NetXMS Support Forum

Hi!

Check where critical status comes from - it could be either interface object below host or outstanding alarm. Terminate alarm(s) if they are not relevant. For interface object, set expected state to IGNORE. Also you may want to disable configuration polls for that node.

Best regards,
Victor

Hi!

You probably need to show your event processing policy. I suppose there is a flaw in event processing logic.

Best regards,
Victor

You can add special threshold type "data collection error" on DCIs where you want to detect collection errors.

Best regards,
Victor

Hi!

There are different functionality in NetXMS that is called proxy. One is agent proxy - when one agent used as a proxy to connect to another agent - this functionality you don't need. What you need is "proxy node" in DCI configuration - it allows you to configure DCI on one node but specify that real data collection will occur on another (proxy) node. In your case you will configure agent DCIs on ESX node and set proxy node to some ode with agent which can really collect this data. Don't forget to add ESX nodes to the list of trusted nodes on collecting node or disable trusted nodes check altogether.

Best regards,
Victor

Quote from: scuq on March 31, 2014, 02:05:45 AM

Code Select Expand

if (AddrInSubnet($node->ipAddr, "10.16.0.0", "255.255.0.0")) {
return;
}

is this valid syntax?

Syntax is valid, but depending on context (for example if you are writing a filter script) you may need to return TRUE or FALSE instead of null value.

Best regards,
Victor

Then to clear flag you can use the following construct:


  newFlags = node.getFlags() & ~Node.NF_DISABLE_NXCP
  #newFlags = node.getFlags() & ~(Node.NF_DISABLE_NXCP | Node.NF_DISABLE_SNMP)
  md = NXCObjectModificationData(node.getObjectId())
  md.setObjectFlags(newFlags)
  session.modifyObject(md)


Best regards,
Victor

Можете подцепить gdb к процессу и сделать stack trace всех потоков?

А Windows английский или русский? Попробуйте запустить агента как сервис с дебагом и посмотреть PDH ошибки в логе.

Quote from: raypetter on March 29, 2014, 11:50:16 PM
Victor, please update the postgresql package so that it does NOT depend on postgresql 8.4

Most of us are running 9.1 and newer. For every update I need to extract the package and remove/update the package dependency (DEBIAN/control).

Thanks

I've changed dependency for Debian 7 package. Next release should be ok.

Best regards,
Victor

Hi!

If you can disable SNMP (by setting appropriate bit flag), then you can enable it back by clearing same bit flag. If you show me the code I probably will be able to suggest how to change it so it will work.

Newly discovered nodes comes with all communication channels enabled. They are checked on each configuration poll to detect any changes. System cannot predict why SNMP is not responding for example - vecause it is not enabled and never will, or because of misconfiguration, or it will e enabled next day.

With latest patches (that will be part of 1.2.14 release) it is also possible to disable let say SNMP for each new discovered node using NXSL script called on SYS_NODE_ADDED event.

Best regards,
Victor

Маленькое дополнение - в 1.2.13 починили HTTPS в portcheck субагенте, теперь ServiceCheck.HTTPS тоже будет работать.

А можно включить debug 6 (не останавливая сервер) и прислать мне кусок лога за минут 5-10. Плюс результат команд на дебаг консоли сервера

sh q

sh po

Выложил еще одну версию по тому-же адресу - попробуйте.

Пришлите пожалуйста результат SQL запроса

SELECT transformation FROM items WHERE transformation is not null AND transformation<>'';

Hi!

Just test it and got object details for map itself (which is also wrong but in a different way). Will check it later.

Best regards,
Victor