Destroy cost too much time if all pd were down
Open
Nobody has claimed this yet.
type/bug
- Dominant language
- Go
- Stars
- 466
- Forks
- 338
- Avg merge
- 3d 7h
- Merged PRs (30d)
- 8
Description
Bug Report
Please answer these questions before submitting your issue. Thanks!
- What did you do?
0. deploy a simple cluster with [tests/tiup-cluster/topo/full.yaml](tests/tiup-cluster/topo/full.yaml)
1. stop all pd nodes
```bash
tiup cluster stop cluster -R pd
```
2. destroy the cluster
```bash
time tiup cluster destroy cluster -y
...
Destroyed cluster `cluster` successfully
real 4m39.339s
user 0m2.308s
sys 0m1.491s
```
-
What did you expect to see?
It should takes less than one or two minutes.
-
What did you see instead?
-
What version of TiUP are you using (
tiup --version)?
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Reproduce the report by deploying tests/tiup-cluster/topo/full.yaml, stopping all PD nodes with tiup cluster stop cluster -R pd, and timing tiup cluster destroy cluster -y. Trace the cluster-destroy entry point used by that command and identify why the all-PD-down case takes about five minutes. Done means the same scenario completes in under one or two minutes and still reports successful destruction.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- go
- Domain
- cli
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 30/100