global config prox_url not respected in alertmanager configuration

[arnelasse]

What did you do?

• Defined a proxy_url in the Helm Chart under config.global.http_config.proxy_url.

What did you expect to see?

• msteamsv2_configs posting alert notification to Teams Channel

What did you see instead? Under which circumstances?

• No Notifications in Teams Channel, although Prometheus was firing alerts
• After i have added the proxy in the msteamsv2_configs section the Alerts it was working again

System information

Wer are running kube-prometheus-stack: 86.2.3 with quay.io/prometheus/alertmanager:v0.33.0

Alertmanager version

alertmanager, version 0.33.0 (branch: HEAD, revision: 5d3ceb55bf3775ea152dcdf3803bbbb2b4afed54)
  build user:       root@7a8b562c66e8
  build date:       20260612-15:35:56
  go version:       go1.26.4
  platform:         linux/amd64
  tags:             netgo

Alertmanager configuration file

config:
    global:
      http_config:
        proxy_url: "{{ .Environment.Values.proxy }}"
    route:
      group_by:
        - alertname
      group_interval: 5m
      receiver: "teams"
      routes:
        - matchers:
            - alertname =~ "InfoInhibitor|Watchdog|CPUThrottlingHigh"
          receiver: "null"

    receivers:
      - name: "null"
      - name: "teams"
        msteamsv2_configs:
          - webhook_url: "<webhook_url>"
            http_config:
              http_headers:
                teams_channel_id:
                  values:
                    - {{ .Values.monitoring.alertmanager.teams_channel_id }}

Prometheus version

Prometheus configuration file

Logs

`Logs error time=2026-07-06T08:25:45.731Z level=ERROR source=dispatch.go:580 msg="Notify for alerts failed" component=dispatcher aggrGroup="{}:{alertname=\"AlertmanagerFailedToSendAlerts\"}" num_alerts=1 err="teams/msteamsv2[0]: notify retry canceled after 3 attempts: Post \"<redacted>\": dial tcp <public_microsoft_ip>:443: connect: connection timed out"`

CPU profiles

No response

Memory profiles

No response

Number of alerts

No response

Number of silences

No response

prometheus-operator checklist

• I have confirmed this issue is not caused by a missing feature in prometheus-operator.
• I have reproduced this issue when running Alertmanager directly (without prometheus-operator).