routeros-diagnostics-monitoring
Diagnostics and monitoring on MikroTik RouterOS 7. Covers the system log (topics, logging rules and actions, debug buffers, disk, syslog over UDP, TCP and TLS, CEF, email and script actions), ping and path MTU discovery, traceroute, torch, the packet sniffer with TZSP streaming and pcapng files, interface counters and monitor-traffic, Netwatch probes and scripts, SNMP agent, communities, SNMPv3 and traps, Traffic Flow (NetFlow and IPFIX), graphing, resource and CPU views, profiler, health sensors, watchdog, Detect Internet, bandwidth test and speed test, IP scan, and the traffic generator. Use when a link, host or service misbehaves and you need to find where, when collecting logs or metrics from a router, setting up alerts, or measuring throughput.
How do I install this agent skill?
npx skills add https://github.com/wiresage/routeros-skills --skill routeros-diagnostics-monitoringIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill is a comprehensive documentation set for MikroTik RouterOS diagnostics and monitoring. It is generally safe but presents a minor vulnerability surface for indirect prompt injection when processing untrusted network logs or packet captures.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
RouterOS diagnostics and monitoring
Start with the cheapest tool that can answer the question. Most tools below change router behavior while they run (sniffer and torch turn off fast path and FastTrack, bandwidth test saturates the link), so say so before running them on a production router and stop them afterwards.
| Question | Tool |
|---|---|
| Is the host or the internet reachable? | /ping (add src-address, interface, vrf, count) |
| Where does the path break, loop or slow down? | /tool/traceroute (use use-dns=yes for names) |
| What happened and when? | /log/print, /system/logging |
| Who uses the bandwidth right now? | /tool/torch <interface> |
| What exactly is on the wire? | /tool/sniffer (save to pcapng for Wireshark) |
| Is a host or service down (and act on it)? | /tool/netwatch |
| Long-term metrics | SNMP, /ip/traffic-flow, /tool/graphing, syslog |
| How fast is the link? | /tool/speed-test, /tool/bandwidth-test |
| CPU, memory, temperature, voltage | /system/resource, /tool/profile, /system/health |
| Router hangs or lost uplink must reboot it | /system/watchdog |
Ping and traceroute
/ping 1.1.1.1 count=3 # internet; then a name to test DNS
/ping 198.51.100.10 size=1500 do-not-fragment count=2 # path MTU: lower size until replies arrive
/ping 192.168.2.10 src-address=192.168.88.1 count=3 # needed for IPsec or policy-routed tests
/tool/traceroute 198.51.100.10 count=3 use-dns=yes
sizecounts the whole IP packet (1500 equalsping -l 1472on Windows). Withoutdo-not-fragmenta large ping is fragmented and hides an MTU problem. A firewall that drops "fragmentation needed" messages turns the symptom into plain timeouts. Fix with an MSS clamp (routeros-firewall).- Status words:
timeout(host down, ICMP filtered or reply slower thaninterval),net unreachableandhost unreachable(the router inHOSThas no route or rejects),admin prohibited(reject-with rule),TTL exceeded(usually a routing loop). interface=is needed for ARP ping (hosts that block ICMP), ND ping and link-local IPv6.vrf=accepts only real VRFs.- Traceroute is
mtr-like: ICMP probes, up to 30 hops, statistics per hop. Loss at one hop that does not continue at later hops is ICMP rate limiting, not path loss. The same router repeating means a loop; a silent tail after one hop means a missing route or a missing route back.
Log
Rules (/system/logging) select by topics and send to an action (/system/logging/action): memory (1000 lines, lost at reboot), disk,
echo, remote (syslog), email, script. Defaults keep info, error, warning in memory and print critical. Debug is off.
/log/print where topics~"interface|dhcp"
/log/print follow-only where message~"login failure"
/system/logging/action/add name=dhcpdebug target=memory memory-lines=2000 # debug into its own buffer
/system/logging/add topics=dhcp,debug action=dhcpdebug
/log/print where buffer=dhcpdebug
/system/logging/action/set remote remote=192.0.2.10 remote-log-format=syslog
/system/logging/add topics=info action=remote
- A rule matches entries that have all its topics:
topics=info,warningmatches nothing. Use one rule per topic.!excludes a topic. - Debug logs can contain secrets (
ssh,debuglogs session keys). Keep them in a local buffer, never send them to syslog or email. - Syslog formats:
default,syslog(RFC 3164 only),cef. Over TCP and TLS usecef(the others have no message delimiter). TLScheck-certificate=yesverifies the CA but not the server name. Setsyslog-time-format=iso8601across time zones and keep NTP right (routeros-system-utilities). Action names accept only letters and digits. - Email actions send one mail per entry; a brute-force attack floods the mailbox, so stop attacks in the firewall. A script action reads
$topicsand$message; the manual's example collects failed-login sources into an address list. - Entries show twice in
/log/printwhen several actions store them: addwhere buffer=memory.
