NetXMS Support Forum

Currently it looks normal (except for DCI queue time peak - it can be caused by high CPU load at the moment or excessive number of simultaneous polls - you should watch if it will happen again). In any case, single peak is not an indicator of problem - problem is when these parameters instantly high.

For performance - from my experience, MSSQL performs better on large database. But maybe I just don't know how to tune MySQL...

Best regards,
Victor

Could you please send me configuration of all 4 DCIs and thresholds, and historical data for them?

Best regards,
Victor

Currently there are no better solution. In next release, I'll add some kind of "ignore for topology" attribute. I didn't decide yet how it should be implemented - as global subnet exclusion list or interface exclusion list for node, or even interface object's attribute. What way you would prefer?

Best regards,
Victor

Надо сначала настроить агента. В конфиге агента надо добавить записи о каждом устройстве в секции UPS, например:


*UPS

Device = 0:"COM1:":APC


Первое поле - это индекс устройства, он потом используется как instance в DCI. Второе поле - порт. Третье - протокол. Варианты протоколов - APC, BCMXCP, MICRODOWELL.
Протокол BCMXCP используется устройствами производства HP, PowerWare, возможно кем-то еще.

Уточнение - надо создавать аларм из события, которое было создано из трапа. Т.е. цепочка такая:

SNMP Trap -> Event -> Alarm

Вообще-то такая проверка там есть, и такой ситуации быть не должно. А проблема точно в одинаковых id? Кстати версия 0.2.21 теперь пишет и текст ошибки которую вернул SQL сервер, будет легче такие ситуации разбирать.

Можно будет сделать. Приму к сведению.

Unfortunately I cannot reproduce this problem myself. Is it only for PDF printer or for physical printer too? About crash when copying to clipboard - is final 0.2.21 console affected as well?

Best regards,
Victor

When you save graph as predefined, systems saves current graph window title, whatever it is. It does not relate to menu item name. If you create multiple graphs by changing DCIs and saving them under different names, you can get this behaviour. However, it should be considered a bug, because it's not correct - there should be at least prompt for graph window title when it saved as predefined. I will register this as a bug in our bug tracking system.

Best regards,
Victor

Can you see appropriate events in event log? If they are, than your problem is in event processing policy. If not, problem is with thresholds or DCI configuration.

For thresholds on strings - you can use "equal", "not equal", "like", and "not like" operations in thresholds for string DCIs. Or use transformation script to transform them into numbers if it's possible.

Best regards,
Victor

By default, if node already has a name, NetXMS will not change it automatically. You can set server configuration parameter SyncNodeNamesWithDNS to 1, then NetXMS will synchronize node names with DNS names during configuration poll, but not with system name available through SNMP (nobody asks for it ). I can add this type of name sync in next release.

Best regards,
Victor

NetXMS consists of multiple projects, depending on each other. libnetxms is a core library used by each NetXMS component. Did you compile libnetxms first?

Best regards,
Victor

I just check it, and it looks like all .dsp and .dsw files in source package are broken - all CR/LF pairs changed to single LF characters, most likely because we create source packages on UNIX. I'll fix it and create an updated source package.

Best regards,
Victor

Hello!

Unfortunately maps is probably the worst part of a system which requires a lot of improvements. Your problem cannot be solved in current release. Work on new map implementation is under way, but it is not ready yet. However, I plan to fix at least big problems in current map implementation. Next release of NetXMS should contain some improvements there.

Best regards,
Victor

ок, сделаю.