After TiFlash OOM, it keeps crashing and won't start again

Translated
1

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

Original topic: tiflash oom之后，一直crash，再也起不来

| username: gejibin

TiDB Version: 5.4.0
After TiFlash OOM, it enters a continuous crash—>running—>crash… infinite loop.

Even after disabling queries to TiFlash, TiFlash still starts up. All three TiFlash instances OOM almost simultaneously, and after that, none of the TiFlash instances can start up.

Refer to the logs below:

error.log (1.3 MB)
tiflash.log-1 (3.0 MB)
serverlog.log-1 (21.8 MB)

2
| username: tidb狂热爱好者 | Original post link

Drop the flash table. Is your TiFlash deployed separately or together with TiDB?

3
| username: Minorli-PingCAP | Original post link

What does the deployment architecture look like? And the configuration details.
Can replicas be disabled on-site to start?

4
| username: xfworld | Original post link

First, remove the existing synchronization configuration, then scale down all TiFlash nodes, and scale them back up to see if it works.

5
| username: gejibin | Original post link

Then my 3TB of data is gone, right?

6
| username: xfworld | Original post link

Data can be resynchronized as long as TiKV has the data.

7
| username: tidb狂热爱好者 | Original post link

The data is redundant. Just delete it first and then create it again.