Connect to your Contabo VPS over SSH and use top to watch overall CPU and memory alongside the process running your YouTube loop. Use htop when you want an interactive view, then compare what you see while the loop runs with a quiet period if one is available.
Those readings are observations, not a verdict that a server is adequate or overloaded. Interpret them against the resources assigned to your particular VPS and the work its media command performs: copying an already encoded stream is different from decoding, filtering, scaling or re-encoding video.
What CPU and RAM monitoring can tell you
A live process monitor helps answer practical questions: which processes are active, whether one is using more CPU than others, and how much memory the system reports as used or available. It can help you notice a change that coincides with a stream starting, a media file changing, or another service running. It does not explain the cause on its own.
The distinction between system-wide and process-level readings matters. Overall CPU activity includes the loop and every other running task. A loop process row gives you a more direct clue about the media workload, but it is not the whole server: operating-system tasks and other applications can also use resources. Read both views together rather than assuming the busiest process accounts for everything.
Memory needs similar care. A summary gives a view of the system as a whole; process rows help show how memory is attributed among tasks. Linux may use memory for caching, so a single displayed figure labelled “used” is not, by itself, evidence that the system is running out. Look at the available-memory context shown by the tool and whether the picture changes over time. If swap is shown, note it too, but do not treat one reading as a diagnosis.
A terminal snapshot is useful for immediate inspection, not a controlled benchmark. top refreshes as the system runs, and the figures can move with scene changes, media activity, background tasks and other conditions. A few observations across relevant moments tell you more than a glance at one screen. Neither a brief quiet period nor a momentary spike establishes what your channel will need on another day.
This is why no general CPU or RAM threshold can settle whether a Contabo VPS is “enough” for a continuous loop. You need the allocation of your own plan, the actual command and files in use, and observations made while that work is happening. Contabo describes different VPS resource allocations in its VPS documentation; check the current specification for the instance in your account rather than inferring capacity from a product family name.
Connect to your Contabo VPS over SSH
SSH opens a text-based session on the Linux VPS so you can run commands there. Use the server address and login configured for your instance, together with an SSH client available on your computer. The exact address, login method and client depend on your account and operating system; obtain those details from your VPS setup information rather than guessing them.
Once connected, you should see a shell prompt. If you manage more than one server, first confirm that the session is on the intended VPS. Then run:
top
The command runs on the VPS, so the numbers describe that machine rather than the computer from which you opened SSH. Keep the terminal session open while you inspect it. To leave top, use q. If you disconnect, you will no longer have that live view; this does not itself stop a separately running loop, but do not assume how your particular loop was launched without checking its setup.
If SSH does not connect, resolve that access issue before interpreting monitoring output. Verify the address and login details, and check that you are using the right account and server. Avoid copying a command or login instruction meant for a different distribution or VPS into your session without confirming that it applies.
The Contabo guide to checking running processes on a Linux VPS is a useful companion for thinking about what is active on a cloud machine while you stream. For the YouTube side of the setup, the guide to locating the stream server URL covers a different part of the chain: the destination settings do not tell you how much CPU your encoder uses.
Read CPU and memory in top
top presents a changing summary followed by rows for individual processes. The top portion is the system view; the rows below it are a process view. Exact labels and formatting can vary with the Linux distribution and version, so focus first on the broad categories and compare readings in the same environment rather than expecting every screen to look identical.
Look at the CPU summary to understand activity across the system. Depending on the display, it may distinguish time spent running user tasks, system tasks, and idle time, among other categories. The percentages are not a direct measure of video quality or YouTube delivery. They describe CPU activity during the sampling interval, and the loop's row may be one contributor among several.
Next, look at the memory summary. top commonly reports total, used, free, and other memory figures; the exact fields depend on the version. Consider the overall picture, including available memory where shown, and then check the process list for the loop's resident memory. Do not add together unrelated fields or compare differently labelled values as if they meant the same thing. If you need to track a trend, note the labels and figures at intervals under similar conditions.
The process list is usually ordered by a resource measure, often CPU by default. A process moving near the top may be consuming more CPU at that moment, but sorting can change and a brief rise is not proof of a persistent problem. Watch whether the same process remains prominent over several refreshes, and check what else is running. A media loop can also spawn helper processes, so a single row may not represent every part of its work.
For a first pass, record the time and context rather than trying to turn a display into a score: loop on or off, the media command or mode, and whether other work was running. That small note makes later comparison more useful. A long-running channel may have different activity during startup, a file transition or normal steady playback; distinguish those circumstances when you describe what you observed.
Use htop for an interactive process view
htop offers a more interactive process list and can make it easier to scan CPU, memory and swap indicators, sort processes, and search or inspect process relationships. Contabo's Linux VPS process guide describes these kinds of views. If it is already available, run:
htop
If the shell says the command is unavailable, that means it is not installed or not on the command path; it does not mean the VPS is unhealthy. Contabo's guidance gives package installation examples for particular Linux families: sudo apt install htop for Ubuntu or Debian, and sudo yum install htop or sudo dnf install htop for CentOS/RHEL-family systems. Confirm your distribution and package manager first. Do not run an installation command merely because it appears in a guide for another system.
In htop, CPU bars can make per-core activity easier to see than a single system summary. Memory and swap indicators add context, while sorting and searching can help locate a process among many rows. The display is still a view of current activity, not a forecast. Interpret the bars against the VPS's allocated resources and the media work underway, and note the displayed labels rather than assuming that colours or layouts mean the same thing across configurations.
Use the interactive features to inspect, not to experiment by stopping processes. htop can offer controls for sending signals or changing process priority, but those are operational actions, not monitoring steps. Ending a service to see whether a number falls could interrupt the stream or other software. If you do not know what a process does, identify it before changing it.
