TiCDC Task Rebuild Failure

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: TiCDC重建同步任务失败

| username: TiDBer_hwEZA4rV

v6.5.2
Found ticdc synchronization task failure
{
“id”: “kunta2bigdata”,
“namespace”: “default”,
“summary”: {
“state”: “failed”,
“tso”: 441737524319092736,
“checkpoint”: “2023-05-26 17:55:26.144”,
“error”: {
“addr”: “192.168.86.93:8300”,
“code”: “CDC:ErrGCTTLExceeded”,
“message”: “[CDC:ErrGCTTLExceeded]the checkpoint-ts(441737524319092736) lag of the changefeed({default kunta2bigdata}) has exceeded the GC TTL”
}
}
},
Deleted the task and recreated it, but failed
tiup cdc cli changefeed create --disable-gc-check --server=http://XXXXX:8300 --sink-uri=“mysql://XXXXX:XXXXX@tidb-XXXXX:4000?safe-mode=true” --changefeed-id=“XXXXX2bigdata” --start-ts=441737524319092736 --config=cdc.yaml
tiup is checking updates for component cdc …
Starting component cdc: /home/admin/.tiup/components/cdc/v6.5.2/cdc cli changefeed create --disable-gc-check --server=http://ticdc-XXXXX:8300 --sink-uri=mysql://repl:XXXXX@tidb-XXXXX:4000?safe-mode=true --changefeed-id=kunta2bigdata --start-ts=441737524319092736 --config=cdc.yaml
Replicate lag (65h39m57.4s) is larger than 1 day, large data may cause OOM, confirm to continue at your own risk [Y/N]
Y
Error: [CDC:ErrMetaListDatabases]meta store list databases: [tikv:9006]GC life time is shorter than transaction duration, transaction starts at 2023-05-26 17:55:26.144 +0800 CST, GC safe point is 2023-05-29 11:25:16.295 +0800 CST

| username: xfworld | Original post link

This error description is very clear.

What is your scenario? What is your request?
It feels like the template was set up for nothing… still have to ask again…

| username: TiDBer_hwEZA4rV | Original post link

The request is to resynchronize the data to the downstream from the point of failure.

| username: 考试没答案 | Original post link

Parse the binlog. Manually import it into the database up to the most recent time, then specify a recent point in time to check.

| username: zhanggame1 | Original post link

It seems we need to start over.

| username: tidb菜鸟一只 | Original post link

The GC time is not enough, and the GC at the time of your issue has already been cleaned up. You can only resynchronize.

| username: xingzhenxiang | Original post link

It feels like monitoring and alerts need to be added. It’s been several days apart, and issues should be fixed promptly to avoid the end of the GC cycle when there is no relevant data left.