TiCDC doesn't report any error when log disk is full

[sleepymole]

What did you do?

1. Use tiup playground to deploy cluster with one TiCDC.

tiup --tag cdc-test playground --tiflash 0  --ticdc 1 v4.0.14

2. Stop cluster and enter the data directory.

cd .tiup/data/cdc-test
mv ticdc-0/ticdc.log .
mount -t tmpfs -o size=1M tmpfs ticdc-0
mv ticdc.log ticdc-0/ 
dd if=/dev/zero of=ticdc-0/zero  # should output "dd: writing to 'ticdc-0/zero': No space left on device"

3. Restart the cluster for several times.

What did you expect to see?

Every time I restart the cluster, some new log like "Welcome xxx" should be outputted to the log file or report some error on stdout or stderr.

What did you see instead?

After a few attempts, log file doesn't update any more and there is no error report on stdout or stderr.

Versions of the cluster

v4.0.14 (I think it probably exists in all versions, but I only tested it with v4.0.14)

[sleepymole]

It seems https://github.com/uber-go/zap/blob/aea2a5051d079dbe4606d990119c8b68f574ff59/options.go#L75 is not set to stdout or stderr.

[overvenus]

fixed by #3778