]>
git.ipfire.org Git - thirdparty/knot-resolver.git/log
Tomas Krizek [Wed, 8 Jan 2020 13:46:22 +0000 (14:46 +0100)]
Merge branch 'ci-fixes' into 'master'
ci: fixes
See merge request knot/knot-resolver!918
Tomas Krizek [Wed, 8 Jan 2020 13:31:28 +0000 (14:31 +0100)]
tests/pytests: create and collect html report
The report has been previously removed, most likely by a mistake.
Tomas Krizek [Wed, 8 Jan 2020 13:27:46 +0000 (14:27 +0100)]
ci: re-run valgrind test on failure
Vladimír Čunát [Tue, 7 Jan 2020 13:45:09 +0000 (14:45 +0100)]
Merge !911: daemon/main: decrease suggested nofile hard limit to 512k
Closes #476
Tomas Krizek [Mon, 16 Dec 2019 13:15:09 +0000 (14:15 +0100)]
daemon/main: decrease suggested nofile hard limit to 512k
512k (or more) seems to be hard limit valued used by multiple distros
(fedora, arch, ubuntu, debian).
Closes #476
Lukas Jezek [Tue, 7 Jan 2020 12:53:42 +0000 (13:53 +0100)]
Merge branch 'https_client_improvemens' into 'master'
HTTPS client improvemens
See merge request knot/knot-resolver!913
Petr Špaček [Mon, 30 Dec 2019 15:56:11 +0000 (16:56 +0100)]
utils: add human-readable text to HTTP client library errors
Petr Špaček [Mon, 30 Dec 2019 13:31:00 +0000 (14:31 +0100)]
prefill: retry download immendiatelly if cached data cannot be imported
Retry will typically happen if data on disk are corrupted, and in that
case it does not make sense to wait full retry interval.
Petr Špaček [Mon, 30 Dec 2019 13:06:37 +0000 (14:06 +0100)]
prefill: remove zone file from disk on failure
Empty zone file was causing errors after restart and user had to wait to
retry interval before re-download. Let's not polute filesystem with
incomplete files.
Petr Špaček [Mon, 30 Dec 2019 11:16:52 +0000 (12:16 +0100)]
prefill: allow to use system-wide CA store
It also improves error reporting from store:add() call.
Sometimes the error message from lua-ossl is incomplete. This is fixed
by https://github.com/wahern/luaossl/pull/176.
Petr Špaček [Mon, 30 Dec 2019 10:43:03 +0000 (11:43 +0100)]
utils: log errors from HTTPS client library
Previous code inconsistently thrown some errors and returned as string
other ones, so we now return all errors as strings in classic Lua-style.
Petr Špaček [Tue, 7 Jan 2020 10:00:28 +0000 (11:00 +0100)]
Merge branch 'ta-test-update-cert' into 'master'
daemon/trust_anchors.test: Turn off expiration date in certificates
See merge request knot/knot-resolver!916
Lukáš Ježek [Tue, 7 Jan 2020 09:47:24 +0000 (10:47 +0100)]
daemon/trust_anchors.test: Turn off expiration date in certificates
Petr Špaček [Mon, 23 Dec 2019 19:08:34 +0000 (20:08 +0100)]
Merge branch 'doc-quickstart' into 'master'
Quick start documentation
Closes #500, #499, and #498
See merge request knot/knot-resolver!867
Petr Špaček [Mon, 23 Dec 2019 17:30:51 +0000 (18:30 +0100)]
docs: quick start guide: final polish
Ales Mrazek [Thu, 17 Oct 2019 11:52:17 +0000 (13:52 +0200)]
docs: quick start documentation third draft corrections
Ales Mrazek [Mon, 7 Oct 2019 11:37:30 +0000 (13:37 +0200)]
docs: quick start documentation third draft
Ales Mrazek [Thu, 3 Oct 2019 08:02:52 +0000 (08:02 +0000)]
docs: apply suggestion to doc/startguide.rst
Ales Mrazek [Mon, 30 Sep 2019 12:11:35 +0000 (14:11 +0200)]
docs: quick start documentation second draft
Ales Mrazek [Wed, 11 Sep 2019 12:34:25 +0000 (14:34 +0200)]
docs: quick start documentation first draft
Petr Špaček [Fri, 20 Dec 2019 15:11:26 +0000 (16:11 +0100)]
Merge branch 'freebind' into 'master'
daemon/network: add freebind support
See merge request knot/knot-resolver!898
Petr Špaček [Fri, 20 Dec 2019 14:18:00 +0000 (15:18 +0100)]
tests: sanity check for net.listen() and net.list()
Tomas Krizek [Fri, 13 Dec 2019 14:32:29 +0000 (15:32 +0100)]
modules/http: fix net.listen in config tests
Tomas Krizek [Wed, 11 Dec 2019 17:09:46 +0000 (18:09 +0100)]
tests/config: add freebind test
Tomas Krizek [Wed, 11 Dec 2019 17:07:01 +0000 (18:07 +0100)]
bindings/net: throw lua error ofr net.listen() failures
To avoid configuration errors, throw a lua error and crash
if it's not possible to bind as specified in net.listen().