Sniffer and torch
/tool/sniffer/start interface=bridge ip-address=192.168.88.50/32 # capture on the LAN side, before NAT hides the host
/tool/sniffer/stop
/tool/sniffer/save file-name=capture.pcap # pcapng, open in Wireshark
/tool/sniffer/quick interface=ether1 port=bootps,bootpc duration=10s
/tool/torch bridge src-address=192.168.88.21/32
- The sniffer sees received packets before the firewall and sent packets after it, so drops show as received only. Hardware-offloaded bridge
traffic never reaches the CPU, so it is invisible to both tools (
routeros-bridging-switching). Filters combine with AND across different filters. Memory is 100 KiB by default;memory-scroll=yeskeeps the newest packets. Captures contain user traffic: treat files as private. - TZSP streaming (
streaming-enabled=yes streaming-server=... filter-stream=yes) sends frames unencrypted over UDP 37008. - Torch:
src-addressis the far side of the watched interface, so on a LAN interface filter hosts withsrc-address.RXis what the interface receives from that address. It shows traffic the firewall drops (input side), which helps decide whether a rule or the path is at fault.
Netwatch
/tool/netwatch/add name=wan-dns host=1.1.1.1 type=icmp thr-avg=100ms interval=10s \
up-script=":log info \"wan up\"" down-script=":log warning \"wan down\""
/tool/netwatch/print stats
Types: simple (default interval 60 s), icmp, tcp-conn, http-get, https-get, dns. Scripts run as a restricted internal user with
only read, write, test, reboot policies and have private globals. ICMP probes have default thresholds (for example thr-avg=100ms), so
a slow link reports down. http-get is down on a redirect unless http-codes=100-399. After a reboot probes wait startup-delay (5 m).
SNMP
/snmp/set enabled=yes contact="NOC" location="Server room"
/snmp/community/add name=monit addresses=192.0.2.0/24 security=none read-access=yes
/snmp/community/set [find name=public] addresses=192.0.2.0/24 # or remove it
/snmp/set trap-target=192.0.2.50 trap-community=monit trap-version=2 trap-generators=interfaces
The default public community is readable from any address. security=authorized|private means SNMPv3 (v1/v2c polls get no reply), with
passwords of at least 8 characters. A request from outside addresses gets no answer. write-access=yes allows changing the identity,
running scripts and rebooting: only on v3 or a narrow subnet. print oid at any menu lists OIDs. Test with /tool/snmp-get.
Flow export, graphs, resources
/ip/traffic-flow/set enabled=yes
/ip/traffic-flow/target/add dst-address=192.168.0.2 port=2055 version=9
/ip/traffic-flow/monitor duration=3
/tool/graphing/interface/add interface=ether1 allow-address=192.168.88.10/32
/tool/graphing/resource/add allow-address=192.168.88.10/32
/system/resource/print
/tool/profile cpu=all freeze-frame-interval=1 duration=5
- Traffic Flow exports NetFlow v1, v5, v9 or IPFIX over UDP; hardware-offloaded traffic is not counted. Graphs live at
http://router/graphs/with no login: always setallow-address(default is every IPv4 address) and keep the page off WAN. - Interface counters:
/interface/print stats,monitor-traffic, and/tool/traffic-monitorto run a script when a rate crosses a threshold. Rising regular counters with flatfp-*counters mean traffic leaves the fast path. - Health (
/system/health) shows temperature, voltage, fans; threshold traps needtrap-generators=temp-exception.
Throughput tests
/tool/speed-test address=<peer> test-duration=5s (latency, jitter, TCP and UDP both ways) and /tool/bandwidth-test need a RouterOS peer
with /tool/bandwidth-server on, a user in a group with the test and winbox policies, device mode bandwidth-test enabled, TCP 2000 and
UDP 2000-65535 allowed, and for UDP receive the client must accept the server's packets. Limit local-tx-speed and remote-tx-speed on
live links. Remove the test user and firewall rules afterwards. flood-ping and the traffic generator need the traffic-gen device-mode flag.
Watchdog and Detect Internet
/system/watchdog reboots a hung router (watchdog-timer=yes) and, when watch-address is set, a router whose uplink address stops
answering six pings. Pick an address inside the tunnel or on the far side of the link, set ping-start-after-boot longer than the link needs,
and keep another way in: a bad address means a reboot every few minutes. /interface/detect-internet fills dynamic lists (internet, wan,
lan) from route and MikroTik cloud checks; the lists change by themselves, so do not use them in accept rules.
Where to look
| Need | Open |
|---|---|
| Log topics list, TLS syslog, disk files, script actions | references/guides/diagnostics-monitoring-and-troubleshooting/log.md |
| Ping, traceroute, torch, sniffer | ping.md, traceroute.md, torch.md, packet-sniffer.md in the same folder |
| Netwatch, SNMP, traffic flow, graphing | netwatch.md, snmp.md, traffic-flow.md, graphing.md |
| Bandwidth test, speed test, traffic generator | bandwidth-test.md, speed-test.md, traffic-generator.md |
| Resource, profiler, health, watchdog, detect internet | resource.md, profiler.md, health.md, watchdog.md, detect-internet.md |
| Properties | references/cli/tool/, references/cli/system/logging/, references/cli/snmp/, references/cli/ip/traffic-flow/ |
| Everything else | references/INDEX.md |
Related skills: routeros-troubleshooting (supout.rif and support), routeros-system-utilities (clock, scheduler, email), routeros-firewall
(counters, logging rules), routeros-scripting, routeros-hardware (sensors and PoE), routeros-queues (graphing queues).
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/wiresage/routeros-skills/routeros-diagnostics-monitoring">View routeros-diagnostics-monitoring on skillZs</a>