Hi... I just updated my nabox to the latest version and tried to add an API key for my Claude LLM... but I have yet to respond to anything... testing the connection in Nabox works fine... but from there I am only getting "Unknown error"... it seems to "think" for awhile... but then just error... I know this is beta, but I expected a bit more 😉
#MCP in Nabox?
crystal crow
chilly arch
@magic isle
magic isle
So your LLM is connected successfully and once saved, the Test button validates ?
crystal crow
Correct 🙂
magic isle
can you check what containers are running ? dc ps ?
crystal crow
most of them I guess?
nabox4 ~ # dc ps NAME IMAGE COMMAND SERVICE CREATED STATUS PORTS alertmanager alertmanager "/bin/alertmanager -…" alertmanager 3 hours ago Up 3 hours 9093/tcp caddy caddy "caddy run --config …" rproxy 3 hours ago Up 3 hours 0.0.0.0:80->80/tcp, :::80->80/tcp, 0.0.0.0:443->443/tcp, :::443->443/tcp, 443/udp, 2019/tcp cadvisor cadvisor "/usr/bin/cadvisor -…" cadvisor 3 hours ago Up 3 hours (healthy) 8080/tcp grafana grafana-oss "/run.sh" grafana 5 minutes ago Up 5 minutes 127.0.0.1:3000->3000/tcp grafana-mcp mcp-grafana "/app/mcp-grafana --…" grafana-mcp 5 minutes ago Up 5 minutes 8000/tcp, 127.0.0.1:8086->8082/tcp harvest-mcp harvest-mcp "./bin/harvest-mcp s…" harvest-mcp 3 hours ago Up 3 hours 127.0.0.1:8084->8082/tcp havrest havrest:latest "/havrest" havrest 3 hours ago Up 3 hours 8080/tcp node-exporter node-exporter "/bin/node_exporter …" node-exporter 3 hours ago Up 3 hours 9100/tcp ontap-mcp ontap-mcp "./ontap-mcp start -…" ontap-mcp 3 hours ago Up 3 hours 127.0.0.1:8085->8082/tcp victoria-metrics victoria-metrics "/victoria-metrics-p…" victoria-metrics 3 hours ago Up 3 hours (healthy) 127.0.0.1:8428->8428/tcp vmagent vmagent "/vmagent-prod -remo…" vmagent 3 hours ago Up 3 hours 8429/tcp vmalert vmalert "/vmalert-prod -rule…" vmalert 3 hours ago Up 3 hours 8880/tcp
magic isle
What about journalctl -fu naboxd when interacting?
crystal crow
A lot of gets of local stuff which I guess is not related? Then these...
Apr 08 15:09:17 nabox4 naboxd[17105]: time=2026-04-08T15:09:17.180Z level=INFO source=agent.go:261 msg="llm first token" iteration=1 ttft=4.601s
Apr 08 15:09:17 nabox4 naboxd[17105]: time=2026-04-08T15:09:17.180Z level=INFO source=agent.go:274 msg="llm requested tool" iteration=1 tool=get_interest args="{"id": "morning-coffee"}" elapsed=4.601s
Apr 08 15:09:17 nabox4 naboxd[17105]: time=2026-04-08T15:09:17.185Z level=INFO source=agent.go:286 msg="llm stream complete" iteration=1 duration=4.606s tool_calls=1
Apr 08 15:09:17 nabox4 naboxd[17105]: time=2026-04-08T15:09:17.185Z level=INFO source=agent.go:415 msg="tool completed" tool=get_interest type=internal duration=0s
Apr 08 15:09:17 nabox4 naboxd[17105]: time=2026-04-08T15:09:17.186Z level=INFO source=agent.go:526 msg="tool round complete" iteration=1 tools_called=1 tools_duration=0s iteration_duration=4.607s
crystal crow
the last one includes some errors..
magic isle
and reload the page
crystal crow
Apr 08 15:10:11 nabox4 naboxd[17105]: time=2026-04-08T15:10:11.094Z level=ERROR source=agent.go:310 msg="llm stream error" error="anthropic stream: json: error calling MarshalJSON for type anthropic.MessageParam: json: error calling MarshalJSON for type anthropic.ContentBlockParamUnion: json: error calling MarshalJSON for type *anthropic.ToolUseBlockParam: json: error calling MarshalJSON for type json.RawMessage: unexpected end of JSON input" iteration=2
Yep.. the first 5 lines again... then it waits for awhile.... then the error is logged... no other errors apparently... only some GET calls to the nabox...
...that was after the restart naboxd...
I have one 4 node ONTAP cluster, a 3 node StorageGRID and two Cisco Nexus switches defined in this setup...
magic isle
Can you try and carch the startup of naboxd with the messages connecting to mcp servers ?
crystal crow
so a restart and the journalctl after? or at once? 🙂
crystal crow
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.236Z level=INFO source=main.go:64 msg="Starting NAbox 4" version=v4.2.0-0-gf2056a6
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.257Z level=INFO source=main.go:150 msg="no upgrade in progress"
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.264Z level=INFO source=main.go:291 msg="starting http server" bind=:8080
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.272Z level=INFO source=router.go:103 msg="mcp connected" server=harvest-mcp endpoint=http://localhost:8084/ tools=17
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.273Z level=INFO source=chatbot.go:119 msg="chatbot: connected to MCP server" server=harvest-mcp endpoint=http://localhost:8084/
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.283Z level=INFO source=router.go:103 msg="mcp connected" server=ontap-mcp endpoint=http://localhost:8085/ tools=27
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.284Z level=INFO source=chatbot.go:119 msg="chatbot: connected to MCP server" server=ontap-mcp endpoint=http://localhost:8085/
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.317Z level=INFO source=router.go:103 msg="mcp connected" server=grafana-mcp endpoint=http://localhost:8086/mcp tools=42
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.318Z level=INFO source=chatbot.go:119 msg="chatbot: connected to MCP server" server=grafana-mcp endpoint=http://localhost:8086/mcp
Apr 08 15:20:11 nabox4 naboxd[20815]: time=2026-04-08T15:20:11.318Z level=INFO source=chatbot.go:186 msg="chatbot: initialised" provider=anthropic model=claude-sonnet-4-6 capabilities=3
crystal crow
ok so it's not just me? 😉