The $SYSTEMTESTTOP shell variable if often set to .. in various shell
scripts inside bin/tests/system/, but most of the time it is only
used one line later, while sourcing conf.sh. This hardly improves
code readability.
$SYSTEMTESTTOP is also used for the purpose of referencing
scripts/files living in bin/tests/system/, but given that the
variable is always set to a short, relative path, we can drop it and
replace all of its occurrences with the relative path without adversely
affecting code readability.
There were several problems with rbt hashtable implementation:
1. Our internal hashing function returns uint64_t value, but it was
silently truncated to unsigned int in dns_name_hash() and
dns_name_fullhash() functions. As the SipHash 2-4 higher bits are
more random, we need to use the upper half of the return value.
2. The hashtable implementation in rbt.c was using modulo to pick the
slot number for the hash table. This has several problems because
modulo is: a) slow, b) oblivious to patterns in the input data. This
could lead to very uneven distribution of the hashed data in the
hashtable. Combined with the single-linked lists we use, it could
really hog-down the lookup and removal of the nodes from the rbt
tree[a]. The Fibonacci Hashing is much better fit for the hashtable
function here. For longer description, read "Fibonacci Hashing: The
Optimization that the World Forgot"[b] or just look at the Linux
kernel. Also this will make Diego very happy :).
3. The hashtable would rehash every time the number of nodes in the rbt
tree would exceed 3 * (hashtable size). The overcommit will make the
uneven distribution in the hashtable even worse, but the main problem
lies in the rehashing - every time the database grows beyond the
limit, each subsequent rehashing will be much slower. The mitigation
here is letting the rbt know how big the cache can grown and
pre-allocate the hashtable to be big enough to actually never need to
rehash. This will consume more memory at the start, but since the
size of the hashtable is capped to `1 << 32` (e.g. 4 mio entries), it
will only consume maximum of 32GB of memory for hashtable in the
worst case (and max-cache-size would need to be set to more than
4TB). Calling the dns_db_adjusthashsize() will also cap the maximum
size of the hashtable to the pre-computed number of bits, so it won't
try to consume more gigabytes of memory than available for the
database.
FIXME: What is the average size of the rbt node that gets hashed? I
chose the pagesize (4k) as initial value to precompute the size of
the hashtable, but the value is based on feeling and not any real
data.
For future work, there are more places where we use result of the hash
value modulo some small number and that would benefit from Fibonacci
Hashing to get better distribution.
Notes:
a. A doubly linked list should be used here to speedup the removal of
the entries from the hashtable.
b. https://probablydance.com/2018/06/16/fibonacci-hashing-the-optimization-that-the-world-forgot-or-a-better-alternative-to-integer-modulo/
Make sure bin/tests/system/run.sh returns a non-zero exit code if any of
the following happens:
- the test being run produces a core dump,
- assertion failures are found in the test's logs,
- ThreadSanitizer reports are found after the test completes,
- the servers started by the test fail to shut down cleanly.
This change is necessary to always fail a test in such cases (before the
migration to Automake, test failures were determined based on the
presence of "R:<test-name>:FAIL" lines in the test suite output and thus
it was not necessary for bin/tests/system/run.sh to return a non-zero
exit code).
While the creation and publication times of the various keys
in this policy are nearly at the same time there is a chance that
one key is created a second later than the other.
The `set_keytimes_algorithm_policy` mistakenly set the keytimes
for KEY3 based of the "published" time from KEY2.
this changes most visble uses of master/slave terminology in tests.sh
and most uses of 'type master' or 'type slave' in named.conf files.
files in the checkconf test were not updated in order to confirm that
the old syntax still works. rpzrecurse was also left mostly unchanged
to avoid interference with DNSRPS.
it is now an error to have two primaries lists with the same
name. this is true regardless of whether the "primaries" or
"masters" keywords were used to define them.
as "type primary" is preferred over "type master" now, it makes
sense to make "primaries" available as a synonym too.
added a correctness check to ensure "primaries" and "masters"
cannot both be used in the same zone.
This test ensures that named will correctly shutdown
when receiving multiple control connections after processing
of either "rncd stop" or "kill -SIGTERM" commands.
Before the fix, named was crashing due to a race condition happening
between two threads, one running shutdown logic in named/server.c
and other handling control logic in controlconf.c.
This test tries to reproduce the above scenario by issuing multiple
queries to a target named instance, issuing either rndc stop or kill
-SIGTERM command to the same named instance, then starting multiple rndc
status connections to ensure it is not crashing anymore.
