TiUP deploys TiDB Dashboard with multiple PD URLs, but Dashboard only supports one PD endpoint

[mayjiang0203]

Bug Report

Please answer these questions before submitting your issue. Thanks!

What did you do?

What did you expect to see?

What did you see instead?
When deploying a standalone TiDB Dashboard with TiUP to a cluster that has multiple PD nodes, TiUP generates a --pd argument containing all PD URLs:

--pd="http://127.0.0.1:23791,http://127.0.0.1:23792,http://127.0.0.1:23793" 

However, TiDB Dashboard treats the whole value as one PD endpoint. It passes the complete comma-separated string to the etcd client instead of splitting it into multiple endpoints.
As a result, Dashboard cannot connect to PD.

Dashboard logs show:

dial tcp: address 127.0.0.1:23791,http:: too many colons in address

It also repeatedly logs:

Failed to load dynamic config from etcd
error="context deadline exceeded"

Dashboard is shown as Up by tiup cluster display because its HTTP port is listening, but PD-dependent APIs fail.

What version of TiDB Dashboard are you using (./tidb-dashboard --version)?
[2026/08/19 14:24:37.760 +08:00] [INFO] [version.go:33] ["TiDB Dashboard started"] [internal-version=v8.5.1] [standalone=Yes] [pd-version=N/A] [build-time="2025-01-10 06:50:53"] [build-git-hash=5d1f7280d5d8a7f97b5791b31c7ee300ff31229b]

**Root Cause
TiUP collects all PD client URLs and joins them with commas before generating the Dashboard startup script:

PD: strings.Join(pds, ","),

Dashboard accepts --pd as a single string and creates the etcd client with:

Endpoints: []string{config.PDEndPoint},

Dashboard does not split the comma-separated value into separate endpoints.