SYS_AGENT_UNREACHABLE

[tomaskir]

Rather than just making the alarms go away, I would suggest investigating the root issue, if NetXMS agent really is unreachable or not during those alarms.

You can start with some simple debugging like this:
Setup monitoring of TCP port 4700 from the NetXMS server to the node (with for example 10 sec poll time), and see if the server does or doesnt actually have connectivity issues to the NetXMS agent on that node.

[Nomis]

Thank you, you're probably right. Unfortunately I'm not an experienced NetXMS admin. While I have successfully set up monitoring of system services, I can't figure out how to monitor a certain port (cannot find it in the Admin Guide or via Google either). Would you be so kind to point out the direction for me?

[tomaskir]

1. Enable PortCheck subagent
   Add this line to the main section of the agent config file on the server

SubAgent = portcheck.nsm

Restart the agent so it loads the new config and the SubAgent.

2. Create a DCI on the monitored server node.
   Origin "NetXMS Agent"
   Click "Select" next to "Parameter"
   Find "ServiceCheck.Custom" and use it.
   Modify the "Parameter" like this (substitute node IP for 1.2.3.4):

ServiceCheck.Custom("1.2.3.4", "4700", "1000")

Set "Polling mode" to custom and to 1 second.
Set "Source node" to your NetXMS server.

3. Set threshold to create alarms when NetXMS Agent connection is unavailable
   Under the DCI, go to "Thresholds" section
   "Add" and just change "Operation" to "!= : not equal to" and keep "Value" at 0

[Nomis]

Thank you very much, tomaskir. I've created a rule according to your guide. It does not set off any alarm. The previous alarm keeps coming and going as before, but the one I created is silent.

[victor]

Hi,

try to switch agent logging to file if it's not done already (by changing LogFile parameter in nxagentd.conf), set debug level to 6 (by adding DebugLevel = 6 to nxagentd.conf) and run agent for some time. After you got some agent unreachable errors, post your log file (or send it to [email protected]).

Best regards,
Victor

[Nomis]

Thank you. I did as you suggested. Where will the log file end up?

[victor]

You should set full path to log file in LogFile parameter in nxagentd.conf.

Best regards,
Victor

[Nomis]

Hm. This is quite weird. If I add the line "LogFile = C:\Program\NetXMS\etc\" (without quotation marks, and there's nothing wrong with the path), I am unable to restart the NetXMS service, or start it when it is stopped. If I delete the line, I can restart/start the service.

I came to this in my fruitless attempts to generate a log file, I assumed that the service needed a restart to have the file created after the parameter was added.

[tomaskir]

You need a file path, if you give it just directory, it will of course fail to start.

Use for example:
"LogFile = C:\Program\NetXMS\etc\nxagentd.log"

Then restart the Agent service to apply that change.

[Nomis]

Thank you, tomaskir. When I at first wrote [Path]\logfile.txt, NetXMS created a folder called logfile.txt, so I assumed that that last part should be a folder, but obviously the file extension .log is the key. -- I'll get back with the log file contents!

[Nomis]

Nothing seems to get logged except the start of the service. Since then there have been several "Native agent is not responding" alarms, but nothing the logging picked up on.

[25-Jan-2016 11:52:02.347] Log file opened
[25-Jan-2016 11:52:02.347] [INFO ] Additional configs was loaded from F:\Program\NetXMS\etc\nxagentd.conf.d
[25-Jan-2016 11:52:02.347] [INFO ] Debug level set to 0
[25-Jan-2016 11:52:02.550] [INFO ] DB Library: Database driver "sqlite.ddr" loaded and initialized successfully
[25-Jan-2016 11:52:02.659] [INFO ] Subagent "WINNT.NSM" loaded successfully
[25-Jan-2016 11:52:02.659] [INFO ] Subagent "ecs.nsm" loaded successfully
[25-Jan-2016 11:52:02.659] [INFO ] Subagent "ping.nsm" loaded successfully
[25-Jan-2016 11:52:02.659] [INFO ] Subagent "logwatch.nsm" loaded successfully
[25-Jan-2016 11:52:02.659] [INFO ] Subagent "portcheck.nsm" loaded successfully
[25-Jan-2016 11:52:07.573] [INFO ] Subagent "winperf.nsm" loaded successfully
[25-Jan-2016 11:52:07.573] [INFO ] Subagent "wmi.nsm" loaded successfully
[25-Jan-2016 11:52:07.573] [INFO ] Subagent "ups.nsm" loaded successfully
[25-Jan-2016 11:52:08.587] [INFO ] Listening on socket 0.0.0.0:4700
[25-Jan-2016 11:52:08.587] [INFO ] Listening on socket [::]:4700
[25-Jan-2016 11:52:09.601] [INFO ] NetXMS Agent started

[tomaskir]

Did you set "DebugLevel = 6"?

[Nomis]

Yes, "DebugLevel = 6".

[tomaskir]

@nomis said in SYS_AGENT_UNREACHABLE:

Yes, "DebugLevel = 6".

You log file shows

[25-Jan-2016 11:52:02.347] [INFO ] Debug level set to 0

[Nomis]

Am I missing something obvious or is this the Twilight Zone. I once more opened the nxagentd.conf file, ensured that DebugLevel is 6, saved and closed, restarted the Agent, and the log file once again says "0". This is the content of the conf-file:

#
# NetXMS agent configuration file
# Created by agent installer at Thu Sep 29 15:41:41 2011
#

MasterServers = 192.168.25.17

FileStore = F:\Program\NetXMS\var
LogFile = F:\Program\NetXMS\etc\nxagentd.log
DebugLevel = 6
SubAgent = ecs.nsm
SubAgent = ping.nsm
SubAgent = logwatch.nsm
SubAgent = portcheck.nsm
SubAgent = winperf.nsm
SubAgent = wmi.nsm
SubAgent = ups.nsm

[tomaskir]

Please also check "F:\Program\NetXMS\etc\nxagentd.conf.d" if there is some config there or not.

[Nomis]

That folder is empty.

[victor]

Hi,

check if you have -D0 in service command line. If yes, remove it.

Best regards,
Victor

[Nomis]

That did the trick, Victor. Now there's a lot of logging going on. I'll get back tomorrow with the result.

[Nomis]

I'm attaching an hour's worth of the log file, between 09:00 and 10:00. During that time there were five alarms, approximately 09:17, 09:21, 09:23, 09:32, 09:42, and 09:48. Thanks for taking time!

---

NomisLog.txt