The restart time under disagg is too long if some meta is not yet ready #8946
Labels
affects-7.5
affects-8.1
component/storage
severity/moderate
type/bug
The issue is confirmed as a bug.
Bug Report
Please answer these questions before submitting your issue. Thanks!
1. Minimal reproduce step (Required)
Deploy a disagg arch cluster with following tiflash wn config
2. What did you expect to see? (Required)
3. What did you see instead (Required)
The main thread after restart, we can see that
Waiting for restore checkpoint info from S3
block for 10 minutesThe thread of uploading wn checkpoint. In 02:30:57:364, the initialization from S3 is skipped because tmt context is not ready yet. Because
profiles.default.remote_checkpoint_interval_seconds
is set to 600 seconds, the retry happened at 02:40:57. So it block the main thread from restarting.The retry of
UniversalPageStorageService::uploadCheckpoint
after restart should be more frequent4. What is your TiFlash version? (Required)
master
The text was updated successfully, but these errors were encountered: