METRICS_REFERENCE.md

proxtop Metrics Reference

This document provides a comprehensive reference of all metrics collected by proxtop, organized by collector. Each metric includes its data source, description, collection cycle, and availability mode.

Legend

Symbol	Meaning
🔄 lookup	Metric is collected during the lookup phase (VM discovery)
📊 collect	Metric is collected during the collect phase (periodic sampling)
📝 verbose	Only available with --verbose flag
🔒 internal	Internal metric, not exported but used for calculations
⚠️ root	Requires root/sudo privileges

---

Table of Contents

1. Base VM Metrics
2. CPU Collector
3. Memory Collector
4. Network Collector
5. Disk Collector
6. I/O Collector
7. Host Collector
8. PSI Collector
9. Metric Calculations

---

Base VM Metrics

These metrics are always collected for each virtual machine, regardless of which collectors are enabled.

Metric	Source	Description	Unit
UUID	libvirt/QMP	Unique identifier for the virtual machine	string
name	libvirt/QMP	Human-readable name of the virtual machine	string
PID	/proc	Process ID of the QEMU process on the host	integer

---

CPU Collector (--cpu)

Monitors CPU utilization, frequency, and steal time for both the host and virtual machines.

Host Metrics

Metric	Source	Description	Unit	Cycle
cpu_cores	/proc/cpuinfo	Total number of physical CPU cores	count	🔄 lookup
cpu_curfreq	/sys/devices/system/cpu/cpu*/cpufreq	Current mean frequency across all CPU cores	MHz	🔄 lookup
cpu_user	/proc/stat	Time spent executing user-space processes	%	📊 collect
cpu_system	/proc/stat	Time spent executing kernel-space code	%	📊 collect
cpu_idle	/proc/stat	Time spent idle (no tasks running)	%	📊 collect
cpu_steal	/proc/stat	Time stolen by hypervisor for other VMs	%	📊 collect

Verbose Mode Host Metrics 📝

Metric	Source	Description	Unit	Cycle
cpu_minfreq	/sys/devices/system/cpu/cpu*/cpufreq	Minimum supported CPU frequency	MHz	🔄 lookup
cpu_maxfreq	/sys/devices/system/cpu/cpu*/cpufreq	Maximum supported CPU frequency	MHz	🔄 lookup
cpu_nice	/proc/stat	Time spent on niced (low priority) user processes	%	📊 collect
cpu_iowait	/proc/stat	Time spent waiting for I/O completion	%	📊 collect
cpu_irq	/proc/stat	Time spent handling hardware interrupts	%	📊 collect
cpu_softirq	/proc/stat	Time spent handling software interrupts	%	📊 collect
cpu_guest	/proc/stat	Time spent running guest VMs	%	📊 collect
cpu_guestnice	/proc/stat	Time spent running niced guest VMs	%	📊 collect

VM Metrics

Metric	Source	Description	Unit	Cycle
cpu_cores	libvirt/QMP	Number of virtual CPU cores assigned to VM	count	🔄 lookup
cpu_total	calculated	Total CPU utilization across all vCPUs	%	📊 collect
cpu_steal	calculated	CPU time stolen due to host contention	%	📊 collect

Verbose Mode VM Metrics 📝

Metric	Source	Description	Unit	Cycle
cpu_other_total	calculated	CPU overhead from I/O and emulation threads	%	📊 collect
cpu_other_steal	calculated	Steal time for overhead threads	%	📊 collect

Internal Metrics 🔒

Metric	Source	Description	Cycle
cpu_threadIDs	libvirt + /proc	List of thread IDs for vCPU threads	🔄 lookup
cpu_otherThreadIDs	libvirt + /proc	List of thread IDs for non-vCPU threads	🔄 lookup
cpu_times_${pid}	/proc/${pid}/schedstat	CPU time counter for each vCPU thread	📊 collect
cpu_runqueues_${pid}	/proc/${pid}/schedstat	Run queue wait time for each vCPU thread	📊 collect
cpu_other_times_${pid}	/proc/${pid}/schedstat	CPU time for overhead threads	📊 collect
cpu_other_runqueues_${pid}	/proc/${pid}/schedstat	Run queue wait for overhead threads	📊 collect

---

Memory Collector (--mem)

Monitors memory allocation, usage, and page faults.