For special use-cases, freebind=true should be used instead.
Tomas Krizek [Wed, 11 Dec 2019 15:30:06 +0000 (16:30 +0100)]
daemon/io: rename to family_to_freebind_option()
Tomas Krizek [Fri, 29 Nov 2019 15:19:10 +0000 (16:19 +0100)]
doc: document freebind option for net.listen()
Tomas Krizek [Fri, 29 Nov 2019 15:04:10 +0000 (16:04 +0100)]
daemon/bindings: add freebind option to net.list()
Tomas Krizek [Fri, 29 Nov 2019 14:24:17 +0000 (15:24 +0100)]
daemon/network: add freebind support
Petr Špaček [Fri, 20 Dec 2019 13:32:33 +0000 (14:32 +0100)]
Merge branch '521-replace-lua-socket-depedency-with-lua-http' into 'master'
replace lua-socket depedency with lua-http
Closes #512 and #521
See merge request knot/knot-resolver!894
Petr Špaček [Thu, 19 Dec 2019 18:10:02 +0000 (19:10 +0100)]
prefill: fix timer restart
event.reschedule() is not a good idea for long-running downloads and in
general I have a bad feeling that it did not work as intended even for
retries after errors.
Lukáš Ježek [Thu, 19 Dec 2019 13:15:56 +0000 (14:15 +0100)]
daemon/lua/kluautil.lua: remove from global namespace
Lukáš Ježek [Thu, 19 Dec 2019 10:28:02 +0000 (11:28 +0100)]
modules/prefill and TA: Move https_fetch to daemon/lua/kluautil.lua
Lukáš Ježek [Tue, 17 Dec 2019 07:59:06 +0000 (08:59 +0100)]
packages: fix dependencies
Vladimír Čunát [Mon, 16 Dec 2019 09:43:12 +0000 (10:43 +0100)]
CI ASAN: switch from gcc to clang
Trying to avoid https://github.com/google/sanitizers/issues/1010
Lukáš Ježek [Fri, 13 Dec 2019 12:06:57 +0000 (13:06 +0100)]
Update NEWS and doc
Lukáš Ježek [Fri, 13 Dec 2019 12:00:37 +0000 (13:00 +0100)]
packages: remove lua-socket
Lukáš Ježek [Fri, 13 Dec 2019 11:24:36 +0000 (12:24 +0100)]
modules/graphite: Use module 'cqueues.socket' instead 'socket'
Lukáš Ježek [Fri, 13 Dec 2019 11:07:15 +0000 (12:07 +0100)]
modules/policy: Use module 'cqueues.socket' instead 'socket'
Lukáš Ježek [Fri, 13 Dec 2019 10:21:11 +0000 (11:21 +0100)]
daemon/lua: add kluautil.lua file for helpers functions
Lukáš Ježek [Fri, 6 Dec 2019 13:19:44 +0000 (14:19 +0100)]
packages: remove lua-sec and add lua-http
Lukáš Ježek [Thu, 5 Dec 2019 14:42:34 +0000 (15:42 +0100)]
daemon/lua/trust_anchors: Use module 'http.request' instead 'ssl.https'
Lukáš Ježek [Thu, 5 Dec 2019 09:35:16 +0000 (10:35 +0100)]
modules/prefill: support large zone file
Lukáš Ježek [Tue, 3 Dec 2019 13:34:00 +0000 (14:34 +0100)]
modules/prefill: Add the functionality to specify a custom CA file
Lukáš Ježek [Fri, 22 Nov 2019 15:18:51 +0000 (16:18 +0100)]
modules/prefill: Remove unused ltn12
Lukáš Ježek [Fri, 22 Nov 2019 14:33:04 +0000 (15:33 +0100)]
modules/prefill: Fetch root zone file asynchronously
Lukáš Ježek [Fri, 22 Nov 2019 14:27:04 +0000 (15:27 +0100)]
modules/prefill: Use module 'http.request' instead 'ssl.https'
Petr Špaček [Thu, 19 Dec 2019 09:06:29 +0000 (10:06 +0100)]
Merge branch 'multiple-config-files' into 'master'
daemon/main: support multiple config files
See merge request knot/knot-resolver!909
