[Bug] Cluster down during regress/createdb

[Ray-Eldath]

Cloudberry Database version

No response

What happened

https://github.com/cloudberrydb/cloudberrydb/actions/runs/6876511531/job/18702700167?pr=291

regress failed in just a few minutes. createdb failed and so does many subsequent, dependent tests. cluster verification also failed due to a down in a mirror.

What you think should happen instead

not sure if it's related to external-fts. but I'll say the possibility is high.

How to reproduce

n/a

Operating System

ci

Anything else

No response

Are you willing to submit PR?

• Yes, I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct.

[Ray-Eldath]

it isn't. singlenode just exposed the same failure. https://github.com/cloudberrydb/cloudberrydb/actions/runs/6902756041/job/18780540313

[Ray-Eldath]

same port-in-use errors happened here as in #262. maybe they have a same root cause.

edit: very likely due to ci environment error. in the case of createdb failure (https://github.com/cloudberrydb/cloudberrydb/actions/runs/6902756041/job/18780540313), there aren't any restart of the cluster. it's a fresh environment, but standby still cannot bind to address.