Host Metrics

Metric	Source	Description	Unit	Cycle
ram_Total	/proc/meminfo	Total physical RAM installed	KB	📊 collect
ram_Free	/proc/meminfo	Completely unused RAM	KB	📊 collect
ram_Available	/proc/meminfo	RAM available for new allocations	KB	📊 collect

Verbose Mode Host Metrics 📝

Metric	Source	Description	Unit	Cycle
ram_Buffers	/proc/meminfo	Memory used for kernel buffers	KB	📊 collect
ram_Cached	/proc/meminfo	Memory used for page cache	KB	📊 collect
ram_SwapCached	/proc/meminfo	Swap memory also in RAM	KB	📊 collect
ram_Active	/proc/meminfo	Recently used memory	KB	📊 collect
ram_Inactive	/proc/meminfo	Less recently used memory	KB	📊 collect
ram_ActiveAnon	/proc/meminfo	Active anonymous memory	KB	📊 collect
ram_InactiveAnon	/proc/meminfo	Inactive anonymous memory	KB	📊 collect
ram_ActiveFile	/proc/meminfo	Active file-backed memory	KB	📊 collect
ram_InactiveFile	/proc/meminfo	Inactive file-backed memory	KB	📊 collect
ram_Unevictable	/proc/meminfo	Memory that cannot be reclaimed	KB	📊 collect
ram_Mlocked	/proc/meminfo	Memory locked with mlock()	KB	📊 collect
ram_SwapTotal	/proc/meminfo	Total swap space	KB	📊 collect
ram_SwapFree	/proc/meminfo	Unused swap space	KB	📊 collect
ram_Dirty	/proc/meminfo	Memory waiting to be written to disk	KB	📊 collect
ram_Writeback	/proc/meminfo	Memory actively being written to disk	KB	📊 collect
ram_AnonPages	/proc/meminfo	Anonymous mapped memory	KB	📊 collect
ram_Mapped	/proc/meminfo	Files mapped into memory	KB	📊 collect
ram_Shmem	/proc/meminfo	Shared memory (tmpfs, etc.)	KB	📊 collect
ram_Slab	/proc/meminfo	Kernel slab allocator memory	KB	📊 collect
ram_SReclaimable	/proc/meminfo	Reclaimable slab memory	KB	📊 collect
ram_SUnreclaim	/proc/meminfo	Non-reclaimable slab memory	KB	📊 collect
ram_KernelStack	/proc/meminfo	Kernel stack memory	KB	📊 collect
ram_PageTables	/proc/meminfo	Memory for page tables	KB	📊 collect
ram_NFSUnstable	/proc/meminfo	NFS pages not yet committed	KB	📊 collect
ram_Bounce	/proc/meminfo	Bounce buffer memory	KB	📊 collect
ram_WritebackTmp	/proc/meminfo	Temporary writeback memory	KB	📊 collect
ram_CommitLimit	/proc/meminfo	Total memory available for allocation	KB	📊 collect
ram_CommittedAS	/proc/meminfo	Total memory committed	KB	📊 collect
ram_VmallocTotal	/proc/meminfo	Total vmalloc address space	KB	📊 collect
ram_VmallocUsed	/proc/meminfo	Used vmalloc space	KB	📊 collect
ram_VmallocChunk	/proc/meminfo	Largest contiguous vmalloc block	KB	📊 collect
ram_HardwareCorrupted	/proc/meminfo	Memory with hardware errors	KB	📊 collect
ram_AnonHugePages	/proc/meminfo	Anonymous huge pages	KB	📊 collect
ram_ShmemHugePages	/proc/meminfo	Shared memory huge pages	KB	📊 collect
ram_ShmemPmdMapped	/proc/meminfo	Shared memory PMD mapped	KB	📊 collect
ram_HugePagesTotal	/proc/meminfo	Total huge pages configured	count	📊 collect
ram_HugePagesFree	/proc/meminfo	Free huge pages	count	📊 collect
ram_HugePagesRsvd	/proc/meminfo	Reserved huge pages	count	📊 collect
ram_HugePagesSurp	/proc/meminfo	Surplus huge pages	count	📊 collect
ram_Hugepagesize	/proc/meminfo	Size of each huge page	KB	📊 collect
ram_Hugetlb	/proc/meminfo	Total huge page memory	KB	📊 collect
ram_DirectMap4k	/proc/meminfo	Memory mapped with 4K pages	KB	📊 collect
ram_DirectMap2M	/proc/meminfo	Memory mapped with 2M pages	KB	📊 collect
ram_DirectMap1G	/proc/meminfo	Memory mapped with 1G pages	KB	📊 collect

