NetXMS Support Forum

Hi!

What OS you are using? Can you provide crash dump?

Best regards,
Victor

Hi,

it is defined by agent parameter TunnelKeepaliveInterval, but same interval also used for sending keepalive packets in existing tunnels.

Best regards,
Victor

Hi,

to get the dump, do the following:

1. Stop netxmsd process
2. Run command

gdb netxmsd

3. You'll get "gdb>" prompt - run command

run -D4

4. Wait for the crash. When process crashed, you'll get "gdb>" prompt again.

5. Run command

bt

and send me it's output.

If you have installed server from deb packages, make sure that package netxms-server-dbg is installed.

Best regards,
Victor

Hi!

You can use macros for forming date/time based file name (possible macros can be found here: http://www.cplusplus.com/reference/ctime/strftime/) or you can use output of shell script as file name using backticks, like this:


<file>C:\FX_CHUB\LOGS\`C:\gen_file_name.cmd`</file>


In this example file name will be taken from output of script C:\gen_file_name.cmd

Best regards,
Victor

You can open filter in event processing policy editor and type rule comments, event name, or object name to show only rule(s) containing search term.

Best regards,
Victor

Hi,

currently the only way I can think of is to create nxshell script that will request summary table and push DCI value. Summary tables are not accessible from NXSL scripts so simple script DCI is not possible.

Best regards,
Victor

Hi,

I don't see anything wrong in the dump - all threads seems to be waiting on something. Try to comment out all subagents in nxagentd.conf and restart agent. If CPU usage stay low, enable them one by one to find out which is responsible.

Best regards,
Victor

Hi,

You can do this with nxshell script, like this:


for node in [o for o in session.getAllObjects() if isinstance(o, objects.Node)]:
  print 'Processing ', node.getObjectName()
  session.pollNode(node.getObjectId(), NodePollType.CONFIGURATION_NORMAL, None)
  md = NXCObjectModificationData(node.getObjectId())
  md.setObjectFlags(node.getFlags() | Node.NF_DISABLE_CONF_POLL)
  session.modifyObject(md)


You can do mass enable of configuration poll like this:


for node in [o for o in session.getAllObjects() if isinstance(o, objects.Node)]:
  print 'Processing ', node.getObjectName()
  md = NXCObjectModificationData(node.getObjectId())
  md.setObjectFlags(node.getFlags() & ~Node.NF_DISABLE_CONF_POLL)
  session.modifyObject(md)


Best regards,
Victor

It actually looks like SNMP is not detected at all with new server version. Could you set debug level to 6, run configuration poll again, and check or post here relevant lines from the log (usually those with name or IP address of the node in question)?

Best regards,
Victor

Hi,

SSH credentials currently not accessible as macro or from the script. Feel free to register feature request for that in our tracker.

Best regards,
Victor

Quite strange. Looks like it get empty interface list from device. What device driver is used? Can you provide SNMP walk results on ifMib?

Best regards,
Victor

Hi,

please provide output of configuration and status poll for such node.

Best regards,
Victor

Hi,

can you create process dump when agent start using 100% CPU and send it to me?

Best regards,
Victor

Hi!

1) You can use actions with timers. For example, execute action on node down event with timer set to 5 minutes, and cancel this timer on "node up" event. If node will recover within 5 minute interval, timer will be cancelled and action never executed.

2) Yes, diff will be correct method in this case.

Best regards,
Victor
 

Fix is already made, I plan to made release 2.2.13 within a week.

Best regards,
Victor