Skip to main content
Version: Next

Release notes v0.1.7

Here are some highlights of this release. For a full list of updates available for Release v0.1.7, check out here.

YAML configuration supports multiple values

In previous releases, the Pulsar Function or connector configurations were declared as map[string]string, and nested or complex configurations shown below were not supported.

sinkConfig:
topics:
a: xxx
b: xxxx
settings:
x: 123
y: foo

In this release, multiple values for related parameters are supported. You can define untyped Pulsar Function or connector configurations in the YAML format, as shown below.

sinkConfig:
topics:
players:
operation:
type: write
total-timeout: 0
mapping:
bins:
type: multi-bins
map:
name:
source: value-field
field-name: name
key-field:
source: key
namespace:
mode: dynamic
source: value-field
field-name: namespace
ttl:
mode: dynamic
source: value-field
field-name: ttl
retired-players:
operation:
type: delete
mapping:
key-field:
source: key
namespace:
mode: static
value: test
services:
use-services-alternate: true

Pulsar source connectors can pass message properties to a target topic

In previous releases, Pulsar source connectors could not pass message properties to a target topic because the forwardSourceMessageProperty was not applied to the source connector.

In this release, the forwardSourceMessageProperty parameter is added to the source connectors. You can use Function Mesh to pass message properties to the target topic.

Function Mesh provides multiple options for auto-scaling the number of Pods

In previous releases, Function Mesh supported scaling Pods (Function instances or Connector instances) based on the CPU utilization automatically.

In this release, Function Mesh auto-scales the number of Pods based on the CPU usage, memory usage, and metrics.

  • CPU usage: auto-scale the number of Pods based on 80%, 50% or 20% CPU utilization.
  • Memory usage: auto-scale the number of Pods based on 80%, 50% or 20% memory utilization.
  • Metrics: auto-scale the number of Pods based on a single metric.

For details, see scaling.