Skip to content

Periodical dd-task-scheduler errors - Unable to create StatsD client - localhost:8125 #78

@drobakowski

Description

@drobakowski

Hi there,

I'm not sure if the following errors origins from this library or java-dogstatsd-client, but we're seeing a lot of dd-task-scheduler errors from Java lambdas like this:

[dd.trace 2022-04-18 22:16:49:700 +0000] [dd-task-scheduler] ERROR datadog.communication.monitor.DDAgentStatsDConnection - Unable to create StatsD client - localhost:8125

We are using version 1.4.4 of this library in combination with the latest serverless-datadog-plugin - v5.0.0 which uses the lambda layer v21 of datadog-lambda-extension and v4 of the dd-trace-java layer and java11. For instrumenting things we've followed the official DataDog documentation. The errors doesn't seem to be specific to the Serverless plugin, as we also get this errors when we integrate the layers and configs directly without the plugin. Also the log level doesn't seem to take effect, as we're seeing the logs unaggregated on INFO level, despite the fact that we have configured logLevel: warn or DD_LOG_LEVEL: WARN.

Our serverless.yml config looks like this:

...
provider:
  environment:
    DD_ENV: ${self:provider.stage}
    DD_SERVICE: ${self:service}
    DD_TRACE_SAMPLE_RATE: ${self:custom.datadogTracingSampleRate.${opt:stage, self:provider.stage}}
    DD_VERSION: ${git:sha1}
...
custom:
  datadogTracingSampleRate:
    dev: 1.0
    int: 0.8
    prod: 1.0
  ...
  datadog:
    addExtension: true
    apiKey: ${ssm:.../apiKey}
    site: datadoghq.eu

    flushMetricsToLogs: true
    addLayers: true
    logLevel: warn
    enableTags: true
    injectLogContext: true
    integrationTesting: false

Any ideas what we could do to fix the error?

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions