Prometheus-master can not query any metrics when labelmap and hashmod are both configured in Prometheus-slave

[heayin]

I am using Prometheus Federation in k8s.

The master configuration is as follows：

global:
  scrape_interval: 10s
  scrape_timeout: 10s
scrape_configs:
- job_name: 'dc_prometheus'
  honor_labels: true
  metrics_path: /federate
  params:
    match[]:
    - '{__name__=~".+"}'
  static_configs:
    - targets:
      - 10.181.12.19:30092
      - 10.181.12.19:30093

"10.181.12.19:30093" is the ip-address of slave1.
"10.181.12.19:30092" is the ip-address of slave0.

The slave1 configuration is as follows:

global:
  external_labels:
    slave: 1
  scrape_interval: 10s
  scrape_timeout: 10s
scrape_configs:
- job_name: 'kubernetes-pod'

  kubernetes_sd_configs:
  - role: pod

  relabel_configs:
  - action: labelmap
    regex: __meta_kubernetes_pod_label_(.+)
  - source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
    action: replace
    target_label: __metrics_path__
    regex: (.+)
  - source_labels: [__address__]
    modulus: 2
    target_label: __tmp_hash
    action: hashmod
  - source_labels: [__tmp_hash]
    regex:     ^1$
    action: keep

I can query metrics successfully in slave1. But I can query nothing in master.
When querying in master, I tcpdumped "10.181.12.19:30093" . It showed that the slave sended the metrics to master.

However, I can query successfully in master when labelmap is removed in slave1 configuration .
The configuration is follows:

global:
  external_labels:
    slave: 1
  scrape_interval: 10s
  scrape_timeout: 10s
scrape_configs:
- job_name: 'kubernetes-pod'

  kubernetes_sd_configs:
  - role: pod

  relabel_configs:
  - source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
    action: replace
    target_label: __metrics_path__
    regex: (.+)
  - source_labels: [__address__]
    modulus: 2
    target_label: __tmp_hash
    action: hashmod
  - source_labels: [__tmp_hash]
    regex:     ^1$
    action: keep