---
title: Responding to monitoring alerts
slug: how-tos/vtWP-responding-to-monitoring-alerts
docTags: n5hUNKFejaPplyvTaPYRJ,vg8QkSZx_1TOzYPB3tXmx
createdAt: 2026-08-10T09:12:36.000Z
---
An alert has fired. This page covers what to do next: deciding whether it is actionable, tuning it if it is noisy, and interpreting the alert message.
Two monitoring paths exist, and the response differs:
- **Standard monitoring** via the Factry Historian Portal, auto-configured per installation.
- **Custom monitoring** via your own Grafana alerting, for implementation-specific checks such as recalculations.
Work through the section that matches the alert you received.
:::hint{type="info"}
This page is about responding to alerts. For the Factry Historian Portal, setting up alerts, creating notification channels and assigning recipients is covered in [Managing alerts & notifications](docId\:XOCaCmXp7iiYvOYpTG32D). Portal roles, and who may acknowledge or silence an alert, are covered in [Managing users](docId:5z5IefC9GPfya92qP1H46).
:::
## Factry Historian Portal alerts
### Is the alert actionable?
Rule out the two most common non-issues first.
- **Many alerts at once, especially across one site.** This is almost always a single shared cause: the site's PCs or gateways are offline, or the network is down, so every collector alerts together. Restore the connection and the alerts clear together.
- **A known false positive.** Check the table below before investigating further.
:::hint{type="warning"}
**Acknowledging is not a fix, and it does not last.** An acknowledgement holds only while the alert stays active. As soon as the alert returns to normal and fires again, you are notified again. A silence lasts 7 days at most. For a condition that will keep coming back, such as a collector service that is deliberately stopped, pause the collector instead.
:::
### Common false positives
| Alert | What it usually means | Action |
| ------------------------------ | --------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Collecting insufficient points | Data arrives in bursts, not continuously (SQL, CSV, SAP and batch collectors). The alert is added automatically on install. | Increase the activation delay, or do not attach this alert to notification |
| Collector health / Last seen | The collector service is stopped on purpose | Pause the collector and leave it paused until the connection is set up. Acknowledging is not enough: it holds only while the alert is active, and a silence expires after 7 days |
:::hint{type="warning"}
**Collector health / Last seen** is not always false positive. See the table below for more information.
:::
### Alerts that usually need action
| Alert | What it usually means | Action |
| ------------------------------ | ------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Disk space above threshold | The drive is genuinely filling up | Free space on the machine or allocate more disk space. On customer-hosted servers that Factry cannot reach, the customer has to act |
| Memory or CPU above threshold | A process is consuming more than the default allows | Find the process.
- If that level is normal for this machine, ask Factry to raise the threshold instead of chasing it every time
- If the level is not normal for the machine, investigate the issue and troubleshoot |
| Collector health / Last seen | The collector should be running but is down or unreachable | Check the PC, the collector logs (for Last seen, logs can be found on collector server), the gateway and the VPN, then restart the collector service |
| Collecting insufficient points | A collector that normally streams continuously has slowed down or stopped | Check the source system and the collector configuration |
| License expiry | The license really is close to expiring | Contact the person in charge of the license and have them confirm with Factry that the license is updated before it expires |
### Tuning a noisy alert
:::hint{type="info"}
Alerts created automatically by the Factry Historian Portal can only be edited by Factry. You can see the current threshold and activation delay, but not change them.
:::
Two options:
1. Ask Factry to adjust the threshold or the activation delay on the automatic alert.
2. Create your own alert with your own threshold, and link it to the notification channel instead.
Adjustments that come up repeatedly:
- **Raise the threshold** when a resource sits stably above the default. Memory on a shared collector PC is often normal at 90% or 95%.
- **Increase the activation delay** for bursty collectors, so short gaps do not trigger.
- **Silence** during planned maintenance, then lift the silence as soon as the work is done. See Managing alerts & notifications.
:::hint{type="info"}
A Portal silence lasts 7 days at most. Maintenance that runs longer has to be re-silenced.
:::
### An alert fires but nobody is notified
Not every alert is linked to a notification channel by default. If an alert appears on the overview page but no email arrives, first check the link between the alert and the channel, then check the activation delay. See [Managing alerts & notifications](docId\:XOCaCmXp7iiYvOYpTG32D).
## Custom Grafana alerts
### Common false positives
| Alert | What it usually means | Action |
| -------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| No data, although the measurement has values | The rule filters on `status = Good`. A stopped collection or a bad status leaves no Good data behind, so the rule fires even though the Historian still shows values (Bad status) | Check the measurement status in the Historian. If a no-data gap should not alert, change the rule's no-data handling. For example choose Normal or keep the last state |
| Rapid switching between Alerting and Normal | An alert-rule execution error, for example the Grafana database file was briefly locked | Check the Grafana logs. Usually transient |
### Tuning a noisy alert
Custom rules are yours to edit, so tuning does not go through Factry.
- **Raise the threshold** when the measured value sits stably above the default.
- **Increase the pending period** so short excursions do not trigger.
- **Set the no-data handling** deliberately per rule: alerting, Normal, or keep the last state.
**Checking notification delivery**
Connect the alert directly to a contact point. That covers almost every case.
One alert routes to one contact point. If you need several delivery types, for example email and Slack, put them all inside that single contact point.
Routing one alert to several contact points is possible with notification policies. That is advanced use, and the setup differs between Grafana versions, so follow the **Grafana alerting documentation** for the version you run.
### Customizing the alert message
- Edit the text under **Contact points > Templates tab**, not under the Alertmanager section. The template only takes effect once it is selected in the contact point settings.
- To point the alert back at its panel, add the **Dashboard UID** and **Panel ID** under "Add details for your alert". This produces a link, not a graph. Use `{{ .PanelURL }}` in the template for a clickable link.
- To deliver to Microsoft Teams, create a "**Post to a channel when a webhook request is received**" workflow and use the webhook URL it returns.
:::hint{type="info"}
Showing an actual graph image inside the alert message requires the **Grafana Image Renderer service**. Ask Factry to set this up.
Grafana's alerting interface changes between versions. Check the **[Grafana alerting documentation](https://grafana.com/docs/grafana/latest/alerting/)** for the version you are running before following the steps above.
:::
## Related pages
- [Managing alerts & notifications](docId\:XOCaCmXp7iiYvOYpTG32D)
- [Managing users](docId:5z5IefC9GPfya92qP1H46)
- [Factry Historian Portal](docId\:VdQsSjkjfNn9rFDfkl4ty)