VM Metrics

Metric	Source	Description	Unit	Cycle
ram_total	libvirt/QMP	Maximum memory the VM can use	KB	🔄 lookup
ram_used	libvirt/QMP	Currently allocated memory	KB	🔄 lookup

Verbose Mode VM Metrics 📝

Metric	Source	Description	Unit	Cycle
ram_vsize	/proc/${pid}/stat	Virtual memory size of QEMU process	bytes	📊 collect
ram_rss	/proc/${pid}/stat	Resident set size (physical memory used)	bytes	📊 collect
ram_minflt	/proc/${pid}/stat	Minor page faults (no disk I/O)	count	📊 collect
ram_cminflt	/proc/${pid}/stat	Minor faults including children	count	📊 collect
ram_majflt	/proc/${pid}/stat	Major page faults (required disk I/O)	count	📊 collect
ram_cmajflt	/proc/${pid}/stat	Major faults including children	count	📊 collect

---

Network Collector (--net)

Monitors network traffic across physical and virtual interfaces.

Host Metrics

Metric	Source	Description	Unit	Cycle
net_host_receivedBytes	/proc/net/dev	Total bytes received (sum over relevant interfaces)	bytes	📊 collect
net_host_transmittedBytes	/proc/net/dev	Total bytes transmitted (sum over relevant interfaces)	bytes	📊 collect
net_host_speed	/sys/class/net/${dev}/speed	Network device maximum link speed	Mbps	🔄 lookup

Verbose Mode Host Metrics 📝

Metric	Source	Description	Unit	Cycle
net_host_receivedPackets	/proc/net/dev	Total packets received	count	📊 collect
net_host_receivedErrs	/proc/net/dev	Receive errors	count	📊 collect
net_host_receivedDrop	/proc/net/dev	Dropped incoming packets	count	📊 collect
net_host_receivedFifo	/proc/net/dev	FIFO buffer errors (receive)	count	📊 collect
net_host_receivedFrame	/proc/net/dev	Framing errors on receive	count	📊 collect
net_host_receivedCompressed	/proc/net/dev	Compressed packets received	count	📊 collect
net_host_receivedMulticast	/proc/net/dev	Multicast frames received	count	📊 collect
net_host_transmittedPackets	/proc/net/dev	Total packets transmitted	count	📊 collect
net_host_transmittedErrs	/proc/net/dev	Transmit errors	count	📊 collect
net_host_transmittedDrop	/proc/net/dev	Dropped outgoing packets	count	📊 collect
net_host_transmittedFifo	/proc/net/dev	FIFO buffer errors (transmit)	count	📊 collect
net_host_transmittedColls	/proc/net/dev	Packet collisions during transmit	count	📊 collect
net_host_transmittedCarrier	/proc/net/dev	Carrier losses during transmit	count	📊 collect
net_host_transmittedCompressed	/proc/net/dev	Compressed packets transmitted	count	📊 collect

