====== OS Agent - Monitor ======

The **Monitor** tab lists every host that pushes metrics to this Collector, with their status and last activity. It is also where you can link a host to a System, deactivate it, or remove it.

===== Host list =====

Each row shows:

| Column | Meaning |
| **Hostname** | The host's network name as reported in the ''host'' tag of the push. Case-insensitive, used as the unique identifier inside the Collector. |
| **System** | The linked System short name, or ''-'' if the host is //standalone//. |
| **Status** | ''OK'' / ''ERROR'' / ''OFF''. See below. |
| **OS** | ''linux'', ''windows'', or empty when not detected. |
| **Last Push** | Time of the last received push. ''Never'' if the host has been registered but never pushed. |
| **Active** | ''ON'' (accepts pushes) or ''OFF'' (rejects pushes with //403//). |

Click a row to **expand the detail** for the host.

==== Status values ====

  * ''OK'' (green) - the host is active and has pushed within the last 5 minutes
  * ''ERROR'' (red) - the host is active but has not pushed in the last 5 minutes
  * ''OFF'' (grey) - the host is set inactive in the Collector - all incoming pushes for this host are rejected

The 5-minute threshold is fixed in the Collector. Telegraf default flush interval is 15 seconds, so a missed push of more than 20 cycles is a real problem (network down, Telegraf crashed, host shut down).

===== Host detail (expanded) =====

When you click a host row, the panel below shows the **last received values**:

==== Top metrics grid ====

Four big cards: **CPU**, **Memory**, **Swap**, **Load Average**. Each shows the latest known value (used percent for CPU/Mem/Swap, 1-minute load with 5m/15m below).

==== Disk Usage ====

A table with one row per mount point (Linux) or drive letter (Windows), with the used percent. Color codes:

  * < 75% → green
  * 75-90% → orange
  * > 90% → red

==== Top Processes ====

A table with the top processes ranked by **CPU**, **Memory**, or **Disk I/O** (button toggles at the top right of the panel). Up to 20 processes are listed.

  * **CPU %** - per-process CPU usage as reported by Telegraf
  * **Memory** - resident set size (RSS) - actual physical memory used
  * **Disk I/O** - read and write bytes since the process started

If you do not see any process, the most likely causes are:

  * The **process** input is not enabled in the Configuration tab → enable it and redeploy the agent
  * The agent has not pushed yet → wait 30 seconds (procstat sends every 30s by default)
  * On Linux Docker without ''SYS_PTRACE'', the process list works but read/write bytes will be zero (see the [[products:promonitor:latest:userguide:administration:os-agent:configuration#process_monitoring_procstat|process monitoring section]])

==== Bottom info row ====

  * **Processes** - total / running / zombies (from the ''processes'' input - Linux only)
  * **Network** - bytes sent (↑) and received (↓) per interface, in human-readable units
  * **First Seen** - when the host was first auto-discovered

===== Actions per host =====

==== Remap to a System ====

Use the **Remap to System** dropdown to link the host to a System (or set it back to //Standalone//).

  * When a host is linked to a System, its metrics get the System tags automatically (sid, system name, etc.) - dashboards by System include them
  * When standalone, metrics are stored with only the host tag - they still go through the pipeline but they are not associated to any System

Click **Save** to apply the link.

==== Activate / Deactivate ====

The **Activate / Deactivate** button (pause icon when ON, play when OFF) toggles the **Active** flag on the host:

  * **OFF** rejects every incoming push from this host with //403 Forbidden//
  * **ON** accepts pushes again

Useful when you want to silence a host temporarily without uninstalling Telegraf.

==== Delete ====

Removes the host from the database. The next push from this host will re-create it as a new entry (with //Standalone// link, so you will need to re-map it).

Use this to clean up old hosts that no longer exist.

===== Refresh =====

A **Refresh** button at the top of the panel reloads the host list and detail. The list does not auto-refresh - click Refresh after waiting a few seconds to see the new values.

===== Auto-discovery =====

Hosts appear automatically the first time they push successfully:

  - The Collector parses the ''host'' tag from the first metric line
  - It checks the in-memory map and the database
  - If unknown, tries to **resolve a System** by matching the hostname (case-insensitive) against the existing connectors
  - If it matches a connector, the host is auto-linked to that System
  - Otherwise the host is //standalone//

You do **not** need to pre-register hosts. Just install Telegraf and they show up.