Currently outstanding_global_views values are only decremented during
the end of epoch cleanup, but not when handle_end_of_result_collection
is called for the specific uid that actually triggered the result
collection (which is specifically NOT a cleanup event).
This changes outstanding_global_views values to be a set of outstanding
uids, instead of a count. This allows handle_end_of_result_collection
to remove any uids from the set as it sees them.
Previously, recent_global_view_keys was only tracked on workers causing
a popular key to be sent up and handled by the manager once for each
worker.
This records the key inside recent_global_view_keys on the manager after
the first update, making the rest of the updates no-ops.
Additionally, since the counter value was never used, it has been
changed from a table to a set.
This adds an event that is raised once Catch & Release ceases the
block management for an IP address because the IP has not been seen in
traffic during the watch interval.
This allows users who use their own logic on the top of catch and
release know when they will have to start re-blocking the IP if it
occurs in traffic again.
This change introduces error events for Table and Event readers. Users
can now specify an event that is called when an info, warning, or error
is emitted by their input reader. This can, e.g., be used to raise
notices in case errors occur when reading an important input stream.
Example:
event error_event(desc: Input::TableDescription, msg: string, level: Reporter::Level)
{
...
}
event bro_init()
{
Input::add_table([$source="a", $error_ev=error_event, ...]);
}
For the moment, this converts all errors in the Asciiformatter into
warnings (to show that they are non-fatal) - the Reader itself also has
to throw an Error to show that a fatal error occurred and processing
will be abort.
It might be nicer to change this and require readers to mark fatal
errors as such when throwing them.
Addresses BIT-1181
Instead of assuming the logger node always has the name "logger", now
broctl will set a boolean which the cluster framework scripts can use
to determine if there is a logger node or not.
Also removed one line from the manager node script, because it has to do
with logging, which the logger.bro script handles.
Great work, and great documentation!
I'm getting one test failure with
scripts.base.frameworks.netcontrol.catch-and-release-cluster Going
ahead and commiting, Jenkins will show the details I assume.
BIT-1584 #merged
* origin/topic/johanna/netcontrol-improvements:
SMTP does not need to pull in the notice framework.
Write NetControl framework documentation.
Use NetControl for ACTION_DROP of notice framework.
NetControl: slightly update catch and release logging
NetControl: fix several small logging issues
NetControl: more catch and release logging and cluster fix
NetControl: rewrite catch and release and small fixes.
NetControl: find_rules_subnet works in cluster mode
NetControl: fix acld whitelist command
NetControl: add rule exists as state besides added and failure.
NetControl: Suppress duplicate "plugin activated" messages.
NetControl: make new broker plugin options accessible
NetControl: add predicates to broker plugin
Changed the cluster framework scripts by adding a new Bro node type
for doing logging (this is intended to reduce the load on the manager).
If a user chooses not to specify a logger node in the cluster
configuration, then the manager will write logs locally as usual.
When inserting, existance of the given subnet is checked using exact
matching instead of longest prefix matching. Before, inserting a subnet
would have updated the subnet item, which is the longest prefix of the
inserted subnet, if present.
In all versions so far, the identifier string that was used for
comparisons might have been different from the identifier string that
was added (when certain notices are used).
This commit rewrites the way that weirds are logged and fixes a number
of issues on the way. Most prominently, flow weirds now actually log
information about the flow that they occur in (before this change, they
only logged the name of the weird, which is only marginally helpful).
Besides restructuring how weird logging works internally, weirds can now
also be generated by calling Weird::weird with the info record directly,
allowing more fine-granular passing of information. This is e.g. used
for DNS weirds, which do not have the connection record available any
more when they are generated (before data like the connection ID was
just not logged in these instances).
Addresses BIT-1578
File Analysis Framework related code has been moved into a separate
script. Using redefinitions of the corresponding records causes the
file-related columns to appear last.
forgotten messages are only logged on the manager (or standalone host)
now. Logs are not written by default anymore when Bro encounters traffic
that should have been blocked.
This commit rewrites catch and release, fixing issues with it and making
it fully cluster capable. A dedicated netcontrol_catch_release.log is
also added.
This is not quite done yet; a few more log messages are missing. There
should hopefully not be many big issues left.
This introduces two new events, NetControl::rule_new and
NetControl::rule_destroyed, which are raised when rules are first added
and then deleted from the internal state tracking.
- When a log record is being "unrolled" (sub-records flattened
out into a single record), it's now possible to choose the
character/string to separate the outer name from the inner
name. This can be used to work around the problems
with ElasticSearch 2.0 not supporting dots "." in field names.
This value can be provided per-filter as well as a global
default value.
- Log fields can be renamed by providing a table per-filter
(or a global default) to rename fields for any log writer.
The name translation is performed after unrolling so the
value in the field name table must match whatever is being
used to separate field names.
For example if the unrolling separator was set to "*":
redef Log::default_unrolling_sep = "*";
The field name map would need to reflect it:
redef Log::default_field_name_map = {
["id*orig_h"] = "src",
["id*orig_p"] = "src_port",
["id*resp_h"] = "dst",
["id*resp_p"] = "dst_port",
};
The extension mechanism is basically the one that Seth introduced with
his intel extensions. The main difference lies in using a hook instead
of an event. An example policy implements whitelisting.
When Bro was compiled with broker disabled, then some Bro scripts
were referencing functions and types that were not defined. Fixed
by adding @ifdefs to several scripts. Removed one @ifdef because
it was causing several unit tests to fail.
Also fixed the @TEST-REQUIRES check in tests that rely on broker so
that such tests are skipped when broker is disabled.
(Cleaned up some code a little bit.)
* origin/topic/seth/stats-improvement:
Fixing tests for stats improvements
Rename the reporting interval variable for stats.
Removing more broken functionality due to changed stats apis.
Removing some references to resource_usage()
Removing Broker stats, it was broken and incomplete.
Fixing default stats collection interval to every 5 minutes.
Add DNS stats to the stats.log
Small stats script tweaks and beginning broker stats.
Continued stats cleanup and extension.
More stats collection extensions.
More stats improvements
Slight change to Mach API for collecting memory usage.
Fixing some small mistakes.
Updating the cmake submodule for the stats updates.
Fix memory usage collection on Mac OS X.
Cleaned up stats collection.
BIT-1581 #merged
BIT-1449 #merged
* origin/topic/dnthayer/ticket1449:
Rename broker BIF wrapper functions in a few more places
Sync the core/leaks/broker/data.bro test with broker/data.bro
Add missing tests for broker data BIFs
Code cleanup for some broker tests
Add script wrapper functions for broker data BIFs
Add script wrapper functions for broker BIFs
Also renamed the "print" function to "send_print" and the "event"
function to "send_event" because Bro shows a syntax error when a
Bro script function is named "event" or "print".
