NetXMS Support Forum

Quote from: testos on February 07, 2013, 12:51:25 PM
Centos kernel is not supposed to panic under any circumstance.

That's pretty much standard trace issued by oomkiller.

Kevin: try to use attached script to get proper valgrind output. Script idea is that it monitors RSS used by netxmsd, and if it's greater than 1Gb, script try to do graceful shutdown using nxadm – before netxmsd is killed by OOMKiller.
You need to change NXADM variable in the script according to your installation prefix.
Start script, then start netxmsd under valgrind.

Yes, there should be no output with '-d', it's daemon mode - it should silently detach and go to background.
Before running it as daemon, you can check that config files are good and agent is working properly by running it with '-D3' (3 - debug level, can be from 0 to 9)

Maybe sever was initially installed with different prefix.
Check which nxdbmgr you are using: run which nxdbmgr
Also search for duplicates:find / -name nxdbmgr 2>/dev/null

Если есть core файл, то запустите пожалуйста "gdb /path/to/netxmsd core" и покажите результат команды "thread apply all backtrace".
Если core не создается, перед запуском сервера выполните "ulimit -c unlimited" (можно так же можно прописать в /etc/security/limits.conf)

I've added couple more samples to Wiki (http://wiki.netxms.org/wiki/Using_nxshell_to_automate_bulk_operations#Sample_Scripts)

Hello.

Please contact me at [email protected] and I will send you our consulting rates.

Best regards,
Alex.

Yes, all operations done by GUI can be done using API. Right now it's just a raw Java API with some glue for more convenient use, but I plan to make it more Python-ish.

I can add more samples – do you have any particular idea in mind?

Проблема в том, что при запуске под runit, readline получает EOF на каждый попытку прочитать консоль – чем и занимает одно ядро полностью.
В 1.2.6 я добавил параметр "-q", который полностью отключает интерактивную консоль.
Как временное решение - отключить консоль в коде и пересобрать сервер. Diff в аттаче.

Посмотрите историю по двум DCI на самом сервере (они создаются автоматически):
* Database writer's request queue (DCI data) for last minute
* Database writer's request queue (other queries) for last minute

Could you please enable crash dump collecting and send us minidump of the crash?

To enable it, you need to add these options in nxagentd.conf:
CreateCrashDumps=yes
FullCrashDumps=yes
DumpDirectory=c:\netxms\dump

You also need to create "c:\netxms\dump" directory.

It's yet unsupported.

There were no changes in portcheck subagent since 1.2.2.
Return code 3 means that connection was successful, but reply didn't match regexp.

Could you check request/response with tcpdump?

Hello.

After couple of requests we made a  small tool to help automate operations, which are currently available only thru user interface or java api. In essence, it's wrapper around Jython (Java implementation of Python) and NetXMS API, with some glue in between to make it easier to use.

More details in our Wiki: http://wiki.netxms.org/wiki/Using_nxshell_to_automate_bulk_operations

I updated wiki with sample script: http://wiki.netxms.org/wiki/Using_nxshell_to_automate_bulk_operations

Right now it's not possible. Services do not have "expected state", as interfaces do. Right now they are expected to be up, so if normal state for network service is "down", you'll always get critical status.
As a workaround, you can handle SYS_SERVICE_UP / SYS_SERVICE_DOWN (there is default rule for that in Event Processin Policy, so add your rule before that and select "stop processing" checkbox) to create alarm / send notification.
To prevent "critical" status on the node itself as well as whole tree, you can status propagation for service to fixed -> Normal. This way service itself will be shown with red dot, but whole tree will be left intact.