Skip to content

observability: increase Go runtime scheduling latency histogram max #121729

Open
@jbowens

Description

@jbowens

The Go runtime scheduling latency histogram has a max bucket threshold of 100ms. We frequently see histogram values in the max bucket, which suggests there are Go scheduling latency events >100ms. These high latency events can be disruptive depending on the goroutine that goes unscheduled. We should consider lifting the max bucket threshold to capture more data about the tail of scheduling latencies.

Relevant internal slack thread: https://cockroachlabs.slack.com/archives/C063CP41TG9/p1707946050906159

Jira issue: CRDB-37471

Metadata

Metadata

Assignees

No one assigned

    Labels

    A-cluster-observabilityRelated to cluster observabilityC-enhancementSolution expected to add code/behavior + preserve backward-compat (pg compat issues are exception)T-observability

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions