

Except for ./meson.build it's sed \
 -e 's/#ifdef ENABLE_CAP_NG/#if ENABLE_CAP_NG/g' \
 -e 's/#ifdef ENABLE_DOH2/#if ENABLE_DOH2/g' \
 -e 's/defined(ENABLE_COOKIES)/ENABLE_COOKIES/g' \
 -e 's/#ifdef ENABLE_COOKIES/#if ENABLE_COOKIES/g' \
 -i $(git grep -l ENABLE_)

(Tiny nitpicks addressed by vcunat.)

Lots of lines affected, but it gets slightly simpler.

In particular this gets rid of last light user data inside kresd.

It was still causing problems on some systems, for example Debian Sid.
The error was the same: "bad light userdata pointer" from luajit,
but note that the problem can still be triggered by lua libraries,
e.g. cqueues.

Vladimír Čunát authored 4 years ago and Tomas Krizek committed 4 years ago

s/gitlab\.labs\.nic/gitlab.nic/g
Redirects are in place, so it shouldn't be required now, but why not.

Formerly multiple instances could use the same seed,
which prevented the retry logic in Lua modules (e.g. prefill) from
retrying at different times.

AFAIK security impact is zero aside from potential thundering-herd
problem with many kresd instances.

Tomas Krizek authored 4 years ago and Vladimír Čunát committed 4 years ago

From our TCP benchmarks, values over 128 don't seem to have any
measurable benefits, even with hundreds of thousands of connections.

On the contrary, during very high TCP and CPU load, smaller backlog
seems to dramatically improve latency for clients that keep idle TCP
connections.

During normal/low load, smaller backlog doesn't seem to have any
benefits.

When measured against "aggressive" clients that immediately close the
TCP connection once their query is answered, backlog smaller than 128
was measured to hurt performance.

The application's backlog size is ultimately limited by
net.core.somaxconn, which has been set to 128 prior to Linux 5.4.
Therefore, this change only affects newer kernels and those who have
manually set this value to a higher size.

For more, see
https://gitlab.labs.nic.cz/knot/knot-resolver/-/merge_requests/968

Petr Špaček authored 5 years ago and Vladimír Čunát committed 5 years ago

Long GNU GPLv3 boilderplate was automatically replaced
with machine readable tag.