NetXMS Support Forum

Hi,

you right, seems that I forgot to fix it. Fixed now (will be in 1.2.14).

Best regards,
Victor

Hi!

I won't be able to help you much. #1, #2, and #4 need changes in agent (#4 also in server). #1 is implemented in 1.2.14 (event source, id, and severity passed as three additional parameters for event generated). #2 should be relatively easy to implement, I'll add this as feature request for 1.2.15. For #4 we plan to introduce new policy type to distribute parser configurations across nodes in a same way as you can distribute configuration files, but I cannot promise exact release when it will be implemented.

Best regards,
Victor

You can query FileSystem.Total(E:) using nxget to check it.

Best regards,
Victor

First problem is that message is empty - this is because of wrong parameter count in event tag - it should be <event params="1">100049</event>. But spam problem most likely caused by this:

[14-May-2014 10:15:41.670] [DEBUG] LogParser: file size differs for stat(4) and fstat(C:\NetXMS\var\log), assume file rename
[14-May-2014 10:15:41.670] [DEBUG] LogParser: file "C:\NetXMS\var\log" (pattern "C:\NetXMS\var\log") successfully opened
[14-May-2014 10:15:41.670] [DEBUG] LogParser: parsing existing records in file "C:\NetXMS\var\log"

we recently encounter this bug on another installation - agent erroneously assume that file was renamed, and start reading file from the beginning. It is already fixed in current development version.

Best regards,
Victor

Hi!

No, tables from scripts are not supported. However, you can use dbquery subagent to read tables from database (it could be sqlite database for example if you need something local).

Best regards,
Victor

Hi!

Condition (value != null || value != 0) is always true - if value is null then it will not be 0, and vise versa. Correct condition will be (value != null && value != 0), so full script could look like this:


snmp = CreateSNMPTransport($node);
value = SNMPGetValue(snmp, ".1.3.6.1.2.1.16.1.1.1.3." . $1);
return (value != null) && (value != 0);


@sperlm:

you are right, returning 0 is the same as returning false.

Best regards,
Victor

I'm a bit lost - are you sure this is log for parser mentioned in the beginning? From this log file I see that parser correctly found matching line (using StatusPoll.* pattern) and generate event. Parser in the first post had different matching pattern. Let's try to combine everything related in one post - server log and agent log for same time period plus actual parser definition.

Best regards,
Victor

Script looks correct. Could it be that disk E: has non-zero total space (could be the case if this is CD drive with disk inserted or any other read-only device)?

Best regards,
Victor

Судя по сообщениям об ошибке (ПРЕДУПРЕЖДЕНИЕ:  нехватка разделяемой памяти), проблема в настройках PostgreSQL сервера. Увеличте размер shared memory, должно помочь.

Hi,

!= null should work. Can you show full script please?

Best regards,
Victor

Hi!

Actually, if node is compoletely down, and SYS_NODE_DOWN event is generated, all SYS_IF_DOWN and SYS_SERVICE_DOWN events should be correlated to SYS_NODE_DOWN and not cause message sending. However, if node is not down completely, you may get lot of events and notifications if multiple interfaces goes down for example. The only way to create digest messaging is to write your own script which will do that. For example, you may store event messages into text file in event processing policy (using "execute" type action) and run cron job every 5 minutes to send this file as email if it is not empty.

Best regards,
Victor

Can you please show how exactly problematic threshold looks like?

Best regards,
Victor

Hi!

It's definitely a bug - looks like threshold state was not saved across restarts, although it had to be. I'll check this.

Best regards,
Victor

Currently you cannot see what output agent was gathered nor measure response time. Feel free to fill feature request

Best regards,
Victor

Yes, that change already applied - so after upgrade to 1.2.14 you'll get this problem fixed as well.

Best regards,
Victor