Internal Metrics 🔐

Metric	Source	Description	Cycle
net_host_ifs	libvirt	List of relevant physical network interfaces	🔄 lookup

VM Metrics

Metric	Source	Description	Unit	Cycle
net_receivedBytes	/proc/${pid}/net/dev	Bytes received by this VM	bytes	📊 collect
net_transmittedBytes	/proc/${pid}/net/dev	Bytes transmitted by this VM	bytes	📊 collect

Verbose Mode VM Metrics 📝

Metric	Source	Description	Unit	Cycle
net_receivedPackets	/proc/${pid}/net/dev	Packets received	count	📊 collect
net_receivedErrs	/proc/${pid}/net/dev	Receive errors	count	📊 collect
net_receivedDrop	/proc/${pid}/net/dev	Dropped incoming packets	count	📊 collect
net_receivedFifo	/proc/${pid}/net/dev	FIFO buffer errors (receive)	count	📊 collect
net_receivedFrame	/proc/${pid}/net/dev	Framing errors	count	📊 collect
net_receivedCompressed	/proc/${pid}/net/dev	Compressed packets received	count	📊 collect
net_receivedMulticast	/proc/${pid}/net/dev	Multicast frames received	count	📊 collect
net_transmittedPackets	/proc/${pid}/net/dev	Packets transmitted	count	📊 collect
net_transmittedErrs	/proc/${pid}/net/dev	Transmit errors	count	📊 collect
net_transmittedDrop	/proc/${pid}/net/dev	Dropped outgoing packets	count	📊 collect
net_transmittedFifo	/proc/${pid}/net/dev	FIFO buffer errors (transmit)	count	📊 collect
net_transmittedColls	/proc/${pid}/net/dev	Packet collisions	count	📊 collect
net_transmittedCarrier	/proc/${pid}/net/dev	Carrier losses	count	📊 collect
net_transmittedCompressed	/proc/${pid}/net/dev	Compressed packets transmitted	count	📊 collect

Internal Metrics 🔐

Metric	Source	Description	Cycle
net_interfaces	libvirt	List of virtual network interfaces for this VM	🔄 lookup

---

Disk Collector (--disk)

Monitors block storage devices and I/O operations. Use --storedev to manually specify host storage devices. This collector provides esxtop-style storage metrics.

Host Metrics

Metric	Display Name	Source	Description	Unit	Cycle
disk_device_reads	READS/s	/proc/diskstats	Read operations per second	ops/s	📊 collect
disk_device_writes	WRITES/s	/proc/diskstats	Write operations per second	ops/s	📊 collect
disk_device_mbread	MBRD/s	/proc/diskstats	MB read per second (sectors × 512)	MB/s	📊 collect
disk_device_mbwrite	MBWR/s	/proc/diskstats	MB written per second (sectors × 512)	MB/s	📊 collect
disk_device_ioutil	%UTIL	calculated	Device busy percentage (time doing I/Os)	%	📊 collect
disk_device_currentops	QDEPTH	/proc/diskstats	Current I/O queue depth (ops in progress)	count	📊 collect
disk_device_queuelen	QLEN	calculated	Average queue length (weighted time / elapsed)	count	📊 collect
disk_device_servicetime	SVCTM	calculated	Average service time per I/O	ms	📊 collect
disk_device_await	AWAIT	calculated	Average wait time per I/O (queue + service)	ms	📊 collect

Verbose Mode Host Metrics 📝

Metric	Source	Description	Unit	Cycle
disk_device_readsmerged	/proc/diskstats	Adjacent reads merged	count	📊 collect
disk_device_sectorsread	/proc/diskstats	Sectors read	count	📊 collect
disk_device_timereading	/proc/diskstats	Time spent reading	ms	📊 collect
disk_device_writesmerged	/proc/diskstats	Adjacent writes merged	count	📊 collect
disk_device_sectorswritten	/proc/diskstats	Sectors written	count	📊 collect
disk_device_timewriting	/proc/diskstats	Time spent writing	ms	📊 collect
disk_device_timeforops	/proc/diskstats	Total time spent on I/Os	ms	📊 collect
disk_device_weightedtimeforops	/proc/diskstats	Weighted time doing I/Os	ms	📊 collect
disk_device_count	/proc/diskstats	Number of relevant disks	count	🔄 lookup

