Broker::NetworkInfo$address is of type string. The Broker::peers() BiF treated
it as an address, which caused ZVal's type validation to flag a
mismatch and abort.
Includes a small tweak to the broker.unpeer test that makes it invoke
Broker::peers(), which breaks in absence of the fix.
* origin/topic/vern/ZAM-prep: (45 commits)
whoops overlooked the need to canonicalize filenames
another set of tweaks per review comments
addressed a number of code review comments
baseline updates for merge
support "any" coercions for "-O gen-C++"
better descriptions for named record constructors
test suite baseline updates for "-a opt" optimize-AST alternative
test suite baseline updates for "-a xform" alternative / AST transformation
error propagation fix for AST reduction
updates to "-a inline" test suite alternative baseline
updates for the main test suite baseline
updates to test suite tests for compatibility with upcoming ZAM functionality
"-O compile-all" option to specify compilation of inlined functions
compile inlined functions if they're also used indirectly
provide ZAM-generated code with low-level access to record fields
fix for cloning records with fields of type "any"
direct access for ZAM to VectorVal internal vector
ZVal constructors, accessors & methods in support of ZAM
switch ZVal representation of types from Type objects to TypeVal's
revised error-reporting interface for ZVal's, to accommodate ZAM inner loop
...
- Extended a btest to cover tables/sets with index types
(this was originally broken until fixed by GH-1514)
* origin/topic/jsiwek/gh-1506-fix-broker-func-indices:
GH-1506: Fix Broker unserialization of set/table function indices
Zeek function types are serialized as a broker::vector, but the
unserialization logic for Zeek set/table types that use a function for
as an index incorrectly identified it as a composite-index, which also
use broker::vector, rather than a singleton-index, and makes such
unserialization fail.
A general example where this failure can happen in practice is when
trying to unserialize a connection record for which there's a
Conn::RemovalHook, since that's a set[function], and a specific case of
that is use of the Intel Framework in a Zeek cluster.
Broker::create_master() and Broker::create_clone() now return
a valid value even when there's a failure to open the backend database
(e.g. SQLite filesystem error). In that case, the returned value can
still be passed into other data store operations, but they'll fail
immediately with an error. Broker::is_closed() can now also be used to
determine whether the data store creation calls failed.
This change adds support for complex indexes for sets and tables. With
this change, sets with indexes like:
set[string, count, count]
will function. Before this change, Zeek raised an error message in these
cases.
Addresses GH-1033
There was a race that previously may cause it to not truly test
whether the connecting-process exited due to SSL authentication
failure or just because the listening-process wasn't available yet
(e.g. due to process scheduling variability).
The explicit sorting function definition was a temporary stability
workaround with a better fix now being available via
https://github.com/zeek/btest/pull/25
- Use `-b` most everywhere, it will save time.
- Start some intel tests upon the input file being fully read instead of
at an arbitrary time.
- Improve termination condition for some sumstats/cluster tests.
- Filter uninteresting output from some supervisor tests.
- Test for `notice_policy.log` is no longer needed.
Particularly, the final output order of a table/set is sensitive to
order of input/insertions and some tests were converting
std::unordered_{set,map} to Zeek table/set and iteration over those
standard containers may not always loop through elements in the same
order across all platforms.
Haven't checked different build configurations yet, but all except
a few SumStats tests are stable for me now. The external tests
are also completely failing, but haven't looked at those yet.
* origin/topic/johanna/table-changes: (26 commits)
TableSync: try to make test more robust & add debug output
Increase timeouts to see if FreeBSD will be happy with this.
Try to make FreeBSD test happy with larger timeout.
TableSync: refactor common functionality into function
TableSync: don't raise &on_change, smaller fixes
TableSync: rename auto_store -> table_store
SyncTables: address feedback part 1 - naming (broker and zeek)
BrokerStore <-> Zeek Tables: cleanup and bug workaround
Zeek Table<->Brokerstore: cleanup, documentation, small fixes
BrokerStore<->Zeek table: adopt to recent Zeek API changes
BrokerStore<->Zeek Tables Fix a few small test failures.
BrokerStore<->Zeek tables: allow setting storage location & tests
BrokerStore<->Zeek tables: &backend works for in-memory stores.
BrokerStore<->Zeek table - introdude &backend attribute
BrokerStore<->Zeek tables: test for clones synchronizing to a master
BrokerStore<->Zeek tables: load persistent tables on startup.
Brokerstore<->Tables: attribute conflicts
Zeek/Brokerstore updates: expiration
Zeek/Brokerstore updates: add test that includes updates from clones
Zeek/Brokerstore updates: first working end-to-end test
...
This addresses PR feedback. The main component in this commit is to
disable &on_change notifications when &backend loads a table from sqlite
on startup.
* origin/topic/jsiwek/gh-1024-broker-store-handle-type-checks:
Improve Broker store API's handling of invalid arguments
Add builtin_exception() functions
GH-1024: fix crash on passing wrong types to Broker store API
This was a bit of a bigger merge since Zeek changed inbetween the time
of the PR and me actually merging it.
I put the new functions into the zeek::detail namespace -- since it
seems unlikely that those will be used by something external.
I also renamed them to fit better with the naming scheme of the new
error functions.
Fixes GH-1024
This commit adds script/c++ documentation and fixes a few loose ends.
It also adds tests for corner cases and massively improves error
messages.
This also actually introduces type-compatibility checking and introduces
a new attribute that lets a user override this if they really know what
they are doing. I am not quite sure if we should really let that stay in
- but it can be very convenient to have this functionality.
One test is continuing to fail - the expiry test is very flaky. This is,
I think, caused by delays of the broker store forwarding. I am unsure if
we can actually do anything about that.
