Class: Datadog::Tracing::Sampling::PrioritySampler
- Inherits:
-
Object
- Object
- Datadog::Tracing::Sampling::PrioritySampler
- Defined in:
- lib/datadog/tracing/sampling/priority_sampler.rb
Overview
Instance Attribute Summary collapse
-
#pre_sampler ⇒ Object
readonly
NOTE: We do not advise using a pre-sampler.
-
#priority_sampler ⇒ Object
readonly
NOTE: We do not advise using a pre-sampler.
Class Method Summary collapse
-
.sampled?(priority_sampling) ⇒ Boolean
Check if the Priority Sampling decision is to keep or drop the trace.
Instance Method Summary collapse
-
#initialize(opts = {}) ⇒ PrioritySampler
constructor
A new instance of PrioritySampler.
-
#sample!(trace) ⇒ Object
DEV-2.0:We should get rid of this complicated interaction between @pre_sampler and @priority_sampler.
- #update(rate_by_service, decision: nil) ⇒ Object
Constructor Details
#initialize(opts = {}) ⇒ PrioritySampler
Returns a new instance of PrioritySampler.
21 22 23 24 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 21 def initialize(opts = {}) @pre_sampler = opts[:base_sampler] || AllSampler.new @priority_sampler = opts[:post_sampler] || RateByServiceSampler.new(decision: Sampling::Ext::Decision::AGENT_RATE) end |
Instance Attribute Details
#pre_sampler ⇒ Object (readonly)
NOTE: We do not advise using a pre-sampler. It can save resources, but pre-sampling at rates < 100% may result in partial traces, unless the pre-sampler knows exactly how to drop a span without dropping its ancestors.
Additionally, as service metrics are calculated in the Datadog Agent, the service’s throughput will be underestimated.
19 20 21 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 19 def pre_sampler @pre_sampler end |
#priority_sampler ⇒ Object (readonly)
NOTE: We do not advise using a pre-sampler. It can save resources, but pre-sampling at rates < 100% may result in partial traces, unless the pre-sampler knows exactly how to drop a span without dropping its ancestors.
Additionally, as service metrics are calculated in the Datadog Agent, the service’s throughput will be underestimated.
19 20 21 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 19 def priority_sampler @priority_sampler end |
Class Method Details
.sampled?(priority_sampling) ⇒ Boolean
Check if the Priority Sampling decision is to keep or drop the trace. Other factors can influence the sampling decision; this method is only responsible for interpreting the Sampling Priority decision.
90 91 92 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 90 def self.sampled?(priority_sampling) priority_sampling >= Ext::Priority::AUTO_KEEP end |
Instance Method Details
#sample!(trace) ⇒ Object
DEV-2.0:We should get rid of this complicated interaction between @pre_sampler and @priority_sampler. DEV-2.0:If the user wants to configure a custom sampler, we should only allow them to provide a complete DEV-2.0:sampling suite, not having this convoluted support for mixing arbitrary provided samplers in DEV-2.0:the PrioritySampler. Ideally, the PrioritySampler is only used by Datadog. DEV-2.0:There are too many edge cases and combinations to work around currently in this class.
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 31 def sample!(trace) # The priority that was set before the sampler ran. # This comes from distributed tracing priority propagation. distributed_sampling_priority = priority_assigned?(trace) # If pre-sampling is configured, do it first. (By default, this will sample at 100%.) # NOTE: Pre-sampling at rates < 100% may result in partial traces; not recommended. trace.sampled = pre_sample?(trace) ? preserving_priority_sampling(trace) { @pre_sampler.sample!(trace) } : true if trace.sampled? # If priority sampling has already been applied upstream, use that value. return true if priority_assigned?(trace) # Check with post sampler how we set the priority. sample = priority_sample!(trace) # Check if post sampler has already assigned a priority. return true if priority_assigned?(trace) # If not, use agent priority values. priority = sample ? Sampling::Ext::Priority::AUTO_KEEP : Sampling::Ext::Priority::AUTO_REJECT assign_priority!(trace, priority) else # If discarded by pre-sampling, set "reject" priority, so other # services for the same trace don't sample needlessly. assign_priority!(trace, Sampling::Ext::Priority::AUTO_REJECT) end trace.sampled? ensure if trace.sampling_priority && trace.sampling_priority > 0 # Don't modify decision if priority was set upstream. if !distributed_sampling_priority && !trace.has_tag?(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER) # If no sampling priority being assigned at this point, a custom # sampler implementation is configured: this means the user has # full control over the sampling decision. trace.set_tag( Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER, Sampling::Ext::Decision::MANUAL ) end else # The sampler decided to not keep this span, removing sampling decision. trace.clear_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER) end end |
#update(rate_by_service, decision: nil) ⇒ Object
79 80 81 |
# File 'lib/datadog/tracing/sampling/priority_sampler.rb', line 79 def update(rate_by_service, decision: nil) @priority_sampler.update(rate_by_service, decision: decision) end |