NetXMS Support Forum

Hi!

This is queue for processing discovered addresses. If you have discovery filters some (or most) of them could be filtered out and then re-discovered again and again. This can create permanently non-empty discovery queue, but this is fine.

Best regards,
Victor

Dashboards created in old versions (likely before 4.x) may render as empty in GUI client version 4.1.377. We have fixed it in version 4.1.380, so if you experienced that issue, upgrade should have it fixed. No changes to dashboards are required.

Best regards,
Victor

Hi all!

New patch release for version 4.1 - 4.1.377 is just published. Changes since previous patch release:

- Agent configuration option to disable local database
- Linux subagent can read software inventory from OpenWrt package manager
- Added driver for HPE iLO
- Fixed incorrect return code of NXSL function "SNMPSet" and SNMP transport method "set"
- Original DCI string value returned by transformation script is preserved
- Fixed bug in downtime calculation for hierarchical business services
- Fixed issues:
        NX-2262 (syncUserDatabase() does not wait for CMD_USER_DB_EOF)

Best regards,
Victor

Quote from: Kriptoker on June 07, 2022, 06:13:44 PM
As far as disabling the ssh.nsm local agent, that would be by removing that line 'SubAgent = ssh.nsm' from /etc/nxagentd.conf correct?

Correct.

Try to run netxmsd with elevated debug by adding
DebugLevel = 6
to /etc/netxmsd.conf and share log file for startup.

Best regards,
Victor

Make sure you put password in single quotes. Also, user should have appropriate access rights ("access server console" in global system access rights).

Best regards,
Victor

There is a bug that NetXMS server attempts to access SSH even with empty credentials. It will be fixed in upcoming patch release.

Also, if you are not using SSH, make sure that ssh.nsm is not loaded on server's local agent.

Another option is to use NXSL configuration poll hook to disable SSH on each node by calling

$node->enableSsh(false);


Best regards,
Victor

Actually 6733 (first part of primary key) should be ID of the node this package belongs to.

Best regards,
Victor

Does it report VLANs correctly with GENERIC driver? If not, does forcing CISCO-SB helps?

VLAN information is most common place where custom driver is required.

Best regards,
Victor

Yes, they are universal. Below is ad-hoc example of script that reads DHCP leases from Mikrotik router and prints IP address of specific node, if found:


for(line : $node->executeSSHCommand("/ip dhcp-server/lease/print"))
{
if (line->contains("raspberrypi") and line ~= '^\s?[0-9]+\s+([0-9.]+).*')
   println("address:", $1);
}


Best regards,
Victor

Hi,

Yes, I think we will use SSH more. Some automation with SSH is already possible - like execute command as a reaction for an event for example, or execute command from script and process output. One obvious direction for improvement is to introduce concept of SSH session to NXSL scripts and allow execution of multiple commands within single session.

Best regards,
Victor

I just checked and there is a bug in SSH polling code - server attempts to connect with empty user name and password if nothing is set. We will fix it until next patch release.

For disabling SSH polling you can use NXSL method enableSsh on node object - either add it to configuration poll hook or run once script that will enumerate all nodes and set SSH flag to false. Also, if you are not using SSH functionality, you can disable SSH subagent on server's local agent and on proxy nodes.

Best regards,
Victor

Hi,

looks like nobody checked that client proxy actually compiles. Do you really use it? If not, just remove --with-client-proxy from configure (we will fix compilation anyway though).

Best regards,
Victor

Hi,

version 4.1 attempts to detect SSH availability during configuration poll. Most likely other switches just don't report failed SSH login attempts, or they have valid credentials. If you don't need SSH access to problematic switches you can disable SSH polling in node properties on "Polling" page.

Best regards,
Victor

Hi,

it's possible to some degree. Few options here:

1. Create circular status map - you can define root object, and status map will show sub-objects and their statuses as sunburst diagram. However, drilldown action is hardcoded and will open list of active alarms for selected object.

2. Create network map - you can place any object on a map, and as drilldown you can set other map or dashboard. This will not be sunburst diagram however - although you can represent objects in some logical layout.

Best regards,
Victor