log backup status is normal and lag more and more after injection pd leader io delay 500ms last for 10mins and then recover

[Lily2025]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

1、run tpcc
2、run log backup
3、inject pd leader io delay 500ms last for 10mins and then recover

2. What did you expect to see? (Required)

lag can recover after fault recover

3. What did you see instead (Required)

log backup lag more and more after injection pd leader io delay 500ms last for 10mins and then recover

sh-5.1# /br log status --task-name=pitr --pd="tc-pd.endless-ha-test-oltp-pitr-tps-7655091-1-478.svc:2379"
Detail BR log in /tmp/br.log.2024-12-27T04.49.35Z
● Total 1 Tasks.

#1 <
name: pitr
status: ● NORMAL
start: 2024-12-25 18:23:57.664 +0000
end: 2090-11-18 14:07:45.624 +0000
storage: s3://tmp/ha-release-test
speed(est.): 0.00 ops/s
checkpoint[global]: 2024-12-26 11:10:37.101 +0000; gap=17h38m58s

4. What is your TiDB version? (Required)

./tidb-server -V
Release Version: v8.5.0
Edition: Community
Git Commit Hash: d13e52e
Git Branch: HEAD
UTC Build Time: 2024-12-23 03:40:21
GoVersion: go1.23.3
Race Enabled: false
Check Table Before Drop: false
Store: unistore
2024-12-26T17:49:31.016+0800

[Lily2025]

/type bug
/severity major
/assign Leavrth