NetXMS Support Forum

V6.1.2

Dear Support Team,
I'm writing to seek technical support for two issues with the network topology functions.
When I right-click a node, select Topology maps, and choose Layer 2 topology or other topology types, the displayed topology coverage is far too large. It appears that the discovery scope is not restricted by the parameter Topology.DefaultDiscoveryRadius, which is currently set to 5.
Regarding the Network Map feature: when using MSTP links with ISP communication devices deployed between two end nodes, the system sometimes fails to recognize the connection relationship between the two terminal nodes. We presume this issue results from improper configuration of the intermediate ISP-side devices.
We sincerely hope you can offer relevant solutions and troubleshooting suggestions. Thank you very much for your help!
Best regards！

Hello NetXMS Development Team,
 
I have a question about the AI tool invocation format of NetXMS.
 
I heard that NetXMS currently only accepts AI tool calls developed based on standard JSON format, and it cannot work properly with models that adopt XML format for tool calling.
 
The Qwen3.6 series large models use XML format to implement tool invocation. I would like to confirm whether NetXMS is indeed incompatible with XML tool calling mode right now.
 
Besides, I sincerely hope that you can add support for XML format AI tool calls in subsequent version updates. This optimization will bring great convenience for users applying Qwen3.6 and other XML-based AI models.
 
Thank you very much for your time and support!

That's wonderful! I'm looking forward to the next version update, and thank you so much for your tremendous help!

I tested the command nxsnmpwalk -v 2c -c stonewell2000 10.61.43.131 1.3.6.1.2.1.17.7.1.2.2.1 separately on version 5.0.8 and 6.1.1, and there are obvious discrepancies between the two results. Only version 5.0.8 can retrieve the complete data normally.The packet capture logs of v6.1.1 have been uploaded as an attachment. Thank you so much!

Quote from: Victor Kirhenshtein on May 04, 2026, 01:53:50 PMIs there really only one row in SNMP walk result? If yes, try to do separate walks on 1.3.6.1.2.1.17.7.1.2.2.1.3 and 1.3.6.1.2.1.17.7.1.2.2.1.2 - there should be more entries. Also try walk on 1.3.6.1.2.1.17.4.3.1.3

This issue does not occur in version 5.X. Could you please assist us with the diagnosis and troubleshooting?

Quote from: Victor Kirhenshtein on May 04, 2026, 01:58:42 PMRegarding AI invocation - I've tried to check qwen3.6-35b-a3b model via OpenRouter, but looks like they don't have providers with tool calling support. From your screenshot it seems that tool call is not recognized and interpreted as part of an answer. Are you sure that runner or service you are using actually follows Open AI API?

Thanks for your guidance. I'll check with my colleague in charge of AI large models.

The SNMP walk result for 1.3.6.1.2.1.17.7.1.2.2.1 indeed returns only one entry, which is rather unusual. Many H3C switches show exactly the same result, whereas this used to work normally before.The SNMP traversal results for other OIDs are listed below. Thanks for your help.

Thank you very much for your reply.

I have two questions to ask:
A previous issue (https://www.netxms.org/forum/general-support/help-h3c-port-security-policy-causes-the-switch-to-fail-to-learn-the-interface/msg32436/#msg32436) seems to have reappeared in the latest version. When a port security policy is configured on a switch interface, the MAC address status on the switch shows as secure instead of learned, and NetXMS appears to ignore MAC addresses in the secure state.
The AI model cannot execute tool calls, as shown in the screenshot. My configuration in netxmsd.conf is as follows:
plaintext
Module=aitools

[AI/Provider/qwen3]
Type = openai
URL = http://IP/v1/chat/completions
Model = qwen3.6-35b-a3b
Token = xxx
ContextSize = 32000
Timeout = 900
Slots = default

[AI]
DefaultProvider = qwen3
Looking forward to your reply. Thank you very much!

Wow, that's awesome, thank you so much!

Thank you very much for your help. I would like to ask if NetXMS supports the Tongyi Qianwen model that complies with the OpenAI specification. The Qwen3-30B model received the chat content sent via AITools, but it finally prompted: Error: Cannot get response from AI assistant.

The content in the netxmsd log is as follows:
2026.02.11 17:26:37.135 *D* [ai.assistant       ] Using provider "qwen3" for slot "interactive"
2026.02.11 17:26:37.135 *D* [llm.chat           ] Added message to chat: role="user", content="server uptime"
2026.02.11 17:26:37.136 *D* [client.session.0   ] Sending compressed message CMD_OBJECT_UPDATE (1920 bytes)
2026.02.11 17:26:37.137 *D* [ai.provider        ] Sending request to http://10.61.7.101:8008/v1/chat/completions: {"model": "qwen3-30b-a3b-instruct-2507", "stream": false, "messages": [{"role": "system", "content": "You are

We are experiencing the same issue. Please provide guidance. Thank you very much.

## Logging
# Log file name
LogHistorySize=7
LogRotationMode=1
LogFile=/var/log/netxmsd

# Increase logging verbosity, 0 (only errors) to 9 (verbose debug)
#DebugLevel=3

## Option #2 - PostgreSQL (recommended):
DBDriver=pgsql.ddr
DBServer=127.0.0.1
DBName=netxms_db
DBLogin=netxms
DBPassword=xm123456


[AI/Provider/openai]
Type = openai
URL = http://10.61.7.101:8008/v1/chat/completions
Model = qwen3-30b-a3b-instruct-2507
Token = axxxxxxxxxxxxx
Slots = default

OK。。 thank you very much!

netxms client versions 4.0.6 and 4.0.7 do not run under 64 bit window7, no interface appears, the log in the .nxmc4 directory is empty, and versions prior to 4.0.5 work, what could be the reason for this, thank you very much!