Refinery - the Honeycomb Sampling Proxy

Release Information

For a detailed list of linked pull requests merged in each release, see CHANGELOG.md. For more readable information about recent changes, please see RELEASE_NOTES.md.

Purpose

Refinery is a tail-based sampling proxy and operates at the level of an entire trace. Refinery examines whole traces and intelligently applies sampling decisions to each trace. These decisions determine whether to keep or drop the trace data in the sampled data forwarded to Honeycomb.

A tail-based sampling model allows you to inspect an entire trace at one time and make a decision to sample based on its contents. For example, your data may have a root span that contains the HTTP status code to serve for a request, and another span that contains information on whether the data was served from a cache. Using Refinery, you can choose to keep only traces that had a 500 status code and were also served from a cache.

Refinery's tail sampling capabilities

Refinery support several kinds of tail sampling:

• Dynamic sampling - This sampling type configures a key based on a trace's set of fields and automatically increases or decreases the sampling rate based on how frequently each unique value of that key occurs. For example, using a key based on http.status_code, you can include in your sampled data:
  • one out of every 1,000 traces for requests that return 2xx
  • one out of every 10 traces for requests that return 4xx
  • every request that returns 5xx
• Rules-based sampling - This sampling type enables you to define sampling rates for well-known conditions. For example, you can keep 100% of traces with an error and then apply dynamic sampling to all other traffic.
• Throughput-based sampling - This sampling type enables you to sample traces based on a fixed upper-bound for the number of spans per second. The sampler will dynamically sample traces with a goal of keeping the throughput below the specified limit.
• Deterministic probability sampling - This sampling type consistently applies sampling decisions without considering the contents of the trace other than its trace ID. For example, you can include 1 out of every 12 traces in the sampled data sent to Honeycomb. This kind of sampling can also be done using head sampling, and if you use both, Refinery takes that into account.

Refinery lets you combine all of the above techniques to achieve your desired sampling behavior.

Setting up Refinery

Refinery is designed to sit within your infrastructure where all traces can reach it. Refinery can run standalone or be deployed in a cluster of two or more Refinery processes accessible via a separate load balancer.

Refinery processes must be able to communicate with each other to concentrate traces on single servers.

Within your application (or other Honeycomb event sources), you would configure the API Host to be http(s)://load-balancer/. Everything else remains the same, such as API key, dataset name, and so on since all that lives with the originating client.

Minimum Configuration

Every Refinery instance should have a minimum of:

• a linux/amd64 or linux/arm64 operating system
• 2GB RAM for each server used
• Access to 2 cores for each server used

In many cases, Refinery only needs one node. If experiencing a large volume of traffic, you may need to scale out to multiple nodes, and likely need a small Redis instance to handle scaling.

We recommend increasing the amount of RAM and the number of cores after your initial set-up. Additional RAM and CPU can be used by increasing configuration values; in particular, CacheCapacity is an important configuration value. Refinery's Stress Relief system provides a good indication of how hard Refinery is working, and when invoked, logs (as reason) the name of the Refinery configuration value that should be increased to reduce stress. Use our scaling and troubleshooting documentation to learn more.

Setting up Refinery in Kubernetes

Refinery is available as a Helm chart in the Honeycomb Helm repository.

You can install Refinery with the following command, which uses the default values file:

helm repo add honeycomb https://honeycombio.github.io/helm-charts
helm install refinery honeycomb/refinery

Alternatively, supply your own custom values file:

helm install refinery honeycomb/refinery --values /path/to/refinery-values.yaml

where /path/to/refinery-values.yaml is the file's path.

Peer Management

When operating in a cluster, Refinery expects to gather all of the spans in a trace onto a single instance so that it can make a trace decision. Since each span arrives independently, each Refinery instance needs to be able to communicate with all of its peers in order to distribute traces to the correct instance.

This communication can be managed in two ways: via an explicit list of peers in the configuration file, or by using self-registration via a shared Redis cache. Installations should generally prefer to use Redis. Even in large installations, the load on the Redis server is quite light, with each instance only making a few requests per minute. A single Redis instance with fractional CPU is usually sufficient.

Configuration

Configuration is controlled by Refinery's two configuration files, which is generally referred to as config.yaml for general configuration and rules.yaml for sampling configuration.

Learn more about config.yaml and all the parameters that control Refinery's operation in our Refinery configuration documentation.

Learn more about rules.yaml and sampler configuration in our Refinery sampling methods documentation.

Running Refinery

Refinery is a typical linux-style command line application, and supports several command line switches.

refinery -h will print an extended help text listing all command line options and supported environment variables.

Environment Variables

Refinery supports the following key environment variables; please see the command line help or the online documentation for the full list. Command line switches take precedence over file configuration, and environment variables take precedence over both.

