NetXMS Support Forum

In recent versions NetXMS is performing ICMP ping to node's primary IP address automatically and collects the statistics visible in Object Details. You can also create DCIs to store the history of these values. https://www.netxms.org/documentation/adminguide/icmp-ping.html

If you are trying to achieve something different, please describe in details.

Just to double-check, can you show your output of
netstat -aunp | grep 514

if you set debuglevel=6 in server config file and start server for a minute, do you see something like
Listening for syslog messages on UDP socket
in server log?

Which version are you using?

Just tried recent one, unexpected interface up is critical on an interface and changes node to critical. Same for interface down.

What about status calculation settings on the interfaces, not on the node, are they set to default?

Hi!

These 40 EEP rules are there just for demonstration, in real production situation there might be more rules.

Systems logic is the following

1) Events are created when something happens. There are many source of events - DCI threshold violation/restore, node status change, log parser, SNMP traps, etc.
Each event stores a number of parameter with it, e.g. DCI parameter name, threshold and actual values, etc. The number of parameters is different for different events.
There are "event prototypes" where we can configure event name and user message, which can have macros (they are expanded when event is created).

2) Events are processed by EPP. It could be that some events are not configured in EPP at all - in this case the event would be just recorded to event log, nothing else would happen.
EPP rules can be very flexible - e.g. one rule can handle multiple events, it can work only for nodes in specific container (or except for some nodes/containers) - check conditions in EPP rule properties.

3) EPP can create an alarm - it's sort of a bug-tracker built in netxms that shows list of current problems. It's also possible to terminate alarms automatically when some other event comes. In order for this to work alarms have alarm key, which is made of some text and macros. Macros can have, e.g. ID of node and ID of a DCI - see default rules 19 and 20.


Now it depends on what exactly hardware checks do you need. Node connectivity and status of network interfaces is checked once per minute via status poll - if it's down, corresponding event will be generated. Other checks are usually configured as DCIs, e.g. cpu temperature, that DCI has threshold and an event specified in in DCI properties is generated on threshold violation. You can use generic SYS_THRESHOLD_REACHED event, but usually it's more convenient to have separate events for various groups of thresholds, e.g. something like DC_SERVER_TEMPERATURE, etc.

Hi!

MIB file provides a more convenient way to select OIDs when creating DCIs (data collection items). If you right-click the node and go to Data Collection Configuration, you can create the DCIs for the information you need from that device.Each DCI has specific OID that provides collected values. You can also "Walk" a section of MIB tree to see which OIDs are available and what values they provide.
In DCI properties there's Performance Tab page which allows to configure, if that DCI should be shown in the performance tab.

Using agent tunnel connection on the same machine does not have any benefits (but should work anyways).
Did you try tunnel connection from agent on another machine? Did it work?
What's exact OS version? If it's debian/ubuntu, what's the output of   dpkg -l | grep libssl

Pls execute "sh st" in Tools -> Server Console and show the output.

Connect to postgres database and show the output of this query:

select schemaname as table_schema,
    relname as table_name,
    pg_size_pretty(pg_total_relation_size(relid)) as total_size,
    pg_size_pretty(pg_relation_size(relid)) as data_size,
    pg_size_pretty(pg_total_relation_size(relid) - pg_relation_size(relid))
      as external_size
from pg_catalog.pg_statio_user_tables
order by pg_total_relation_size(relid) desc,
         pg_relation_size(relid) desc
limit 50;

Do you have any DPI system between agent and server that could intercept ssl connections?

Просто список возможных OIDов, разделенных |

Похоже, что это та самая ошибка, которая починена в 3.8.262, пробуйте эту версию.

Can you have a look into log of Postgres DB. There should be a ERROR: line prior to the moment when
"current transaction is aborted, commands ignored until end of transaction block"
started to occur. This should give us some understanding of what's happened.

Hi!

Which exactly training materials, were they prepared by Raden Solutions or some third party?
I suppose disk_logger is some 3-rd party app that can send syslog messages. Not sure if it's giving any benefits over netxms agent when monitoring a system.

You need to create log parser definition in Configuration -> Syslog parser.

There could be several definitions, each is a filter that only reacts to particular conditions - syslog text, source, etc.
Each definition specifies an event that is triggered.
Then in Event Processing Policy you need to have a rules for these events. These rules can create alarms, terminate alarms, send notifications, etc.

What's exactly netxms agent and server versions?
What kind of connection is between server and agent - is it normal or tunnel?

Если у этих коммутаторов порты называются как-то по разному, то можно в summary table в свойствах колонки включить использование регулярных выражений и написать регулярку с OR.

Второй путь, более универсальный - делать на этих нодах дополнительный скриптовый DCI. В его скрипте определять модель коммутатора, брать последнее значение DCI соответствующего нужному порту и сохранять. Если история по этому DCI нигде не требуется, то для него можно выбрать время хранения "не сохранять в базу".
Ну или тоже самое можно сделать другим путем - DCI может быть не скриптовым, а push. А скриптоваться в transformation script оригинальных DCI и оттуда делать PushDCIData(node,dciId,value). Но так, возможно, более путанно и чуть хуже по производительности.