BloopAI/vibe-kanban · error · ContainerError

Container ref not found

Error message

Container ref not found

What it means

start_execution needs the workspace's container_ref (its on-disk directory path) to locate each repo and capture HEAD info. This error is thrown when the workspace's container_ref column is NULL, meaning the workspace has never been materialized on disk (container not yet created or creation failed).

Source

Thrown at crates/services/src/services/container.rs:1154

        session: &Session,
        executor_action: &ExecutorAction,
        run_reason: &ExecutionProcessRunReason,
    ) -> Result<ExecutionProcess, ContainerError> {
        // Create new execution process record
        // Capture current HEAD per repository as the "before" commit for this execution
        let repositories =
            WorkspaceRepo::find_repos_for_workspace(&self.db().pool, workspace.id).await?;
        if repositories.is_empty() {
            return Err(ContainerError::Other(anyhow!(
                "Workspace has no repositories configured"
            )));
        }

        let workspace_root = workspace
            .container_ref
            .as_ref()
            .map(std::path::PathBuf::from)
            .ok_or_else(|| ContainerError::Other(anyhow!("Container ref not found")))?;

        let mut repo_states = Vec::with_capacity(repositories.len());
        for repo in &repositories {
            let repo_path = workspace_root.join(&repo.name);
            let before_head_commit = self.git().get_head_info(&repo_path).ok().map(|h| h.oid);
            repo_states.push(CreateExecutionProcessRepoState {
                repo_id: repo.id,
                before_head_commit,
                after_head_commit: None,
                merge_commit: None,
            });
        }
        let create_execution_process = CreateExecutionProcess {
            session_id: session.id,
            executor_action: executor_action.clone(),
            run_reason: run_reason.clone(),
        };

View on GitHub (pinned to 4deb7eca8f)

Solutions

  1. Wait for workspace initialization to complete and confirm container_ref is set before starting executions.
  2. Inspect the workspaces row (SELECT container_ref FROM workspaces WHERE id = ...) to verify the path is populated.
  3. Retry workspace creation / setup if the container failed to materialize on disk.
  4. Check that the container creation step succeeded (logs for Docker/FS errors) before re-triggering execution.

Example fix

// before
let process = container.start_execution(&workspace, cmd, ...).await?;
// after
if workspace.container_ref.is_none() {
    anyhow::bail!("workspace {} not initialized yet", workspace.id);
}
let process = container.start_execution(&workspace, cmd, ...).await?;
Defensive patterns

Strategy: validation

Validate before calling

if workspace.container_ref.as_deref().map_or(true, |p| !Path::new(p).is_dir()) {
    anyhow::bail!("workspace {} container not ready on disk", workspace.id);
}

Try / catch

if let Err(e) = start_execution(&workspace, ...).await {
    if e.to_string().contains("Container ref not found") {
        reinitialize_workspace(&workspace).await?;
    } else { return Err(e); }
}

Prevention

When it happens

Trigger: Calling start_execution (directly or via follow_up/start_review/run_setup_script/run_codex_setup/run_cursor_setup) on a workspace whose container_ref is NULL — e.g. before workspace initialization finished, or after it failed.

Common situations: Submitting a follow-up while workspace creation is still in progress; container creation failed earlier (disk, Docker, path error) leaving a NULL ref; DB rows created but setup aborted; race between workspace creation UI and immediate execution.

Related errors

AI-assisted analysis of BloopAI/vibe-kanban@4deb7eca8f (2026-08-29). Data as JSON: /api/errors/8b76ff3bbebddb86. Report an issue: GitHub.