# Unable to debug prometheus to alertmanager connection

**URL:** <https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195>\
**Category:** Prometheus server\
**Created:** [May 2, 2021, 11:31pm UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195 "2021-05-02T23:31:29Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![davehayes](https://dub1.discourse-cdn.com/flex017/user_avatar/discuss.prometheus.io/davehayes/32/90_2.png) [@davehayes](https://discuss.prometheus.io/u/davehayes)\
**Post date:** [May 2, 2021, 11:31pm UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195/1 "2021-05-02T23:31:29Z")

</div>

Trying to set up alertmanager on a running prometheus instance. I can see alerts via Grafana, so I know alerts are happening. However, there are no log entries (as there are with other ideas) that say “alertmanager is not found” or "alertmanager not working because ". I have absolutely no indication that alertmanager is not working except for the lack of alerts.

I would like to open a github issue about this, but I need to know why alertmanager is failing to alert first, so I can ask for the inclusion of any log message that will give users a fighting chance to find out what is wrong.

I have the following in prometheus config:

```
alerting:
  alertmanagers:
    - static_configs:
      - targets:
        - 1.2.3.4:8765

```

I start alertmanager with

```
--cluster.listen-address='' --web.listen-address= 1.2.3.4:8765 --log.level=debug

```

Why isn’t this working?

Thanks in advance.

---

<div class="post-metadata">

**Author:** ![roidelapluie](https://dub1.discourse-cdn.com/flex017/user_avatar/discuss.prometheus.io/roidelapluie/32/6_2.png) [@roidelapluie](https://discuss.prometheus.io/u/roidelapluie)\
**Post date:** [May 11, 2021, 9:32am UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195/2 "2021-05-11T09:32:20Z")

</div>

Do you see your alertmanager in the Runtime page of Prometheus?

Do you see alerts in the Alertmanager status page?

What are your prometheus and alertmanager versions?

---

<div class="post-metadata">

**Author:** ![davehayes](https://dub1.discourse-cdn.com/flex017/user_avatar/discuss.prometheus.io/davehayes/32/90_2.png) [@davehayes](https://discuss.prometheus.io/u/davehayes)\
**Post date:** [May 11, 2021, 9:39pm UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195/3 "2021-05-11T21:39:26Z")

</div>

Ok, yes I see alertmanager in the runtime page. I waited for an actual alert (I have grafana monitoring ALERTS\_FOR\_STATE atm) and -now- I see the same alert but not in the status page…on the alerts page.

I am running 0.20.0 of alertmanager and 2.26.0 of prometheus.

I had no indication (that I was familiar with) that alertmanager was even connected and working. Now that I see an alert, I think I can continue with trying to configure a few destinations.

---

<div class="post-metadata">

**Author:** ![roidelapluie](https://dub1.discourse-cdn.com/flex017/user_avatar/discuss.prometheus.io/roidelapluie/32/6_2.png) [@roidelapluie](https://discuss.prometheus.io/u/roidelapluie)\
**Post date:** [May 12, 2021, 7:58am UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195/4 "2021-05-12T07:58:17Z")

</div>

To clarify, do you see the alerts in the alertmanager UI?

---

<div class="post-metadata">

**Author:** ![davehayes](https://dub1.discourse-cdn.com/flex017/user_avatar/discuss.prometheus.io/davehayes/32/90_2.png) [@davehayes](https://discuss.prometheus.io/u/davehayes)\
**Post date:** [May 12, 2021, 5:42pm UTC](https://discuss.prometheus.io/t/unable-to-debug-prometheus-to-alertmanager-connection/195/5 "2021-05-12T17:42:13Z")

</div>

I only saw one alert briefly.

My issue is that when I restart the alertmanager, I don’t see a log message indicating that. In fact, the log file is purged apparently, so I am left with zero feedback to ensure that a configuration file change was accepted or works. Any previous log file is deleted, which is rather surprising. Apparently, the only way I can test alertmanager is to generate an alert myself with the alert manager UI open.