Internal Metrics 🔐

Metric	Source	Description	Cycle
disk_sources	libvirt	List of relevant disk devices	🔄 lookup

VM Metrics

Metric	Display Name	Source	Description	Unit	Cycle
disk_size_capacity	SIZE	libvirt/QMP	Maximum virtual disk capacity	bytes	🔄 lookup
disk_size_allocation	ALLOC	libvirt/QMP	Currently allocated disk space	bytes	🔄 lookup
disk_ioutil	%UTIL	calculated	Estimated I/O utilization for VM	%	📊 collect
disk_reads	READS/s	libvirt/QMP	Read operations per second	ops/s	📊 collect
disk_writes	WRITES/s	libvirt/QMP	Write operations per second	ops/s	📊 collect
disk_mbread	MBRD/s	libvirt/QMP	MB read per second	MB/s	📊 collect
disk_mbwrite	MBWR/s	libvirt/QMP	MB written per second	MB/s	📊 collect
disk_lat_rd	LAT/rd	libvirt/QMP	Average read latency	ms	📊 collect
disk_lat_wr	LAT/wr	libvirt/QMP	Average write latency	ms	📊 collect
disk_lat_fl	LAT/fl	libvirt/QMP	Average flush latency	ms	📊 collect
disk_lat_avg	LAT/avg	calculated	Average latency across all ops	ms	📊 collect

Verbose Mode VM Metrics 📝

Metric	Display Name	Source	Description	Unit	Cycle
disk_size_physical	PHYSICAL	libvirt	Physical space for virtual disks	bytes	🔄 lookup
disk_stats_flushreq	FLUSH/s	libvirt	Cache flush requests per second	ops/s	📊 collect
disk_stats_rdtotaltimes	RDTM	libvirt	Total time spent on reads	ms	📊 collect
disk_stats_wrtotaltimes	WRTM	libvirt	Total time spent on writes	ms	📊 collect
disk_stats_flushtotaltimes	FLTM	libvirt	Total time spent on flushes	ms	📊 collect
disk_delayblkio	BLKIO	/proc/${pid}/stat	Aggregated block I/O delays	ticks	📊 collect

Internal Metrics 🔐

Metric	Source	Description	Cycle
disk_sources	libvirt	List of virtual disk sources	🔄 lookup

---

I/O Collector (--io) ⚠️

Extends disk metrics with process-level I/O statistics. Requires root access to /proc/${pid}/io.

Host Metrics

No host-level metrics are collected by this collector.

VM Metrics

Metric	Source	Description	Unit	Cycle
io_read_bytes	/proc/${pid}/io	Bytes read directly from disk	bytes	📊 collect
io_write_bytes	/proc/${pid}/io	Bytes originally dirtied in page cache	bytes	📊 collect

Verbose Mode VM Metrics 📝

Metric	Source	Description	Unit	Cycle
io_rchar	/proc/${pid}/io	Bytes read via any read-like syscall	bytes	📊 collect
io_wchar	/proc/${pid}/io	Bytes written via any write-like syscall	bytes	📊 collect
io_syscr	/proc/${pid}/io	Read-like system calls performed	count	📊 collect
io_syscw	/proc/${pid}/io	Write-like system calls performed	count	📊 collect
io_cancelled_write_bytes	/proc/${pid}/io	Bytes "un-dirtied" (e.g., ftruncate)	bytes	📊 collect

---

Host Collector (--host)

Provides host identification information useful for multi-host deployments.

