Live Process Monitoring and Automatic Alert Snapshots

Professional

Beyond the aggregated key process metrics, Bleemeo lets you inspect the complete, live process list of any monitored host — much like running top or htop, but from your browser and with the full history of your fleet at hand. When a CPU or memory alert fires, Bleemeo also captures a snapshot of that process list automatically, so you can find out what was consuming the resources even after the incident is over.

Accessing the process view

From the agent list, open a host and select the Processes tab. The page shows every process reported by the Glouton agent on that host.

A stat strip at the top summarises the whole host at the moment the list was captured:

• CPU — total usage with a user / system / nice / iowait breakdown
• Memory — used / total, with buffers and cached
• Swap — used / total (shown only when swap is present)
• Load — 1, 5 and 15-minute load averages
• Uptime, Users and Tasks — with a running / sleeping / stopped / zombie count

Live and paused modes

The list opens paused: it loads once and then freezes, so sorting, scrolling and opening details are not disrupted. Toggle Live to stream updates — the table then refreshes every 10 seconds and a countdown shows the time to the next refresh. Toggle it back to pause at any time.

Tip

Keep the view paused while you investigate a specific process. Switch to Live only when you want to watch usage evolve in real time.

You can display the processes as a flat list or as a tree that reflects the parent/child (PID/PPID) relationships between processes.

Columns

Column	Description
PID	Process identifier
User	User the process runs as
Res	Resident memory (RSS) used by the process
Status	Running, sleeping, stopped or zombie
CPU	CPU usage in percent
Mem	Memory usage in percent
Time	Total CPU time consumed
Command	Process name and full command line

Processes running inside a container are flagged with their runtime icon (Docker, containerd, …) and link to the corresponding container view. Processes that belong to a discovered service link to that service’s metrics.

Filtering

To find a process quickly you can:

• Search by name or command line
• Filter by status (running, sleeping, stopped, zombie)
• Filter by user

Trends

The Trends panel opens historical charts of the host’s process metrics over time. These are the same process_* metrics described in Key Process Metrics, useful to correlate a spike seen in the live list with its evolution over the last hours or days.

Process snapshots

A snapshot is an automatic capture of the full process list, taken by Bleemeo when a CPU or memory alert fires on the host. It freezes exactly what was running — and how much CPU and memory each process was using — at the instant the alert triggered.

This is invaluable for post-mortem analysis: a memory or CPU alert often clears before you can connect to the host, leaving nothing to look at. With snapshots, you can open the alert afterwards and see precisely which process was responsible.

To review snapshots for a host, open its Processes tab and click Snapshots. The list shows every alert that produced a snapshot, with:

• the metric that triggered the alert (CPU or memory)
• the alert status (warning or critical)
• the recorded value
• the date and time it happened

Select View on any entry to open the captured process list, presented exactly like the live view — sortable and searchable — but frozen at the moment of the alert.

Note

Snapshots are created automatically; there is nothing to configure. They appear in the list after the next CPU or memory alert on the host. See Alerting to learn how CPU and memory thresholds are defined.

Disabling process collection

Collecting process information has a small impact on the agent’s CPU usage. If you do not need the process view and want to save some CPU cycles, you can disable process gathering by adding the following to /etc/glouton/conf.d/99-local.conf:

agent:
  process_exporter:
    enable: false