With this, the basic functionality of &backend seems to be working.
It is not yet integrated with zeekctl, one has to manually specify the
storage location for the sqlite files somewhere when using sqlite.
Usage for memory stores:
global table_to_share: table[string] of count &backend=Broker::MEMORY;
Usage for sqlite stores:
redef Broker::auto_store_db_directory = "[path]";
global table_to_share: table[string] of count &backend=Broker::SQLITE;
In both cases, the cluster should automatically sync to changes done by
any node. When using sqlite, data should also be saved to disk and
re-loaded on startup.
Currently this requires using this with a normal cluster - or sending
messages by yourself.
It, in principle, should also work with SQLITE - but that is a bit
nonsensical without being able to change the storage location.
The &backend attribute allows for a much more convenient way of
interacting with brokerstores. One does not need to create a broker
store anymore - instead all of this is done internally.
The current state of this partially works. This should work fine for
persistence - but clones are currently not yet correctly attached.
When a clone attaches to a master, it just gets the diffs sent as
events. Which is neat because it means that we pretty much don't need
any extra code to handle this.
This currently only handles the most basic case, and is not thoroughly
tested.
When initializing a master store, we now check if there already is data
in it. If yes, we load it directly into the zeek table when the store is
created. We assume that this is happening at Zeek startup - and are
supremely evil and just load it synchronously. Which could block
execution for a bit for larger stores.
That being said - this might sidestep other issues that would arise when
doing this async (like scripts already inserting data).
Next step: check if this approach also works for clones.
* origin/master:
Fix shadowed variable that breaks lookup_hostname()
GH-1025: allow copying/cloning of `opaque of Broker::Store`
Fix "possibly-truncated" compiler warning in BuildJSON snprintf()
Update submodule(s)
Fixed some places where tabs became spaces
Convert to using permissions to check for access to cirrus variables in benchmark script
Integrate review feedback: improve command-line option redef parsing
Fix several issues with command-line option redefs
Remove last_access_time from TableEntryVal.
Minimize data published for expected FTP data channel analysis
Stricter checking if we have a dns field on the connection being processed
Modified the DNS protocol analyzer to add a new parameter to the dns_request event which includes the DNS query in its original case. Added a policy script that will add the original_case to the dns.log file as well. Created new btests to test both.
Place build file in explicit location for benchmarking to work correctly
cmake: Make musl support more distro agnostic
Update highwayhash submodule to upstream.
GH-998: Fix Reporter::conn_weird() to handle expired connections
Makes some attributes conflict with each other. This also needed the
test to change.
The test is a bit flaky - but I can, for the heck of it, not figure out
why. I am punting that for the future after spending a few hours on it.
* Some methods mistakenly returned a bool instead of QueryResult
when passed an invalid `opaque of Broker::Store` handle.
* Now generates a runtime exception for store_name() and is_closed()
calls that pass an invalid `opaque of Broker::Store` handle as any
returned value can't be reasonably used in any subsequent logic.
* Descriptions of any invalid arguments are now given in the error
message.
This commit fixes a few more loose ends to actually make the
Zeek Table<->brokerstore syncing work. This mostly slightly changes the
TableVal assign/remove operators to prevent loops when a remote change
arrives.
The tests inserts a value into a table on the manager, and it pops out
in a table on a clone - which is the easiest case.
Timeouts are still not handled at all; the behavior when inserting into
a clone is untested.
Logs that got sent sparsely or burstily would get buffered for long
periods of time since the logic to flush them only does so on the next
log write. In the worst case, a subsequent log write could never happen
and cause a log entry to be indefinitely buffered.
This fix introduces a recurring event/timer to simply flush all pending
logs at frequency of Broker::log_batch_interval.
Only used in one event, without any way to use the opaque for anything
else. At this point this just seems like a complication that has no
reason to be there.
This one took me way too long to admit. Values were pushed back on
deserialization - instead of assigned. Meaning they were added to the
end of the already 0-assigned vector.
The mean thing here is that estimation still worked - just merging
resulted in 0. And estimation still was correct because m, V, alpha_m
are enough for this - and those were correctly copied...
With this change, all tests pass.
We need this to sender through Broker, and we also leverage it for
cloning opaques. The serialization methods now produce Broker data
instances directly, and no longer go through the binary formatter.
Summary of the new API for types derived from OpaqueVal:
- Add DECLARE_OPAQUE_VALUE(<class>) to the class declaration
- Add IMPLEMENT_OPAQUE_VALUE(<class>) to the class' implementation file
- Implement these two methods (which are declated by the 1st macro):
- broker::data DoSerialize() const
- bool DoUnserialize(const broker::data& data)
This machinery should work correctly from dynamic plugins as well.
OpaqueVal provides a default implementation of DoClone() as well that
goes through serialization. Derived classes can provide a more
efficient version if they want.
The declaration of the "OpaqueVal" class has moved into the header
file "OpaqueVal.h", along with the new serialization infrastructure.
This is breaking existing code that relies on the location, but
because the API is changing anyways that seems fine.
This adds an internal BiF
"Broker::__opaque_clone_through_serialization" that does what the name
says: deep-copying an opaque by serializing, then-deserializing. That
can be used to tests the new functionality from btests.
Not quite done yet. TODO:
- Not all tests pass yet:
[ 0%] language.named-set-ctors ... failed
[ 16%] language.copy-all-opaques ... failed
[ 33%] language.set-type-checking ... failed
[ 50%] language.table-init-container-ctors ... failed
[ 66%] coverage.sphinx-zeekygen-docs ... failed
[ 83%] scripts.base.frameworks.sumstats.basic-cluster ... failed
(Some of the serialization may still be buggy.)
- Clean up the code a bit more.