Petr Špaček [Thu, 19 Dec 2019 08:51:44 +0000 (09:51 +0100)]
clarify errors from luaL_dofile while loading configs
Tomas Krizek [Fri, 13 Dec 2019 15:40:35 +0000 (16:40 +0100)]
daemon/main: remove "-" from config array
Tomas Krizek [Wed, 11 Dec 2019 15:05:12 +0000 (16:05 +0100)]
daemon/main: log config path and workdir
Tomas Krizek [Wed, 11 Dec 2019 14:24:31 +0000 (15:24 +0100)]
lib/utils: create get_workdir() utility function
Tomas Krizek [Wed, 11 Dec 2019 12:09:35 +0000 (13:09 +0100)]
daemon/main: use engine_loadconf() for default config
Tomas Krizek [Tue, 10 Dec 2019 18:22:16 +0000 (19:22 +0100)]
daemon/main: remove useless l_dosandboxfile macro
The exact same function is implemented as luaL_dofile() in Lua 5.1,
there seems to be no reason to use our project-specific macro for it.
https://www.lua.org/manual/5.1/manual.html#luaL_dofile
Tomas Krizek [Tue, 10 Dec 2019 17:58:56 +0000 (18:58 +0100)]
daemon/main: support multiple config files
Petr Špaček [Wed, 18 Dec 2019 15:20:42 +0000 (16:20 +0100)]
Merge branch '520-prefill-remove-depedency-on-lua-filesystem-lfs' into 'master'
prefill: remove depedency on lua-filesystem (lfs)
Closes #520
See merge request knot/knot-resolver!912
Lukáš Ježek [Wed, 18 Dec 2019 14:13:56 +0000 (15:13 +0100)]
prefill: remove depedency on lua-filesystem (lfs)
Tomas Krizek [Thu, 12 Dec 2019 16:27:46 +0000 (17:27 +0100)]
Merge branch 'ci-obs-buildall' into 'master'
ci: allow failure of obs:build:all
See merge request knot/knot-resolver!910
Tomas Krizek [Thu, 12 Dec 2019 16:20:40 +0000 (17:20 +0100)]
ci: allow failure of obs:build:all
This job tends to fail very often, but very frequently due to issues
with OBS itself - outside of our control. The output of the job can
still be useful to check manually, e.g. before releases.
Tomas Krizek [Thu, 12 Dec 2019 16:19:29 +0000 (17:19 +0100)]
ci: document confusing allow_failure: false
This value seems to be the default, but it is important to have
it set explicitly, otherwise when: manual actions could be skipped
https://docs.gitlab.com/ee/ci/yaml/#whenmanual
Petr Špaček [Thu, 12 Dec 2019 14:27:37 +0000 (15:27 +0100)]
Merge branch 'lua_gc' into 'master'
lua: stop trying to tweak lua's GC
See merge request knot/knot-resolver!201
Tomas Krizek [Wed, 9 Oct 2019 12:29:28 +0000 (14:29 +0200)]
lua: stop trying to tweak lua's GC
cherry-picked from
f0ca89ac , original author Vlada Cunat
TL;DR: I believe all lua_gc() calls stemmed from misunderstanding lua
documentation, and the current settings seem potentially dangerous.
First, let me rely on lua 5.1 docs, as luajit 2 is documented to have
done only minor changes in the GC.
http://www.lua.org/manual/5.1/manual.html#lua_gc
http://wiki.luajit.org/New-Garbage-Collector#rationale
Commit
5a709411 claims to have increased the speed of GC to 400 % of
speed of allocation, but LUA_GCSETSTEPMUL is the parameter that
controls that, and that one was lowered to 99 % and later in
0ee2d1d7 even to 50 %. Documentation explicitly says that setting
the value under 100 % may cause problems.
