Constructor and Description |
---|
InMemorySpanStore() |
Modifier and Type | Method and Description |
---|---|
List<DependencyLink> |
getDependencies(long endTs,
Long lookback)
Returns dependency links derived from spans.
|
List<Span> |
getRawTrace(long traceId)
Retrieves spans that share a trace id, as returned from backend data store queries, with no
ordering expectation.
|
List<String> |
getServiceNames()
Get all the
service names . |
List<String> |
getSpanNames(String service)
Get all the span names for a particular
Endpoint.serviceName . |
List<Span> |
getTrace(long traceId)
Get the available trace information from the storage system.
|
List<List<Span>> |
getTraces(QueryRequest request)
Get the available trace information from the storage system.
|
List<Long> |
traceIds() |
public List<List<Span>> getTraces(QueryRequest request)
SpanStore
Results are sorted in order of the first span's timestamp, and contain up to QueryRequest.limit
elements.
public List<Span> getTrace(long traceId)
SpanStore
getTrace
in interface SpanStore
Span.traceId
, or null if not present.public List<Span> getRawTrace(long traceId)
SpanStore
This is different, but related to SpanStore.getTrace(long)
. SpanStore.getTrace(long)
cleans data by
merging spans, adding timestamps and performing clock skew adjustment. This feature is for
debugging zipkin logic or zipkin instrumentation.
getRawTrace
in interface SpanStore
Span.traceId
, or null if not present.public List<String> getServiceNames()
SpanStore
service names
.
Results are sorted lexicographically
getServiceNames
in interface SpanStore
public List<String> getSpanNames(String service)
SpanStore
Endpoint.serviceName
.
Results are sorted lexicographically
getSpanNames
in interface SpanStore
public List<DependencyLink> getDependencies(long endTs, @Nullable Long lookback)
SpanStore
Implementations may bucket aggregated data, for example daily. When this is the case, endTs may be floored to align with that bucket, for example midnight if daily. lookback applies to the original endTs, even when bucketed. Using the daily example, if endTs was 11pm and lookback was 25 hours, the implementation would query against 2 buckets.
Some implementations parse DependencyLinkSpan
from storage and call
DependencyLinker
to aggregate links. The reason is certain graph logic,
such as skipping up the tree is difficult to implement as a storage query.
getDependencies
in interface SpanStore
endTs
- only return links from spans where Span.timestamp
are at or before this
time in epoch milliseconds.lookback
- only return links from spans where Span.timestamp
are at or after
(endTs - lookback) in milliseconds. Defaults to endTs.Copyright © 2015–2016 OpenZipkin. All rights reserved.