Skip to content

SIGSEGV (exit 139) under sustained MCP write load #431

Description

@jrosskopf

During the #635 demo seeding (2026-08-30), dz-escurel on lab (7d uptime, image 0.0.0-dev) segfaulted — container exit 139 — while receiving paced update_page writes over MCP (47 markdown pages, ~1/1.2s with occasional bursts). Not a poison page: the page that coincided with the crash writes fine in isolation; the failure correlates with sustained write pressure. Recovered on restart; the seeding completed with retries + pacing. Worth a look at the CRDT/index write path under load — a segfault in safe-Rust territory usually means an unsafe block or a C dependency (DuckDB?). Evidence: pod dz-escurel-7579d84844-v62nf, lastState terminated 2026-08-30T08:49:01Z, exitCode 139.

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