Host Metrics

Metric	Source	Description	Unit	Cycle
host_name	/proc/sys/kernel/hostname	Hostname of the hypervisor	string	🔄 lookup

Verbose Mode Host Metrics 📝 ⚠️

Metric	Source	Description	Unit	Cycle
host_uuid	/sys/devices/virtual/dmi/id/product_uuid	DMI UUID of the host (requires root)	string	🔄 lookup

VM Metrics

Metric	Source	Description	Unit	Cycle
host_name	/proc/sys/kernel/hostname	Hostname of the hypervisor running this VM	string	🔄 lookup

---

PSI Collector (--pressure)

Monitors Pressure Stall Information (PSI) to detect resource shortages before they cause visible performance degradation. Requires kernel 4.20+.

For more information, see: https://facebookmicrosites.github.io/psi/docs/overview

Host Metrics

Metric	Source	Description	Unit	Cycle
psi_some_cpu_avg60	/proc/pressure/cpu	Time some tasks delayed for CPU (60s window)	%	📊 collect
psi_some_io_avg60	/proc/pressure/io	Time some tasks delayed for I/O (60s window)	%	📊 collect
psi_full_io_avg60	/proc/pressure/io	Time all tasks delayed for I/O (60s window)	%	📊 collect
psi_some_mem_avg60	/proc/pressure/mem	Time some tasks delayed for memory (60s window)	%	📊 collect
psi_full_mem_avg60	/proc/pressure/mem	Time all tasks delayed for memory (60s window)	%	📊 collect

Verbose Mode Host Metrics 📝

Metric	Source	Description	Unit	Cycle
psi_some_cpu_avg10	/proc/pressure/cpu	CPU pressure, some tasks (10s window)	%	📊 collect
psi_some_cpu_avg300	/proc/pressure/cpu	CPU pressure, some tasks (300s window)	%	📊 collect
psi_some_cpu_total	/proc/pressure/cpu	Total CPU delay for some tasks	μs	📊 collect
psi_some_io_avg10	/proc/pressure/io	I/O pressure, some tasks (10s window)	%	📊 collect
psi_some_io_avg300	/proc/pressure/io	I/O pressure, some tasks (300s window)	%	📊 collect
psi_some_io_total	/proc/pressure/io	Total I/O delay for some tasks	μs	📊 collect
psi_full_io_avg10	/proc/pressure/io	I/O pressure, all tasks (10s window)	%	📊 collect
psi_full_io_avg300	/proc/pressure/io	I/O pressure, all tasks (300s window)	%	📊 collect
psi_full_io_total	/proc/pressure/io	Total I/O delay for all tasks	μs	📊 collect
psi_some_mem_avg10	/proc/pressure/mem	Memory pressure, some tasks (10s window)	%	📊 collect
psi_some_mem_avg300	/proc/pressure/mem	Memory pressure, some tasks (300s window)	%	📊 collect
psi_some_mem_total	/proc/pressure/mem	Total memory delay for some tasks	μs	📊 collect
psi_full_mem_avg10	/proc/pressure/mem	Memory pressure, all tasks (10s window)	%	📊 collect
psi_full_mem_avg300	/proc/pressure/mem	Memory pressure, all tasks (300s window)	%	📊 collect
psi_full_mem_total	/proc/pressure/mem	Total memory delay for all tasks	μs	📊 collect

VM Metrics

No VM-level metrics are collected by this collector. PSI data is host-level only.

---

Metric Calculations

Several metrics are derived from raw data using formulas:

CPU Utilization

VM CPU metrics are calculated from scheduler statistics:

cpu_total = Δcpu_times / Δtime × 100
cpu_steal = Δcpu_runqueues / Δtime × 100

Where:

• Δcpu_times: Change in CPU time counters from /proc/${pid}/schedstat
• Δcpu_runqueues: Change in run queue wait time
• Δtime: Time elapsed between measurements

Disk I/O Utilization

Host ioutil:

disk_device_ioutil = (Δtimeforops / Δtime) × 100

VM ioutil (estimated):

disk_ioutil = host_ioutil × (vm_io_requests / total_io_requests)