Implement the 'rndc dnssec -status' command that will output
some information about the key states, such as which policy is
used for the zone, what keys are in use, and when rollover is
scheduled.
Add loose testing in the kasp system test, the actual times are
already tested via key file inspection.
The wait until zones are signed after rndc reconfig is broken
because the zones are already signed before the reconfig. Fix
by having a different way to ensure the signing of the zone is
complete. This does require a call to the "wait_for_done_signing"
function after each "check_keys" call after the ns6 reconfig.
The "wait_for_done_signing" looks for a (newly added) debug log
message that named will output if it is done signing with a certain
key.
Add a note why we don't have a test case for the issue.
It is tricky to write a good test case for this if our tools are
not allowed to create signatures for unsupported algorithms.
if tests that take a particularly long time to complete
(serve-stale, dnssec, rpzrecurse) are run first, a parallel
run of the system tests can finish 1-2 minutes faster.
these keywords were added to the parser as synonyms for "master"
and "slave" but were never hooked in to the configuration of named,
so they were ignored. this has been fixed and the option is now
checked for correctness.
Due to the changes introduced by the Automake migration, system tests
requiring Python (chain, pipelined, qmin, tcp), dynamic loading of
shared objects (dlzexternal, dyndb, filter-aaaa), or LMDB (nzd2nzf)
currently do not work on Windows. Temporarily disable them on that
platform by moving them from the PARALLEL_COMMON list to the
PARALLEL_UNIX list until the situation is rectified.
Without SYSTEMTESTTOP=.. lines in tests.sh scripts, SYSTEMTESTTOP is
being set to an absolute path. On Windows, this means that an absolute
Cygwin path gets passed as a command line argument to native Windows
binaries, which cannot work and causes system tests to break. Fix by
passing SYSTEMTESTTOP through cygpath on Windows, which causes that
variable to be set to an absolute "mixed mode" path (Windows path with
forward slashes).
Make various adjustments necessary to enable "make dist" to build a BIND
source tarball whose contents are complete enough to build binaries, run
unit & system tests, and generate documentation on Unix systems.
Known outstanding issues:
- "make distcheck" does not work yet.
- Tests do not work for out-of-tree source-tarball-based builds.
- Source tarballs are not complete enough for building on Windows.
All of the above will be addressed in due course.
DS records only belong at delegation points and if present
at the zone apex are invariably the result of administrative
errors. Additionally they can't be queried for with modern
resolvers as the parent servers will be queried.
When ./run.sh <test> is invoked, it acts as a wrapper around
`env - TESTS="<test>" make -e check` to preserve the ability to build
files defined only in the `check` target. Unfortunately, cleaning the
full environment had a side-effect of some tests failing due to missing
binaries and libraries. We now preserve the two most important
variables - PATH and LD_LIBRARY_PATH.
Originally, the default value for max-stale-ttl was 1 week, which could
and in some scenarios lead to cache exhaustion on a busy resolvers.
Picking the default value will always be juggling between value that's
useful (e.g. keeping the already cached records after they have already
expired and the upstream name servers are down) and not bloating the
cache too much (e.g. keeping everything for a very long time). The new
default reflects what we think is a reasonable to time to react on both
sides (upstream authoritative and downstream recursive).
The logic in `keymgr_key_has_successor(key, keyring)` is flawed, it
returns true if there is any key in the keyring that has a successor,
while what we really want here is to make sure that the given key
has a successor in the given keyring.
Rather than relying on `keymgr_key_exists_with_state`, walk the
list of keys in the keyring and check if the key is a successor of
the given predecessor key.
This improves keytime testing on CSK rollover. It now
tests for specific times, and also tests for SyncPublish and
Removed keytimes.
Since an "active key" for ZSK and KSK means something
different, this makes it tricky to decide when a CSK is
active. An "active key" intuitively means the key is signing
so we say a CSK is active when it is creating zone signatures.
This change means a lot of timings for the CSK rollover tests
need to be adjusted.
The keymgr code needs a slight change on calculating the
prepublication time: For a KSK we need to include the parent
registration delay, but for CSK we look at the zone signing
property and stick with the ZSK prepublication calculation.
Registration delay is not part of the Iret retire interval, thus
removed from the calculation when setting the Delete time metadata.
Include the registration delay in prepublication time, because
we need to prepublish the key sooner than just the Ipub
publication interval.