For persistent history, a terminal session may not be enough: it shows what is happening while you are looking, not a continuous record you can later review after disconnecting. Contabo support describes a Monitoring add-on for server resources, bandwidth and downtime, but availability and scope should be confirmed for your account and server. Check the current support guidance before depending on it. A dashboard is useful only if it captures the measures and time period relevant to the question you are investigating.
Identify the loop process without guessing
Find the process that actually runs the loop, rather than choosing a row because its name looks familiar. Depending on the setup, the media work might appear under a name such as ffmpeg, a playback program, or a script or supervisor that launched another process. The process command or tree view can help distinguish the launcher from the worker doing the media processing.
In top, inspect the command or process-name column and look for the process associated with the command you configured. In htop, use its search or tree view if available, and check parent-child relationships. A wrapper can launch a separate encoder, so the row for the wrapper may show little resource use even while its child does the work. Conversely, a process with a familiar name could be unrelated to this stream.
If you started the loop yourself, compare the visible command details with the command you intended to run. Do not publish or share command lines carelessly: they can contain stream keys or other credentials. A YouTube stream key is sensitive, so redact it from screenshots, notes and support messages. You can still record the executable, relevant media options and whether encoding or filtering is enabled without exposing the key.
Process names are clues, not proof. If you are uncertain, consult the instructions for the software or the person who configured the VPS. Avoid killing, reprioritising or restarting an unfamiliar process to test a theory. A monitoring exercise should not create the outage it was intended to explain.
The media file itself may matter too. If you convert or prepare a file before putting it on the VPS, the HandBrake conversion guide helps separate preparation work from what happens during the live loop. Conversion performed in advance is not the same as work the live process performs continuously, so confirm which operation is active when you take readings.
Compare loop and non-loop observations
A useful comparison starts with a question, such as whether CPU activity rises when the loop starts or whether memory keeps changing during playback. Observe the overall summaries and the identified loop process while the stream is running. If practical, capture the same views when the loop is not running, with other conditions as similar as you can manage. This is a troubleshooting comparison, not a controlled benchmark.
Write down context with the readings. Note whether the stream is running, the file or playlist in use, whether the command copies or encodes, and any other visible task that was active. Record observations at more than one moment, particularly if the system is transitioning between files or recovering from a restart. Do not pretend that two observations are directly comparable if the workload changed between them.
A quiet comparison period may not be easy to arrange on a channel that needs to remain live. Do not interrupt a broadcast solely to produce an idle baseline. You can still inspect the process and overall view during ordinary operation, and compare different phases that occur naturally, while making clear that the conditions differ. If you have logs or monitoring history, use them as additional context, not as a substitute for understanding what the loop was doing.
If the system appears busy, identify what is consuming resources before changing configuration. Check whether the loop is encoding, whether another process is active, and whether memory availability or swap behaviour is changing. A high-looking momentary figure alone does not establish a fault; persistent symptoms and the task the VPS is performing matter. If there are stream interruptions, investigate delivery and connection symptoms as well as CPU and RAM. The stream health troubleshooting guide addresses a separate set of clues that resource monitoring cannot resolve by itself.
Relate resource use to relay and re-encoding work
A “loop” describes the result viewers see, not necessarily the processing pipeline behind it. One process may send already encoded packets onward; another may decode the input, apply filters or scaling, and encode output again. Those jobs differ, so you cannot infer CPU demand from the fact that a video repeats or from the VPS brand alone.
The FFmpeg documentation describes stream copy as transferring an input stream's encoded packets without decoding, filtering or encoding them. That avoids those media-processing steps, but filters cannot be applied to packets that have not been decoded. See the FFmpeg documentation when checking what an option in your command means. A pipeline that decodes, filters, scales or re-encodes performs additional work; its observed resource use depends on the actual inputs, settings and environment.
Read the command that is actually running, not just a saved example or an old setup note. An option that selects stream copy has different implications from one that selects an encoder or filter. A pre-converted file may make a simpler live pipeline possible, but do not assume that is what your VPS is doing. Confirm the command and the process tree, then compare its resource use during operation.
The VPS allocation is the other half of the interpretation. Contabo's plans and specifications can change, and plan names alone do not tell you what resources your current instance has. Check the current listing or account details for the instance's CPU and RAM allocation. A published specification is not evidence that a particular workload will fit comfortably; only observations with your own command and conditions can inform that judgement. There is no universal threshold here that can substitute for that check.
When recurring resource questions are caused by keeping a personal computer on for the loop, a cloud-run workflow can remove the need to leave that computer operating continuously. StreamNeo can take away that specific always-on-computer burden by running an uploaded video as a YouTube live stream, but it does not change how you should assess a VPS or promise a particular channel outcome. Keep the monitoring question tied to the platform and process you are actually using.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Does top show the CPU and RAM used by my loop alone?
No. Its summary describes the system, while the process rows show individual processes. Find the loop worker and read its row alongside the overall figures, remembering that other processes can contribute to system activity.
Should I install htop if it is missing?
Only after confirming the VPS's Linux distribution and package manager. Contabo's examples differ by distribution family, and a command intended for another system may not apply. You can use top without installing anything extra.
What CPU percentage is too high for a Contabo VPS?
There is no universal threshold supported for every VPS and YouTube loop. Check the resources assigned to your instance, identify whether the command copies or re-encodes, and compare repeated observations with relevant symptoms and conditions.
Does a high CPU reading prove the loop is encoding?
No. It shows activity, not its cause. Inspect the actual process command and any child processes, then check whether the pipeline decodes, filters, scales or encodes rather than assuming from a process name or a single reading.