Disk Queue Metrics

Queue Size:

disk_device_queuesize = Δweightedtimeforops / Δtime

Queue Time:

disk_device_queuetime = Δweightedtimeforops / (Δreads + Δreadsmerged + Δwrites + Δwritesmerged + currentops)

Service Time:

disk_device_servicetime = Δweightedtimeforops / (Δreads + Δreadsmerged + Δwrites + Δwritesmerged)

---

Notes

1. Collection Cycles: Metrics marked with "lookup" are collected during VM discovery (less frequently), while "collect" metrics are gathered every sampling interval.

2. Verbose Fields: In the ncurses UI, verbose-only fields are hidden by default but can be enabled via the field selector (press f). In CLI mode, use --verbose to enable all additional metrics.

3. Human-Readable Mode: Enable with -H command-line flag or press u in the ncurses UI to format byte values as KB/MB/GB/TB. When active, the status bar displays [H]. Changes take effect immediately.

4. Root Requirements: Some metrics (I/O collector, host UUID) require root privileges to access protected proc filesystem entries.

5. Proxmox vs Libvirt: On Proxmox VE, QMP (QEMU Machine Protocol) is used instead of libvirt API for faster metric collection.

6. Rate Metrics: Most counter-based metrics (bytes, packets, etc.) should be interpreted as rates when comparing across time intervals.

7. esxtop Equivalent Metrics: proxtop provides many esxtop-equivalent metrics for VMware-to-KVM/Proxmox migrations:

   • %UTIL = Device busy percentage (like esxtop GAVG/DAVG context)
   • QDEPTH = Current queue depth (like esxtop ACTV)
   • QLEN = Average queue length (like esxtop QUED)
   • SVCTM = Service time (like esxtop DAVG)
   • AWAIT = Wait time (like esxtop KAVG)
   • LAT/rd, LAT/wr, LAT/fl = Per-operation type latencies

8. Sorting: Use </> to change sort column, r to toggle ascending/descending. The sorted column shows ^ (ascending) or v (descending). Numeric columns are sorted numerically.

9. Physical Device Views: Press p for physical network interfaces or s for physical disk devices. These views use the same field selector as VM views - press f to show/hide columns.

10. Physical Network Interface Metrics (p key in ncurses): Per-interface statistics from /proc/net/dev with calculated rates:

    • net_MbRX/s = Megabits/sec received
    • net_MbTX/s = Megabits/sec transmitted
    • net_PKTRX/s = Packets/sec received
    • net_PKTTX/s = Packets/sec transmitted
    • Raw counters (RX-Bytes, TX-Bytes, etc.) are also available via field selector

11. Physical Disk Device Metrics (s key in ncurses): Per-device statistics from /proc/diskstats with calculated rates:

    • dsk_READS/s = Read operations per second
    • dsk_WRITES/s = Write operations per second
    • dsk_MBRD/s = MB/sec read throughput
    • dsk_MBWR/s = MB/sec write throughput
    • dsk_%UTIL = Device busy percentage (0-100%)
    • dsk_QDEPTH = Current queue depth (I/Os in flight)
    • dsk_SVCTM = Average service time per I/O (ms)
    • dsk_AWAIT = Average wait time per I/O including queue time (ms)

12. Host Storage Device Views: Three separate views for different storage types, each with the same metrics as above:

    • Physical Disks (s key): SCSI/SATA (sd*), NVMe (nvme*), virtio (vd*), partitions (sda1, etc.)
    • LVM Volumes (l key): LVM logical volumes (detected via LVM- UUID prefix in /sys/block/dm-*/dm/uuid)
    • Multipath Devices (x key): DM-Multipath devices (detected via mpath- UUID prefix in /sys/block/dm-*/dm/uuid)

    Device mapper devices (dm-*) are displayed with their friendly names from /dev/mapper/ instead of the raw dm-X name.

    Excluded from all views: loop* (loopback), ram* (RAM disks)