NetXMS Support Forum

Hi!

At a first glance, looks like you will need some external script. With built-in mechanisms, you can only check for entire page changes using ECS subagent. This subagent provides parameter which returns hash for given URL - so you can detect change in that page, but you cannot determine what part of the page was changed.

Best regards,
Victor

Hi!

1. do you need to check for file existence or to search for some string inside the file? To check for file existence, you can use File.Count parameter - it returns number of files matching given name or pattern. To search for string inside file, you should setup logwatch subagent for that file.

2. Parameter Process.Count(process_name) will return you number of running processes with given name. You can schedule DCI to be collected once per hour, and generate event if it is running (or not running).

Best regards,
Victor

Hi!

Yes, all custom messages comes to m_pMsgWaitQueue of AgentConnection. You can call waitForMessage in separate thread to wait for these messages and process them. For convenience, I have added new virtual methiod in AgentConnection class - processCustomMessage - which can be overriden in derived classes for processing custom messages. It will be avaiable since 1.0.9. Also, you can use CMD_TRAP message instead of custom one - then AgentConnection::onTrap will be called for each CMD_TRAP.

Best regards,
Victor

Hi!

Theoretically it is possible, but it will require a lot of work, because many places in agent code was written with usage of char instead of TCHAR, and because many system APIs on UNIX systems exist only in multi byte or event single byte form. I think that it can be relatively easy to convert Windows agent to unicode, and much harder to convert UNIX agent to unicode. Why you need unicode in agent?

Best regards,
Victor

Hi!

1.0.9 is not released yet. I'll publish it in a few days.

Best regards,
Victor

Hi!

It's my first try to create .deb files. I plan to add interactive configurator later which will be started if configuration files are absent. I agree that for now at least a warning message is necessary. Will fix it for 1.0.9 release.

Best regards,
Victor

I have added these MIBs to standard MIB collection in NetXMS. There was few changes. Also, APACHE2-MIB cannot be compiled by current version of MIB compiler because of a bug. I have fixed this bug for 1.0.9.

Best regards,
Victor

Hi!

It's our main focus now. It's already available for preview, but it has around 50% of required functionality implemented, and still cannot be full replacement for old console. However, early testing and feedback are welcome

Best regards,
Victor

Hi!

I have applied your patch, next release will have it. However, it will not solve all problems, because there are many places in server code where problems can occur if multibyte encoding is used. The only complete solution is to build server with unicode support. I've just finished converting server to unicode support - all 1.1.x versions will be built using unicode. I will not backport this to 1.0.x branch, as it is considered stable, and latest changes requires extensive testing. It will be great if you and other people from countries with non-latin alphabets can spend some time testing 1.1.x versions. I'll publish 1.1.0 installer on web site soon.

Best regards,
Victor

P.S. Attached is a screenshot of console where Latin, Cyrillic, Japanese, and Arabic characters used simultaneously (and was successfully stored in database).

It's rather "known limitation" - it was implemeted that way. I agree that it's not best way and must be fixed, but I'm not sure I'll fix this in old console - I hope to switch to new console as soon as possible. New console already have line charts implemented, with correct zoom (on both axis).

Best regards,
Victor

Event log show events, and on "Alarms" tab shown active (not terminated) alarms. In short, alarms is an events which requires operator attention. They are created from events in event processing policy. Main difference from event is that alarms has state (outstanding, acknowledged, or terminated) and repeat count - so if for example you have repeated events coming every minute, in alarm browser you will only see one alarm with increasing repeat counter. Alarms described in more details in user manual, section 6.3.

Best regards,
Victor

U eventa severity zestko zadan. Esli est' neskol'ko thresholdov, to obichno delaetsja neskol'ko eventov i v zavisimost' ot thresholda generirujutsja raznie eventi.

It can be done already by creating additional DCI (you can use dummy DCI or one of the parameters you wish to sum up as the base) and sum all required values in transformation script.

Best regards,
Victor

Quote from: paul-k on December 09, 2010, 04:59:57 PM
несколько вопросов в основном про графики:

1) не могу понять, каким образом отображаются графики параметров на perfomance tab, наследуемых от System.Agent'a? в темплейте в DCI параметры не вынесены на вкладку perfomance, но, тем не менее - отображаются в соответствующих нодах (параметров пять, графика, что характерно - 4).

Eto hardcoded konfiguracija. Iznachal'no bila tol'ko ona, potom ja sdelal vozmoznost' dobavit' ljuboj DCI na performance tab. Teper' dumaju chto hardcoded parametri mozno i ubrat'.

Quote from: paul-k on December 09, 2010, 04:59:57 PM
2) можно ли изменить масштаб (scope) на perfomance tab?

Net. Ja planiruju eto sdelat', no skoree vsego uze tol'ko v novoj konsoli.

Quote from: paul-k on December 09, 2010, 04:59:57 PM
3) можно ли строить два (и больше) графика на одном - например совмещать network in/out или нагрузку по разным процам на одном графике?

Net. No toze v planah.

Quote from: paul-k on December 09, 2010, 04:59:57 PM
и вопрос не про графики:
не понял, где задается правило если SYS_IF_DOWN => SYS_NODE_CRITICAL. В Event Proc Policy вроде бы стоит (по умолчанию) - Severity: From Event... (хочется, чтобы по SYS_IF_DOWN статус узла становился Minor)

Status hosta opredeljaetsja kak most critical iz aktivnih alarmov i podchinennih ob'ektov (obichno interfejsov). Sootvetstvenno esli interfejs nedostupen, to ego status CRITICAL, i host stanovitsja CRITICAL. Eto mozno menjat', menjaja Status Calculation u hosta ili Status Propagation u interfejsa - naprimer mozno v Status Propagation interfejsa zadat', chto svoj status CRITICAL naverh on otdaet kak MINOR.

Hi!

Appearance of these graphs depends of availability of appropriate parameters on agent. As on Windows physical memory stats provided by core agent, and CPU and disk stats provided by winperf subagent, most likely you do not have winperf.nsm loaded on your node.

Best regards,
Victor