This adds a Broker-specific script to the cluster framework, loaded only when
Zeek is running in cluster mode. It adds logging in cluster.log as well as
telemetry via a metrics counter for Broker-observed backpressure disconnects.
The new zeek_broker_backpressure_disconnects counter, labeled by the neighboring
peer that the reporting node has determined to be unresponsive, counts the
number of unpeerings for this reason.
Here the node "worker" has observed node "proxy" falling behind once:
# HELP zeek_broker_backpressure_disconnects_total Number of Broker peering drops due to a neighbor falling too far behind in message I/O
# TYPE zeek_broker_backpressure_disconnects_total counter
zeek_broker_backpressure_disconnects_total{endpoint="worker",peer="proxy"} 1
Includes small btest baseline update to reflect @load of a new script.
(cherry picked from commit ead6134501)
This module is loaded by the telemetry framework, which we're now loading via
the cluster framework, i.e. also in bare mode. The resulting additional
thread (for creating reporter.log) trips up a number of btest baselines.
version.zeek doesn't use any of the string helper functions.
(cherry picked from commit d260a5b7a9)
This translates backend-specific node identifiers (like Broker IDs) to
cluster nodes and their names, if available.
(cherry picked from commit 46a11ec37d)
This adds re-peering at the Broker level for peers that Broker decided to
unpeer. We keep this at the Broker level since this behavior is specific to
it (as opposed to other cluster backends).
Includes baseline updates for btests that pick up on the new script's @load.
(cherry picked from commit 0010e65f6d)
This moves the Telemetry framework's BIF-defined functionalit from the
secondary-BIFs stage to the primary one. That is, this functionality is now
available from the end of init-bare.zeek, not only after the end of
init-frameworks-and-bifs.zeek.
This allows us to use script-layer telemetry in our Zeek's own code that get
pulled in during init-frameworks-and-bifs.
This change splits up the BIF features into functions, constants, and types,
because that's the granularity most workable in Func.cc and NetVar. It also now
defines the Telemetry::MetricsType enum once, not redundantly in BIFs and script
layer.
Due to subtle load ordering issues between the telemetry and cluster frameworks
this pushes the redef stage of Telemetry::metrics_port and address into
base/frameworks/telemetry/options.zeek, which is loaded sufficiently late in
init-frameworks-and-bifs.zeek to sidestep those issues. (When not doing this,
the effect is that the redef in telemetry/main.zeek doesn't yet find the
cluster-provided values, and Zeek does not end up listening on these ports.)
The need to add basic Zeek headers in script_opt/ZAM/ZBody.cc as a side-effect
of this is curious, but looks harmless.
Also includes baseline updates for the usual btests and adds a few doc strings.
(cherry picked from commit 71f7e89974)
With Cluster::Node$metrics_port being optional, there's not really
a need for the extra script. New rule, if a metrics_port is set, the
node will attempt to listen on it.
Users can still redef Telemetry::metrics_port *after*
base/frameworks/telemetry was loaded to change the port defined
in cluster-layout.zeek.
(cherry picked from commit bf9704f339)
The controller learns IP addresses from agents that peer with it, but that
information has so far gotten lost when resulting configs get pushed out to the
agents. This makes these updates include that information.
This is quite redundant with the enumeration for Broker ports,
unfortunately. But the logic is subtly different: all nodes obtain a telemetry
port, while not all nodes require a Broker port, for example, and in the metrics
port assignment we also cross-check selected Broker ports. I found more unified
code actually harder to read in the end.
The logic for the two sets remains the same: from a start point, ports get
enumerated sequentially that aren't otherwise taken. These ports are assumed
available; there's nothing that checks their availability -- for now.
The default start port is 9000. I considered 9090, to align with the Prometheus
default, but counting upward from there is likely to hit trouble with the Broker
default ports (9999/9997), used by the Supervisor. Counting downward is a bit
unnatural, and shifting the Broker default ports brings subtle ordering issues.
This also changes the node ordering logic slightly since it seems more intuitive
to keep sequential ports on a given instance, instead of striping across them.
This eliminates one place in which we currently need to mirror changes to the
script-land Cluster::Node record. Instead of keeping an exact in-core equivalent, the
Supervisor now treats the data structure as opaque, and stores the whole cluster
table as a JSON string.
We may replace the script-layer Supervisor::ClusterEndpoint in the future, using
Cluster::Node directly. But that's a more invasive change that will affect how
people invoke Supervisor::create() and similars.
Relying on JSON for serialization has the side-effect of removing the
Supervisor's earlier quirk of using 0/tcp, not 0/unknown, to indicate unused
ports in the Supervisor::ClusterEndpoint record.
If the script layer is able to access the current node's config via
Supervisor::node(), it can handle populating Cluster::nodes. That code
is much more straightforward than an equivalent in-core implementation
(especially with the upcoming change to the cluster table's implementation).
This introduces base/frameworks/cluster/supervisor.zeek and
Cluster::Supervisor::__init_cluster_nodes() for that purpose.
The @load of the Supervisor API in cluster/main.zeek isn't technically
necessary since we already load it explicitly even in init-bare.zeek,
but being explicit seems better.
This allows to read Zeek global variables from inside Spicy code. The
main challenge here is supporting all of Zeek's data type in a
type-safe manner.
The most straight-forward API is a set of functions
`get_<type>(<id>)`, where `<type>` is the Zeek-side type
name (e.g., `count`, `string`, `bool`) and `<id>` is the fully scoped
name of the Zeek-side global (e.g., `MyModule::Boolean`). These
functions then return the corresponding Zeek value, converted in an
appropriate Spicy type. Example:
Zeek:
module Foo;
const x: count = 42;
const y: string = "xxx";
Spicy:
import zeek;
assert zeek::get_count("Foo::x") == 42;
assert zeek::get_string("Foo::y") == b"xxx"; # returns bytes(!)
For container types, the `get_*` function returns an opaque types that
can be used to access the containers' values. An additional set of
functions `as_<type>` allows converting opaque values of atomic
types to Spicy equivalents. Example:
Zeek:
module Foo;
const s: set[count] = { 1, 2 };
const t: table[count] of string = { [1] = "One", [2] = "Two" }
Spicy:
# Check set membership.
local set_ = zeek::get_set("Foo::s");
assert zeek::set_contains(set_, 1) == True
# Look up table element.
local table_ = zeek::get_table("Foo::t");
local value = zeek::table_lookup(t, 1);
assert zeek::as_string(value) == b"One"
There are also functions for accessing elements of Zeek-side vectors
and records.
If any of these `zeek::*` conversion functions fails (e.g., due to a
global of that name not existing), it will throw an exception.
Design considerations:
- We support only reading Zeek variables, not writing. This is
both to simplify the API, and also conceptually to avoid
offering backdoors into Zeek state that could end up with a very
tight coupling of Spicy and Zeek code.
- We accept that a single access might be relatively slow due to
name lookup and data conversion. This is primarily meant for
configuration-style data, not for transferring lots of dynamic
state over.
- In that spirit, we don't support deep-copying complex data types
from Zeek over to Spicy. This is (1) to avoid performance
problems when accidentally copying large containers over,
potentially even at every access; and (2) to avoid the two sides
getting out of sync if one ends up modifying a container without
the other being able to see it.