The default values seem perfectly sane to me and currently I can't see
any particular reason to change them. It's 200 % relative GC speed,
and waiting for allocated size to double before starting another cycle.
I assume the resulting possibility of GC being too slow caused the need
to explicitly force a non-incremental GC cycle once in a while, but
that seems not useful anymore and not good for latency.
Petr Špaček [Wed, 11 Dec 2019 10:02:59 +0000 (11:02 +0100)]
Merge branch 'reuseport-freebsd' into 'master'
daemon/io: use SO_REUSEPORT_LB if available (FreeBSD 12.0+)
See merge request knot/knot-resolver!907
Vladimír Čunát [Tue, 10 Dec 2019 16:08:30 +0000 (17:08 +0100)]
daemon/io: use SO_REUSEPORT_LB if available (FreeBSD 12.0+)
and don't use SO_REUSEPORT on non-Linux. (Free)BSD has a different
meaning for it, which only brings confusion - only the last instance
would be getting packets.
Petr Špaček [Wed, 11 Dec 2019 09:48:22 +0000 (10:48 +0100)]
Merge branch 'tty-logging' into 'master'
daemon/ tty commands: don't log unless --verbose
Closes #528
See merge request knot/knot-resolver!908
Vladimír Čunát [Tue, 10 Dec 2019 17:04:15 +0000 (18:04 +0100)]
daemon/ tty commands: don't log unless --verbose
It's minimalistic: no change if in interactive or --verbose mode.
Petr Špaček [Mon, 9 Dec 2019 15:44:02 +0000 (16:44 +0100)]
Merge branch 'systemd-instance' into 'master'
systemd: add env variable SYSTEMD_INSTANCE
See merge request knot/knot-resolver!906
Tomas Krizek [Fri, 6 Dec 2019 15:20:12 +0000 (16:20 +0100)]
systemd: add env variable SYSTEMD_INSTANCE
Tomas Krizek [Fri, 6 Dec 2019 15:00:50 +0000 (16:00 +0100)]
Merge branch 'ci-update' into 'master'
ci: updates
See merge request knot/knot-resolver!905
Tomas Krizek [Fri, 6 Dec 2019 12:34:39 +0000 (13:34 +0100)]
ci: update respdiff jobs
Tomas Krizek [Fri, 6 Dec 2019 11:48:55 +0000 (12:48 +0100)]
ci: allow odvr release in tag pipelines
Tomas Krizek [Wed, 4 Dec 2019 14:28:25 +0000 (15:28 +0100)]
Merge branch 'release-4-3-0' into 'master'
update NEWS, bump to 4.3.0
See merge request knot/knot-resolver!904
Tomas Krizek [Wed, 4 Dec 2019 13:19:47 +0000 (14:19 +0100)]
update NEWS, bump to 4.3.0
Tomas Krizek [Wed, 4 Dec 2019 13:59:21 +0000 (14:59 +0100)]
Merge branch '518-confidential-issue' into 'master'
Resolve "RRset merge operation is too slow for big RRsets"
Closes #518
See merge request knot/knot-resolver!903
Tomas Krizek [Wed, 4 Dec 2019 13:41:16 +0000 (14:41 +0100)]
NEWS: update
Petr Špaček [Wed, 4 Dec 2019 08:39:45 +0000 (09:39 +0100)]
doc: clarify upgrade instructions for modules
Vladimír Čunát [Thu, 21 Nov 2019 14:28:47 +0000 (15:28 +0100)]
lib/utils kr_ranked_rrarray_add(): clarify merging RRs
Petr Špaček [Wed, 20 Nov 2019 15:05:08 +0000 (16:05 +0100)]
ci: skip MacOS tests in security repo
Vladimír Čunát [Wed, 13 Nov 2019 13:07:46 +0000 (14:07 +0100)]
iterate: better efficiency on huge RRsets
- written relatively defensively - act OK even if the API
isn't used in an ideal way
- CI lint:scan-build: bump the error count;
It's only another instance of the mis-detected array_push().
