CloudLinux charts use several indicators that describe different resources. Understanding them helps you give support useful information and avoid changing plans because of the wrong number. The limits that apply are those of your service; the vendor’s examples are not specifications for a Nuagerie hosting plan.
CPU or SPEED: processing capacity
SPEED expresses a processing limit. When stated as a percentage of one core, 100% corresponds to the equivalent of one core. It does not mean that the entire server processor is busy. The interface can also show a percentage normalized to your allocation. Read the legend and displayed limit before comparing two numbers.
PMEM: physical memory
PMEM concerns physical memory accounted for within the LVE environment. Several processes can contribute to that usage. It differs from PHP’s memory_limit setting, which governs script allocations. Raising that setting does not increase the account’s memory limit and is not a solution by itself.
The CloudLinux PMEM documentation explains this scope. Read a memory error in context: it does not necessarily identify the same mechanism as a PMEM counter. Keep the complete error message when asking for assistance, with sensitive details removed.
IO and IOPS: two measures of disk activity
IO measures disk read and write throughput. IOPS counts read and write operations per second. Many small operations and one large transfer can put different pressure on these limits. These indicators measure neither remaining storage space nor your plan’s network traffic allowance.
EP and NPROC: processes, not visitors
EP counts entries into the LVE. Its interpretation depends on the web server and PHP execution mode; LiteSpeed can behave differently from Apache. Do not translate this number into a count of simultaneous visitors or daily visits.
NPROC concerns processes and threads within the LVE. Child processes can exist without adding the same number of EP entries. The counters therefore answer different questions. CloudLinux explains this distinction between EP and NPROC.
Usage, limits and faults
Compare observed usage, the limit and the Faults counters for the same period. An average can conceal a short peak. Conversely, a high value does not establish that a command failed. A fault counter is not a count of lost visitors.
The consequences depend on the resource. Processing or disk limits can slow work; other limits can prevent new processes or interrupt work. Errors such as 500, 503 or 508 can occur in some circumstances, but the status code alone cannot identify the cause. Consult the official explanation of limits.
Turn measurements into a useful decision
Keep the period, time zone, resource name, usage, limit and observed symptom. Our guide to viewing measurements explains where to find this information. To connect a peak to website activity, continue with our diagnostic guide.
An indicator does not automatically determine whether you need an upgrade. Your plan specifications, optimization options and hosting upgrade criteria complete the assessment. Send your observations through your client area so Nuagerie can help identify a response that fits the actual need.
