Seedpeer and peer request scheduling cluster list bug

[Mr-ZhuXingYi]

Bug report:

I have two scheduler clusters with the following configurations:
manager\scheduler: v2.2.2-rc.0
dfdaemon: v0.2.36

Cluster 1:
scheduler：
IDC: ‘testIDC’
Location: ‘test01’
SchedulerClusterID: 1
Seedpeer:

      IDC: ‘testIDC’
      Location: ‘test01’
      ClusterID: 1

---

Cluster 2:
scheduler：
IDC: ‘testIDC’
Location: ‘test02’
SchedulerClusterID: 2

When I start the scheduler and seedpeer for Cluster 1, everything works fine. However, when I start the scheduler for Cluster 2, the seedpeer from Cluster 1 connects to the scheduler of Cluster 2.

When a peer or seedpeer requests a scheduler list, it does not include the cluster id field. The req does not include the cluster id field. In the SQL query, all healthy schedulers are not filtered by idc and location. The idc and location are indeed passed in the filtering method, but they are not used in the actual operation. The length of Schedulers is only checked, and then the sorting is returned. The peer will receive a cluster list, and it should take the first one.

So, I have two questions

1. Why doesn't seedpeer use clusterID when requesting a dispatch list
2. Why doesn't it use idc and location when filtering

[chlins]

@gaius-qi