- the removed stale note in modules/meson.build isn't really related
Tomas Krizek [Wed, 4 Dec 2019 13:25:41 +0000 (14:25 +0100)]
Merge branch 'cname-limit' into 'master'
iterate: fix limit on CNAME chain length
See merge request knot/knot-resolver!899
Petr Špaček [Wed, 4 Dec 2019 07:05:42 +0000 (08:05 +0100)]
ci: skip Travis build for security repo
Petr Špaček [Tue, 3 Dec 2019 14:09:47 +0000 (15:09 +0100)]
iterate: clarify error messages about CNAME chains
Vladimír Čunát [Mon, 2 Dec 2019 15:00:40 +0000 (16:00 +0100)]
iterate: tests for CNAME chain restrictions
Vladimír Čunát [Fri, 29 Nov 2019 16:03:41 +0000 (17:03 +0100)]
iterate: reduce CNAME length limit: 40 -> 13
Unbound has limit 10, and practically useful numbers are way lower.
Vladimír Čunát [Fri, 29 Nov 2019 15:15:10 +0000 (16:15 +0100)]
iterate: fix limit on CNAME chain length
The accounting was just broken and overly messy anyway.
Vladimír Čunát [Wed, 4 Dec 2019 09:57:18 +0000 (10:57 +0100)]
Merge !896: daemon: support dropping capabilities
Tomas Krizek [Wed, 27 Nov 2019 14:08:11 +0000 (15:08 +0100)]
NEWS: mention dropping capabilities
Tomas Krizek [Wed, 27 Nov 2019 14:03:51 +0000 (15:03 +0100)]
distro/*: add libcap-ng dependency
Tomas Krizek [Wed, 27 Nov 2019 11:55:06 +0000 (12:55 +0100)]
systemd/nosocket: use capabilities
Tomas Krizek [Tue, 26 Nov 2019 11:37:45 +0000 (12:37 +0100)]
daemon/main: add libcap-ng support to drop capabilities
Petr Špaček [Mon, 2 Dec 2019 12:55:10 +0000 (13:55 +0100)]
Merge branch 'perf-lua-ffi_cleaned' into 'master'
performance: lua-related improvements
See merge request knot/knot-resolver!874
Vladimír Čunát [Wed, 14 Aug 2019 15:16:58 +0000 (17:16 +0200)]
modules/ta_signal_query: optimize
Basically the same as the parent commit (just much simpler).
Vladimír Čunát [Wed, 14 Aug 2019 14:53:23 +0000 (16:53 +0200)]
modules/ta_sentinel: optimize
When all lua modules get unloaded, this change makes the module's
contribution to QPS unmeasurable (for me), saving a few percent.
The point is to almost always return very cheaply, in particular without
creating any lua GC object (like FFI for kr_query).
Note: some checks didn't make much sense, so I improved those as well.
Vladimír Čunát [Mon, 12 Aug 2019 14:08:37 +0000 (16:08 +0200)]
modules/policy: optimize special domain processing
Running the full special-domain checks is relatively expensive.
Vladimír Čunát [Wed, 31 Jul 2019 13:55:43 +0000 (15:55 +0200)]
modules/policy: optimize postrules
I've never seen anyone use postrules.
Vladimír Čunát [Wed, 7 Aug 2019 07:36:13 +0000 (09:36 +0200)]
lua FFI: avoid frequent usage of lua_pushpointer()
The new way of transitioning to layer callbacks - done because of
portability (mainly to aarch64) - is a bit expensive. This is a simple
way of recovering that cost. Merge
603a24f regressed speed a bit.
Petr Špaček [Thu, 28 Nov 2019 09:06:17 +0000 (10:06 +0100)]
Merge branch 'packaging-fixes' into 'master'
Packaging fixes
See merge request knot/knot-resolver!895
Tomas Krizek [Wed, 27 Nov 2019 11:46:06 +0000 (12:46 +0100)]
systemd: add proper User/Group
The Group= settings was ommited and default group of User= was
implicitly used. Now the group set at build time is respected.
GC didn't respect the user/group set at build time at all.
Tomas Krizek [Tue, 26 Nov 2019 11:58:02 +0000 (12:58 +0100)]
distro/*: http module requires the same knot-resolver version
Previously, it was possible to update just "knot-resolver" even
when the "knot-resolver-module-http" package was installed, or the
other way around.
Tomas Krizek [Wed, 27 Nov 2019 14:57:22 +0000 (15:57 +0100)]
Merge branch 'ci-timeout' into 'master'
ci: increase test timeout
See merge request knot/knot-resolver!897