gastownhall/beads · error

search %s: rows: %w

Error message

search %s: rows: %w

What it means

Raised when rows.Err() returns non-nil after fully iterating the search result set. This is the idiomatic database/sql check for errors that occurred asynchronously during streaming — typically the connection dropping or the server aborting mid-iteration, not an error at query start.

Source

Thrown at internal/storage/domain/db/issue_search.go:279

	}

	var issues []*types.Issue
	seen := make(map[string]bool)
	for rows.Next() {
		issue, scanErr := scanIssue(rows)
		if scanErr != nil {
			_ = rows.Close()
			return domain.SearchPage{}, fmt.Errorf("search %s: scan: %w", tables.Main, scanErr)
		}
		if seen[issue.ID] {
			continue
		}
		seen[issue.ID] = true
		issues = append(issues, issue)
	}
	_ = rows.Close()
	if err := rows.Err(); err != nil {
		return domain.SearchPage{}, fmt.Errorf("search %s: rows: %w", tables.Main, err)
	}

	sortRowsGoSide(issues, func(i *types.Issue) string { return i.ID }, filter.SortBy, filter.SortDesc)
	items, hasMore, err := finishWindow(issues, window)
	if err != nil {
		return domain.SearchPage{}, err
	}

	if err := r.hydrateIssues(ctx, items, tables, filter.IncludeDependencies, filter.SkipLabels); err != nil {
		return domain.SearchPage{}, fmt.Errorf("search %s: hydrate: %w", tables.Main, err)
	}

	return domain.SearchPage{Items: items, HasMore: hasMore}, nil
}

func (r *issueSQLRepositoryImpl) scanFilterIDs(ctx context.Context, selectKw, fromSQL, whereSQL string, args []any, filter types.IssueFilter, tables filterTables) ([]string, bool, error) {
	orderBy := orderBySQL(filter.SortBy, filter.SortDesc, tables.Main)
	window := searchWindowForFilter(filter)

View on GitHub (pinned to 71377f2769)

Solutions

1. Inspect the wrapped error to distinguish connection loss from driver I/O failure.
2. Retry the search; transient connection drops usually succeed on a second attempt.
3. Reduce result-set size with a tighter IssueFilter (limit/status filters) so iteration completes before timeout.
4. For remote Dolt, increase connection idle/read timeouts in the driver configuration.

Example fix

// before
page, err := store.Search(ctx, q, types.IssueFilter{}) // huge result set, drops mid-stream
// after
page, err := store.Search(ctx, q, types.IssueFilter{Status: "open", Limit: 100})

if err := ctx.Err(); err != nil { return err } // ensure context is still live before searching
// prefer filtered queries over unbounded scans

func isRowsIterationError(err error) bool {
	return err != nil && strings.Contains(err.Error(), ": rows: ")
}

var page domain.SearchPage
err := retry.Do(3, backoff, func() error {
	var e error
	page, e = store.Search(ctx, q, filter)
	if e != nil && isRowsIterationError(e) {
		return e // transient stream failure; retry
	}
	return retry.Stop(e)
})

When it happens

Trigger: Iterating a large search result when the underlying connection is lost, the driver hits I/O failure reading the next batch, or the database process is killed while rows are still open.

Common situations: Searching very large issue sets over a remote Dolt server whose connection times out mid-stream; SQLite file unmounted or network filesystem dropped during iteration; OOM-killer terminating the DB process during a big search.

Related errors

• search %s (id scan): rows: %w
• get next child ID: iterate children: %w
• iterate dependency sources: %w
• iterate dependency targets: %w
• count edges in %s: rows: %w

AI-assisted analysis of gastownhall/beads@71377f2769 (2026-08-30). Data as JSON: /api/errors/3c9ca1f2a6e70f78. Report an issue: GitHub.