Orion-LD crashes when using pick parameter on query -- container stops unexpectedly

[dncampo]

I encountered an unexpected crash when using the pick parameter in an entity query.

Environment:

• Orion 1.10.0
• Mongo 6.0
• Dataset used: Smart Farm located here

When using a FIWARE MCP Server to access data in the CB, a query is performed with a pick parameter:

{
  `params`: `GET /ngsi-ld/v1/entities?type=Animal&q=ownedBy==\"urn:ngsi-ld:Person:person001\";sex==\"Female\";species==\"dairy cattle\"&pick=name,weight,sex,species`
}

It returns:
{"status": 200, "headers": {"Date": "Wed, 29 Oct 2025 09:27:47 GMT", "Content-Type": "application/json", "Content-Length": "2"}, "body": []}
But after doing some pick again:

{
  params: GET /ngsi-ld/v1/entities?type=Animal&q=ownedBy==\"urn:ngsi-ld:Person:person001\"&pick=name,weight,sex,species
}

It crashes:

Response
{"error": "('Connection aborted.', RemoteDisconnected('Remote end closed connection without response'))", "url": "http://localhost:1026/ngsi-ld/v1/entities?type=Animal&q=ownedBy==\"urn:ngsi-ld:Person:person001\"&pick=name,weight,sex,species"}

Killing the container. You can see the "conversation" with the LLM here.

[cfreyfh]

Hi there,

I tried to reproduce the error, but the queries worked for me without crashing the broker. Do you have the log messages from the container with the broker?

Also, when you run the query

{
  `params`: `GET /ngsi-ld/v1/entities?type=Animal&q=ownedBy==\"urn:ngsi-ld:Person:person001\";sex==\"Female\";species==\"dairy cattle\"&pick=name,weight,sex,species`
}

Why are you getting an empty response? There should be at least one animal matching the request (based on my attempts):

[...]
{
        "name": {
            "type": "Property",
            "value": "Trotter"
        },
        "sex": {
            "type": "Property",
            "value": "Female"
        },
        "species": {
            "type": "Property",
            "value": "dairy cattle"
        },
        "weight": {
            "type": "Property",
            "value": 373
        }
    },
    {
        "name": {
            "type": "Property",
            "value": "Petal"
        },
        "sex": {
            "type": "Property",
            "value": "Female"
        },
        "species": {
            "type": "Property",
            "value": "dairy cattle"
        },
        "weight": {
            "type": "Property",
            "value": 404
        }
    }
[...]

Greetings,
Carsten

[kzangeli]

Might be a problem with resources then.
David, do you have a "docker exit code" from the container crash?

[alvaro-max-git]

I'm having the same issue on the same environment (MCP Server and Smart Farm). I ran the requests with "pick" directly through Postman and got "Error:socket hang up".

The container status shows "Exited (139)" after crashing.

The request was /ngsi-ld/v1/entities?type=Animal&q=locatedAt==%22urn:ngsi-ld:AgriParcel:003%22&pick=name,id

[kzangeli]

139 ...

Google says:
Error code 139 in Docker typically indicates that a container has exited due to a segmentation fault (SIGSEGV), which usually occurs when the program tries to access an invalid memory location. This can be caused by various issues, including insufficient memory allocation for the container or bugs in the application running inside it.

So, a crash as it seems.
We will have to try to reproduce it, Carsten and myself.
Often, the most difficult part of solving an issue is to be able to reproduce it ...

We still can't rule out what I first suggested, an out-of-memory issue.

So, try giving your "broker containers" some more RAM and let us know

[cfreyfh]

@alvaro-max-git

I'm having the same issue on the same environment (MCP Server and Smart Farm). I ran the requests with "pick" directly through Postman and got "Error:socket hang up".

The container status shows "Exited (139)" after crashing.

The request was /ngsi-ld/v1/entities?type=Animal&q=locatedAt==%22urn:ngsi-ld:AgriParcel:003%22&pick=name,id

I tried the same and get a list of the animals from the context broker. Did you do make some other requests before sending this reqest or only loading the data to the CB?

[kzangeli]

@dncampo , @alvaro-max-git , try with more RAM and if it still fails, let's set up a call and see it together.

[alvaro-max-git]

@cfreyfh I usually run one "health status" query, but it crashes if it's the first request too.

I tried to edit .wslconfig to allocate more memory to WSL (8GB + 8GB swap with 16GB installed RAM), but it crashed again.

I monitored RAM usage with docker stats and it was using less than 10MB before the crash as you can see in the screenshot, so it doesn't seem a lack of memory.

I ran dmesg to check kernel logs and they show a segfault:

[  100.189401] MHD-connection[3976]: segfault at 0 ip 000000000053579d sp 00007be969feb060 error 4 in orionld[400000+625000] likely on CPU 3 (core 1, socket 0)

The container logs show that the last messages were the pickForEntity instructions:

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=orionldGetEntitiesLocal.cpp[270]:orionldGetEntitiesLocal | msg=KjNode Tree too large to be rendered

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[76]:pickForEntityArray | msg=2 items in pick: 13091776

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[80]:pickForEntityArray | msg=  o name

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[80]:pickForEntityArray | msg=  o id

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[83]:pickForEntityArray | msg=First level records per entity:

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[47]:pickForEntity | msg=  o urn:ngsi-ld:Animal:cow002

time=Tuesday 18 Nov 20:33:17 2025.157Z | lvl=DEBUG | corr=N/A | trans=N/A | from=N/A | srv=N/A | subsrv=N/A | comp=Orion | op=pick.cpp[62]:pickForEntity | msg=    - id (stays)

Environment:

• Windows 11
• WSL2 with vsyscall=emulate enabled in .wslconfig
• Docker Desktop 4.46

Tell me if you need anything else. @dncampo may give better feedback.

[dncampo]

I confirm that the error occurred on WSL within a Windows host OS. I was also able to reproduce it on Ubuntu 22.04.05 LTS.

The error code in the Docker container is indeed 139.