Skip to content

IndexDefect in checkStore when store peers run out kills the module - transient network loss becomes permanent death #87

Description

@johnniedom

Summary

When the node runs out of store peers (for example when the network degrades mid-session), checkStore raises an IndexDefect and the module dies with it. A transient network problem becomes permanent loss of networking for a long-running agent.

Observed

2026-07-30, long-running agent session on the logos.dev preset. Connectivity degraded, the store peer list emptied out, checkStore threw IndexDefect, and delivery stayed dead for the remainder of the process lifetime. In our case this happened partway through a scored test run and quietly converted a healthy score into a meaningless one.

Suggested

Bounds-check the store peer selection and treat an empty store peer list as a retry/backoff state rather than an exception. A node with zero store peers is degraded, not broken.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    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