NetXMS Support Forum

Net, ne pravil'no. Na padenie kazdogo interfeisa generiruetsja otdel'noe sobitie, i kazdoe iz nih prohodit cherez event policy. Prohozdenie sobitija zakanchivaetsja kogda zakanchivaetsja policy (vse pravila projdeni) ili vstrechaetsja pravilo "stop processing". No nezavisimo ot togo kak prekratilas' obrabotka tekuschego sobitija, budet vzjato sledujuschee sobitie iz ocheredi i process povtoritsja.

Mozno poprobovat' etot patch - nado zamenit' ber.cpp v src/libnxsnmp na prilozennij.

Net raznici, 1, 20, ili 100 - kazdoe sobitie obrabativaetsja nezavisimo. Tak chto dlja kazdogo sobitija esli uslovie sootvetstvuet i stoit stop processing - mi obrabotku etogo konkretnogo sobitija prekraschaem. No na drugie sobitija eto nikakogo vlijanija ne okazivaet.

Ja kazetsja ponjal v chem problema - NetXMS nekorrektno obrabativaet encoding OID'ov sostojashih tol'ko iz odnoj cifri. T.e. hotja


nxsnmpget -c public host .1


ne rabotaet, variant


nxsnmpget -c public host .1.3


budet raqbotat'. Mozno eto proverit'? A ja poka-chto pitajus' razobratsja kak pravil'no encodit' OID'i dlinnoj 1.

I didn't play much with new NNM 8i, but NNM 6.x and 7.x uses a slightly different approach - status of a node is depending on number of interfaces being down - for example if less than half of interfaces is down node status will be Minor, and if more than half is down - Major. If talking anout HP management products, we use approach similar to OpenView Operations - by default parent object's status equal to most critical child status, but you can change this by changing status propagation and/or calculation properties for either parent or child object.
For example, setting status calculation mode for switch node object to "multiple thresholds" and setting thresholds to Warning 0.1, Minor 0.2, Major 0.5, Critical 0.9 will give you kind of NNM-like status calcualtion for switch node.

Best regards,
Victor

A mozno tcpdump zapustit' s kljuchami -X -s 1000? Togda ja smogu sravnit' strukturi paketov.

Punkt 2 ja bi sformuliroval kak "prekratit' dal'nejshuju obrabotku sobitija".

V privedennom primere esli prihodit sobitie SYS_IF_DOWN, to budet sozdan alarm, vipolnen action SendMail_Down_IF, i na etom obrabotka sobitija zavershitsja. T.e. esli daze dal'she budet pravilo

Source Any
Event Any
Severity Any
Action SendMail

to vipolneno ono uze ne budet.

T.e. summiruja - dlja kazdogo sobitija server beret pravila po porjadku, proverjaet sobiti e na sootvestvie usloviju, esli uslovie istinno - vipolnjaet ukazannie v pravile dejstvija. V normal'noj situacii kazdoe sobitie propuskaetsja cherez kazdoe pravilo, no "stop processing" mozet dosrochno prervat' obrabotku sobitija (konechno esli sobitie sootvetstvuet usloviju v pravile).

Dlja System.ProcessList:

PID ProcessName

Prichem teoreticheski v imeni processa mogut bit' i probeli, t.e. vse chto idet posle pervogo probela i do konca stroki - eto imja processa.


Dlja System.ActiveUserSessions:

"username" "terminal" "client_name"

Best regards,
Victor

Hello!

You can put interfaces you are not interesting in into "unmanaged" state - then they will be excluded from polling or status calculation. To do so, select interface object in object browser, right-click on it and select "Unmanage". Alternatively, if you need to unmanage multiple interfaces at a time, select switch object in object browser, open "Subordinates" tab, select all interfaces you wish to unmanage, right-click and select "Unmanage".

Best regards,
Victor

Podozrevaju chto problema kak-to svjazana s tem faktom, chto Windows russkij - no proverit' tak shodu ne mogu, russkogo Windows'a pod rukoj net. Na sledujuschej nedele budu obratno v Rige, postavlju virtual'nuju mashinu i proverju.

Zapustit'


nxget -l my_host Agent.SupportedEnums


vidast spisok vozmoznih tablic. S kolonkami sloznee - nuzno znat' chto est' chto. Sprashivajte chto nuzno, budu otvechat'. Tak i zadokumentiruem

Naprimer u menja mnogo pravil takogo vida:

any_host event1 - do actions
any_host event2 - do actions
...
any_host eventN - do actions

I ja hochu iskljuchit' odin konkretnij host iz processinga - ja konechno mogu v kazdom pravile propisat' ego i postavit' "Negate", no proshe do togo sdelat' pravilo

bad_host any_event - stop processing

T.e. mozno ispol'zovat' esli snahala idet specializirovannaja obrabotka eventa dlja hostov, a potom - obschee pravilo dlja vseh ostal'nih. Togda v spec. pravilah stavim "stop processing" i obschee pravilo budet prosche.

Nu i mozno kazdoe pravilo otmechat' flazkom "stop processing" - togda budet logika kak u firewall policy - komu-to eto mozet bit' prosche.

Pro SNMP - ochen' slozno skazat' bez dopolnitel'noj informacii. Mne takuju situaciju poluchit' ne udalos'. Mozet bit' est' vozmoznost' tcpdump'om sobrat' paketi otsilaemie snmpwalk i nxsnmpwalk chtobi ih sravnit'?

Hello!

It's just a built-in alias for \PhysicalDisk(_Total)\% Disk Time performance counter. It is the percentage of elapsed time that all disk drives were busy servicing read or write requests. It can be used to measure physical disk load, but it is not related to % of CPU time spent on I/O.

Best regards,
Victor

Quote from: Alex on February 26, 2008, 02:06:38 PM
Еще одна проблема выскочила. Логинюсь под admin все путем. Но если я логинюсь под обычным пользователем пишет: Unable to connect: Access denied. Хотя в dev2 все работало прекрасно. Что интересно такая же бодяга как на рабочем сервере, так и на тестовом Менял права юзверям, однофигственно. В чем может быть проблема?

Eto bug, vilozil installer s ispravlennoj konsol'ju.