Environment Variable	Configuration Field
REFINERY_GRPC_LISTEN_ADDRESS	GRPCListenAddr
REFINERY_REDIS_HOST	PeerManagement.RedisHost
REFINERY_REDIS_USERNAME	PeerManagement.RedisUsername
REFINERY_REDIS_PASSWORD	PeerManagement.RedisPassword
REFINERY_HONEYCOMB_API_KEY	HoneycombLogger.LoggerAPIKey
REFINERY_HONEYCOMB_METRICS_API_KEY	LegacyMetrics.APIKey
REFINERY_HONEYCOMB_API_KEY	LegacyMetrics.APIKey
REFINERY_QUERY_AUTH_TOKEN	QueryAuthToken

Note: REFINERY_HONEYCOMB_METRICS_API_KEY takes precedence over REFINERY_HONEYCOMB_API_KEY for the LegacyMetrics.APIKey configuration.

Dry Run Mode

When getting started with Refinery or when updating sampling rules, it may be helpful to verify that the rules are working as expected before you start dropping traffic. To do so, use Dry Run Mode in Refinery.

Enable Dry Run Mode by adding DryRun = true in your configuration file (config.yaml). Then, use Query Builder in the Honeycomb UI to run queries to check your results and verify that the rules are working as intended.

When Dry Run Mode is enabled, the metric trace_send_kept will increment for each trace, and the metric for trace_send_dropped will remain 0, reflecting that we are sending all traces to Honeycomb.

Scaling Up

Refinery uses bounded queues and circular buffers to manage allocating traces, so even under high volume memory use shouldn't expand dramatically. However, given that traces are stored in a circular buffer, when the throughput of traces exceeds the size of the buffer, things will start to go wrong. If you have statistics configured, a counter named collect_cache_buffer_overrun will be incremented each time this happens. The symptoms of this will be that traces will stop getting accumulated together, and instead spans that should be part of the same trace will be treated as two separate traces. All traces will continue to be sent (and sampled), but some sampling decisions will be made on incomplete data. The size of the circular buffer is a configuration option named CacheCapacity. To choose a good value, you should consider the throughput of traces (for example, traces / second started) and multiply that by the maximum duration of a trace (such as 3 seconds), then multiply that by some large buffer (maybe 10x). This estimate will give a good headroom.

Determining the number of machines necessary in the cluster is not an exact science, and is best influenced by watching for buffer overruns. But for a rough heuristic, count on a single machine using about 2GB of memory to handle 5,000 incoming events and tracking 500 sub-second traces per second (for each full trace lasting less than a second and an average size of 10 spans per trace).

Stress Relief

Refinery offers a mechanism called Stress Relief that improves stability under heavy load. The stress_level metric is a synthetic metric on a scale from 0 to 100 that is constructed from several Refinery metrics relating to queue sizes and memory usage. Under normal operation, its value should usually be in the single digits. During bursts of high traffic, the stress levels might creep up and then drop again as the volume drops. As it approaches 100, it is more and more likely that Refinery will start to fail and possibly crash.

Stress Relief is a system that can monitor the stress_level metric and shed load when stress becomes a danger to stability. Once the ActivationLevelis reached, Stress Relief mode will become active. In this state. Refinery will deterministically sample each span based on TraceID without having to store the rest of the trace or evaluate rule conditions. Stress Relief will remain active until stress falls below the DeactivationLevel specified in the config.

The stress relief settings are:

• Mode - Setting to indicate how Stress Relief is used. never indicates that Stress Relief will not activate. monitor means Stress Relief will activate when the ActivationLevel and deactivate when the is reached. always means that Stress Relief mode will continuously be engaged. The always mode is intended for use in emergency situations.
• ActivationLevel - When the stress level rises above this threshold, Refinery will activate Stress Relief.
• DeactivationLevel - When the stress level falls below this threshold, Refinery will deactivate Stress Relief.
• SamplingRate - The rate at which Refinery samples while Stress Relief is active.

The stress_level is currently the best proxy for the overall load on Refinery. Even if Stress Relief is not active, if stress_level is frequently above 50, it is a good indicator that Refinery needs more resources -- more CPUs, more memory, or more nodes. On the other hand, if stress_level never goes into double digits it is likely that Refinery is overprovisioned.

Understanding Regular Operation

Refinery emits a number of metrics to give some indication about the health of the process. These metrics should be sent to Honeycomb, typically with Open Telemetry, and can also be exposed to Prometheus. The interesting ones to watch are:

• Sample rates: how many traces are kept / dropped, and what does the sample rate distribution look like?
• [incoming|peer]_router_*: how many events (no trace info) vs. spans (have trace info) have been accepted, and how many sent on to peers?
• collect_cache_buffer_overrun: this should remain zero; a positive value indicates the need to grow the size of Refinery's circular trace buffer (via configuration CacheCapacity).
• process_uptime_seconds: records the uptime of each process; look for unexpected restarts as a key towards memory constraints.

