Skip to main content
Version: 7.26

Resource

The /system/resource menu shows the overall resource usage and identity of the router: uptime, version, memory, disk, CPU model, architecture and the bad-block percentage of the NAND. All values are read-only.

[admin@MikroTik] > /system/resource/print
uptime: 2h31m3s
version: 7.26beta1
build-time: 2026-10-01 13:29:47
minimum-version: 7.0.9
free-memory: 678.7MiB
total-memory: 1024.0MiB
cpu: ARM64
cpu-count: 4
cpu-frequency: 864MHz
cpu-load: 0%
free-hdd-space: 91.9MiB
total-hdd-space: 128.0MiB
write-sect-since-reboot: 437
write-sect-total: 263060
bad-blocks: 0%
architecture-name: arm64
board-name: hAP ax^2
platform: MikroTik

The properties (uptime, version, build-time, memory, hdd space, write-sector and bad-block counters, board and platform names) are described in the /system/resource CLI reference page.

Watch usage live​

/system/resource/monitor prints a live-updating view of CPU usage per core and the free memory (the table refreshes continuously; bound it with duration):

[admin@MikroTik] > /system/resource/monitor duration=2
cpu-used: 0%
cpu-used-per-core: 1%
0%
0%
0%
free-memory: 694948KiB

free-memory here is in KiB (694948 KiB ≈ 678.7 MiB shown by print).

Read per-core breakdown​

/system/resource/cpu shows usage per core, split into user load (load), interrupt handling (irq) and disk I/O (disk):

[admin@MikroTik] > /system/resource/cpu/print
Columns: CPU, LOAD, IRQ, DISK
# CPU LOAD IRQ DISK
0 cpu0 0% 0% 0%
1 cpu1 0% 0% 0%
2 cpu2 1% 0% 0%
3 cpu3 0% 0% 0%

Interrupts (IRQ)​

The /system/resource/irq menu shows all used IRQs on the router, including the CPU the IRQ is assigned to and the interrupt counter rows for each device:

[admin@MikroTik] > /system/resource/irq/print where count>0
Flags: o - READ-ONLY
Columns: IRQ, USERS, CPU, ACTIVE-CPU, COUNT
# IRQ USERS CPU ACTIVE-CPU COUNT
0 6 bam_dma auto 0 68
1 8 78b5000.spi auto 2 157
2 9 glink-native auto 3 116

IRQ assignment with cpu=auto is done by the interrupt routing (which uses NAPI interrupt consolidation). The o flag marks hardware-fixed IRQs, whose CPU assignment cannot be changed.

To pin an IRQ to a specific core:

/system/resource/irq/set [find irq=6] cpu=2

Receive packet steering​

/system/resource/irq/rps is Receive Packet Steering (RPS), similar to Receive Side Scaling (RSS), but implemented in software: it directs packet processing work of an interface to a chosen CPU. RPS is useful when packets require additional processing that uses a relatively large amount of CPU time, such as PPP tunnel termination, VPLS, or firewall processing: the cost of RPS steering is outweighed by spreading the heavy work over the cores.

RPS does not always pay off. If a network device has multiple hardware queues, its receive side scaling (RSS) already maps queues to CPUs, and RPS steering costs more than it saves. Configure RPS when the hardware has fewer queues than CPUs, or when RSS does not cover the path (for example packets decapsulated off a tunnel interface).

Hardware map​

/system/resource/hardware lists devices attached over the PCI, USB and SCSI buses with their bus location, vendor/device identifiers, firmware-reported name, driver owner, negotiated speed, USB version and so on. A board with no enumeration-capable bus (for example the hAP ax²) prints an empty list.

USB device authorization​

When USB device authorization is turned on globally (authorization=yes in /system/resource/hardware/usb-settings), newly attached USB devices stay disabled until the administrator allows them in /system/resource/hardware/authorize.

Reset USB devices​

To reset a misbehaving USB modem or drive without rebooting the router, power-cycle its USB port with /system/resource/hardware/usb-power-reset (the bus and slot come from the device's /system/resource/hardware row):

/system/resource/hardware/usb-power-reset bus=1 slot=2 duration=5s

Technical details​

  • cpu-load sums the usage of all cores of the CPU; use /system/resource/cpu or monitor for the per-core breakdown.
  • write-sect-since-reboot/write-sect-total and bad-blocks help spot flash wear on NAND devices (together with the Health readings and .65-style early warnings).

For all parameters and their defaults, see the /system/resource, cpu, irq, irq/rps, hardware, usb-settings, authorize and usb-power-reset CLI reference pages.