LoginSign Up
← Back to all sparks
P

Pinpoint

ANALYTICS
Velocity0.0

Open-source APM tool for monitoring large-scale distributed systems

pinpoint-apm.gitbook.io ↗

ServerMap rebuilt and application names finally long enough to describe a service.

apmdistributed-tracingservermapjvm-agentsschema-migration
◆Current state
Pinpoint ships a minor roughly once a year with patch releases in between. The 3.1.0 release rebuilt ServerMap as V3 with a redesigned storage layout, a new query path and a new set of map tables, and raised the applicationName ceiling from 24 to 254 characters — gated behind an agent property that requires collector 3.1.0 or higher. The patch line before it is mostly backports and plugin compatibility: Java 26, Kafka Streams, Kafka 4.x, S3, nested Spring Boot JARs.
◆Where it's heading
The plugin surface expands continuously — each release absorbs another client library or runtime version — while the platform work arrives in rare, larger jumps that touch storage schema and require coordinated agent and collector upgrades. The 3.1.0 changes suggest the constraints being addressed are those of large deployments: structured naming schemes that no longer fit, and a topology view whose query path needed redesigning rather than tuning.
◆Prediction
Expect the 3.1 line to spend its patches stabilising the ServerMap V3 storage path and backporting plugin updates, with the next set of runtime and client integrations arriving the same way they always have.

◆Recent moves

  1. 2mo ago

    3.1.0 rebuilds ServerMap and lifts application names to 254 characters

    ⚡ SPARK

    The first release in this window to change the platform rather than its plugin coverage: ServerMap becomes V3 with a redesigned storage layout, new query path and new map tables, and applicationName goes from 24 to 254 characters. Both require coordinated agent and collector upgrades, which is why the notes lead with a property you must add at agent start.

    View source ↗
  2. 4mo ago

    3.0.5: backports and a move to Central Portal publishing

    Entirely backports — agent interceptable targets, an S3 auth header, a gRPC plugin stream getter, span sender state init — plus a migration to Central Portal publishing. Maintenance for users staying on the 3.0 line.

    View source ↗
  3. 8mo ago

    3.0.4 adds Java 26, Kafka Streams and nested Spring Boot JAR support

    Runtime coverage widens where it matters for modern Java deployments: Java 26, Kafka Streams tracing behind a flag, and nested JAR filesystems as used by Spring Boot. Reactor tracing improves and collector CPU usage drops.

    View source ↗
  4. 1y ago

    3.0.3 adds an S3 client plugin and Kafka 4.x compatibility

    New plugin coverage for S3 clients and Kafka 4.x, plus Kafka entry-point and gRPC forwarding fixes. From 3.x, executables are published to Maven Central rather than distributed only as archives.

    View source ↗
  5. 1y ago

    3.0.2 stops collecting SQL comments by default

    SQL comments are no longer captured with queries unless you set profiler.jdbc.removecomments to false — a default change that shrinks what a tracing agent quietly stores about your queries. R2DBC and Kafka 3.x plugin updates accompany it.

    View source ↗
  6. 1y ago

    3.0.1 adds SQL cache bypassing and Ktor support

    Queries over a configurable length skip the SQL cache entirely, cutting memory use and improving hit rates for the queries that do get cached, with long queries also trimmed. Arcus, MariaDB 3.x and Ktor plugins land alongside a DirectByteBuffer leak fix.

    View source ↗