Troubleshooting

Logging

The default logging level of warn is fairly quiet. The debug level emits too much data to be used in production, but contains excellent information in a pre-production environment,including trace decision information. info is somewhere between. Setting the logging level to debug during initial configuration will help understand what's working and what's not, but when traffic volumes increase it should be set to warn or even error. Logs may be sent to stdout or to Honeycomb.

Configuration Validation

Refinery validates its configuration on startup or when a configuration is reloaded, and it emits diagnostics for any problems. On startup, it will refuse to start; on reload, it will not change the existing configuration.

Configuration Query

Check the loaded configuration by using one of the /query endpoints from the command line on a server that can access a Refinery host.

The /query endpoints are protected and can be enabled by specifying QueryAuthToken in the configuration file or specifying REFINERY_QUERY_AUTH_TOKEN in the environment. All requests to any /query endpoint must include the header X-Honeycomb-Refinery-Query set to the value of the specified token.

For file-based configurations (the only type currently supported), the hash value is identical to the value generated by the md5sum command for the given configuration file.

For all of these commands:

• $REFINERY_HOST should be the URL of your refinery.
• $FORMAT can be one of yaml, toml, or json.
• $DATASET is the name of the dataset you want to check.

To retrieve the entire Rules configuration:

curl --include --get $REFINERY_HOST/query/allrules/$FORMAT --header "x-honeycomb-refinery-query: my-local-token"

To retrieve the rule set that Refinery uses for the specified dataset, which will be returned as a map of the sampler type to its rule set:

curl --include --get $REFINERY_HOST/query/rules/$FORMAT/$DATASET --header "x-honeycomb-refinery-query: my-local-token"

To retrieve information about the configurations currently in use, including the timestamp when the configuration was last loaded:

curl --include --get $REFINERY_HOST/query/configmetadata --header "x-honeycomb-refinery-query: my-local-token"

Sampling

Refinery can send telemetry that includes information that can help debug the sampling decisions that are made. To enable, in the configuration file, set AddRuleReasonToTrace to true. This will cause traces that are sent to Honeycomb to include a field meta.refinery.reason, which will contain text indicating which rule was evaluated that caused the trace to be included.

Restarts

Refinery does not yet buffer traces or sampling decisions to disk. When you restart the process all in-flight traces will be flushed (sent upstream to Honeycomb), but you will lose the record of past trace decisions. When started back up, it will start with a clean slate.

Architecture of Refinery itself (for contributors)

Within each directory, the interface the dependency exports is in the file with the same name as the directory and then (for the most part) each of the other files are alternative implementations of that interface. For example, in logger, /logger/logger.go contains the interface definition and logger/honeycomb.go contains the implementation of the logger interface that will send logs to Honeycomb.

main.go sets up the app and makes choices about which versions of dependency implementations to use (eg which logger, which sampler, etc.) It starts up everything and then launches App.

app/app.go is the main control point. When its Start function ends, the program shuts down. It launches two Routers which listen for incoming events.

route/route.go listens on the network for incoming traffic. There are two routers running and they handle different types of incoming traffic: events coming from the outside world (the incoming router) and events coming from another member of the Refinery cluster (peer traffic). Once it gets an event, it decides where it should go next: is this incoming request an event (or batch of events), and if so, does it have a trace ID? Everything that is not an event or an event that does not have a trace ID is immediately handed to transmission to be forwarded on to Honeycomb. If it is an event with a trace ID, the router extracts the trace ID and then uses the sharder to decide which member of the Refinery cluster should handle this trace. If it's a peer, the event will be forwarded to that peer. If it's us, the event will be transformed into an internal representation and handed to the collector to bundle spans into traces.

collect/collect.go the Collector is responsible for bundling spans together into traces and deciding when to send them to Honeycomb or if they should be dropped. The first time a trace ID is seen, the Collector starts a timer. If the root span, which is a span with a trace ID and no parent ID, arrives before the timer expires, then the trace is considered complete. The trace is sent and the timer is canceled. If the timer expires before the root span arrives, the trace will be sent whether or not it is complete. Just before sending, the Collector asks the sampler for a sample rate and whether or not to keep the trace. The Collector obeys this sampling decision and records it (the record is applied to any spans that may come in as part of the trace after the decision has been made). After making the sampling decision, if the trace is to be kept, it is passed along to the transmission for actual sending.

transmit/transmit.go is a wrapper around the HTTP interactions with the Honeycomb API. It handles batching events together and sending them upstream.

logger and metrics are for managing the logs and metrics that Refinery itself produces.

sampler contains algorithms to compute sample rates based on the traces provided.

sharder determines which peer in a clustered Refinery configuration is supposed to handle an individual trace.

types contains a few type definitions that are used to hand data in between packages.