siyuan-note/siyuan · warning · ErrIndexing

indexing

Error message

indexing

What it means

Sentinel error ErrIndexing declared at tree.go:206 ('indexing'). Returned when a block/tree lookup fails because the kernel is actively indexing (task.ContainIndexTask() is true) OR the searchTreeLimiter rate limiter prevents a filesystem fallback reindex (indexTreeInFilesystem at tree.go:332). It signals a TRANSIENT state: the data may become available once indexing completes.

Source

Thrown at kernel/model/tree.go:206

			logging.LogErrorf("decrypt tree [path=%s] failed: %s", localPath, err)
			return
		}
	}

	ret, err = dataparser.ParseJSONWithoutFix(data, luteEngine.ParseOptions)
	if err != nil {
		logging.LogErrorf("parse json to tree [%s] failed: %s", localPath, err)
		return
	}
	return
}

var (
	ErrBoxNotFound   = errors.New("notebook not found")
	ErrBoxClosed     = errors.New("notebook closed")
	ErrBlockNotFound = errors.New("block not found")
	ErrTreeNotFound  = errors.New("tree not found")
	ErrIndexing      = errors.New("indexing")
	ErrBoxUnindexed  = errors.New("notebook unindexed")
	ErrInvalidID     = errors.New("invalid id")
)

func LoadTreeByBlockIDWithReindex(id string) (ret *parse.Tree, err error) {
	return LoadTreeByBlockIDWithReindexInBox(id, "")
}

// LoadTreeByBlockIDWithReindexInBox 与 LoadTreeByBlockIDWithReindex 一致，但按 boxID 路由 blocktree 查询。
func LoadTreeByBlockIDWithReindexInBox(id, boxID string) (ret *parse.Tree, err error) {
	if "" == id {
		logging.LogWarnf("block id is empty")
		return nil, ErrTreeNotFound
	}

	bt := treenode.GetBlockTreeInBox(id, boxID)
	if nil == bt && "" == boxID {
		// boxID 未知时（如通用打开入口），遍历所有已打开的加密笔记本查找

View on GitHub (pinned to 251596fc0d)

Solutions

1. Treat ErrIndexing as transient: retry the operation after indexing completes (poll boot progress or use the indexing-complete WebSocket event).
2. Use errors.Is(err, model.ErrIndexing) to return a 503/retry-after style response to API clients.
3. Avoid flooding lookups during reindex; the rate limiter (3s) intentionally throttles filesystem reindex attempts.
4. If stuck, check task.ContainIndexTask() status and kernel boot progress (util.IsBooted()).

Example fix

// before — treat all errors as fatal
tree, err := LoadTreeByBlockID(id)
if err != nil { return fatal(err) }

// after — retry on transient indexing error
tree, err := LoadTreeByBlockID(id)
if errors.Is(err, model.ErrIndexing) {
    return retryLater()
}

func isStillIndexing() bool {
    return task.ContainIndexTask() || !util.IsBooted()
}

for attempt := 0; attempt < maxAttempts; attempt++ {
    tree, err := model.LoadTreeByBlockID(id)
    if err == nil { return tree, nil }
    if !errors.Is(err, model.ErrIndexing) { return nil, err }
    // wait for indexing progress via WS event, then retry
    waitForIndexProgress()
}

When it happens

Trigger: During boot/notebook reindex, task.ContainIndexTask() returns true so LoadTreeByBlockIDWithReindexInBox/loadTreeByBlockIDInBox0 return ErrIndexing instead of attempting filesystem reindex; alternatively, the rate limiter (1 per 3s) blocks a fallback reindex attempt. API handlers in block.go check errors.Is(err, model.ErrIndexing) to return a 'try again later' response.

Common situations: Opening/operating on a doc immediately after kernel boot while notebooks are still indexing; a large notebook reindex in progress; rapid repeated lookups triggering the rate limiter; importing a large dataset then immediately querying.

Related errors

• refresh OAuth credentials: %w
• Related operations are being processed, please try again lat
• Related operations are being processed, please try again lat
• notebook unindexed
• read custom emoji response failed: %w

AI-assisted analysis of siyuan-note/siyuan@251596fc0d (2026-08-12). Data as JSON: /api/errors/8a13ba8893cd7cfe. Report an issue: GitHub.