NetXMS Support Forum

Поведение системы действительно изменилось в версии 2.2.5 - до этого configuration poll на unmanaged нодах работал. Я проверил историю изменений - это было исправление для https://track.radensolutions.com/issue/NX-1420 - теперь сервер не пытается коммуницировать с нодами в состоянии unmanaged. Я думаю будет логично добавить в принудительный poll сообщение о невозможности опроса в таком состоянии чтобы исключить странное с точки зрения пользователя поведение системы.

Currently it is only possible if you use "umatched trap" event. You can create event processing rule for that event and use the following script as action to generate new event with necessary data:


if (($event->parameters[1] == '.1.3.6.1.2.1.15.7.2') &&
    ($event->parameters[2] ~= "\\.1\\.3\\.6\\.1\\.2\\.1\\.15\\.3\\.1\\.14\\.([0-9]+\\.[0-9]+\\.[0-9]+\\.[0-9]+)"))
{
   PostEvent($node, "BGP_PEER_DOWN", null, $1);
}


This will generate event BGP_PEER_DOWN (you have to define it) with peer address passed as parameter 1 (you can access it in message text via %1).

Best regards,
Victor

Так он же поставлен в "unmanaged" - поэтому собственно ничего и не опрашивается.

Да, и можно скриншот страничек Polling и Communications/SNMP из свойств ноды?

Сейчас обратил внимание, что в примере с nxsnmpwalk указана версия 1, а сервер выбрал 2c. Если в nxsnmpwalk указать версию 2c, тоже работает?

На самом деле добавилось:

2018.07.18 17:33:47.719 *D* Node::getInterfaceList(node=LAN1 [302091]): cannot create SNMP transport

И это очень странно. Сервер напрямую общается с устройством или через прокси?

Actually, I made a mistake here. There are hook for topology poll (it's not created by default for some reason, should be called Hook::TopologyPoll), but not for discovery poll. I'll add discovery poll hook in next release.

Best regards,
Victor

Только что выложили 2.2.7. Для отладки как и перед этим включите уровень 6 перед началом configuration poll.

Hi all!

NetXMS version 2.2.7 is out. Changes since previous release:

- "Convert to hexadecimal string" option for SNMP table columns
- Improved configuration wizard in Windows server installer
- New command line option /CONFIGENTRY in Windows agent installer
- New dashboard element "Object Details" (based on object queries)
- TCP proxy functionality
- Improved "Alarm Details" view in management console
- Added support for agent actions defined in external subagents
- TCP listener can be disabled in agent configuration file
- Push connector can be disabled in agent configuration file
- Virtual nodes detection via agent
- Basic BIOS and hardware platform information provided by Windows agent
- Fixed issues:
        NX-993 (Windows installer should honor current state of services during upgrade)
        NX-1438 (Increase debug level for thread pool messages)
        NX-1453 (Windows installer option to disable database upgrade)
        NX-1455 (Zone UIN in NXSL function CreateNode)
        NX-1456 (DCI values aggregated on cluster may be incorrect on first few polls after server startup)
        NX-1460 (Configure does not detect pthread_setname_np on macOS)
        NX-1464 (Agent connection via proxy is not working for IPv6 destinations)
        NX-1466 (Line length in agent's external table script limited to 255 characters)
        NX-1467 (Agent cannot process requests for parameters longer than 255 characters)

Best regards,
Victor

This is simple hook script that sets attribute "lastDiscoveryPoll" to current time for all node's subnets:


for(o : GetObjectParents($node))
{
   if (o->type == 1)
   {
      SetCustomAttribute(o, "lastDiscoveryPoll", time());
   }
}


Best regards,
Victor

I'm a bit lost - are we talking about hook script or about manually starting discovery poll? Hook script already exist - you can use it to update attributes on all subnets where current node belongs.

Best regards,
Victor

Hi,

what is the output if you run status poll and configuration poll from the GUI?

Best regards,
Victor

Hi,

could you please provide statistics for your server(s)?
- CPU, RAM, disk I/O usage for server machine
- Output of the following commands on debug console:
  - show stats
  - show queues
  - show threads
  - show watchdog
  - show pollers
  - show msgwq
  - show dbcp
  - show dbstats
  - show flags

Best regards,
Victor

Hi,

status is not propagated from maps to map groups. I've created an issue for that (https://track.radensolutions.com/issue/NX-1468).
Best regards,
Victor

Hi,

It is cut to 255 characters. Some time ago it was the limit on parameter name in DCI definition. When DCI name length limit was increased, limit on agent side was forgotten. I just fix it.

Best regards,
Victor