NetXMS Support Forum

Quote from: Anth0ny on May 28, 2009, 02:34:43 PM
nxpush -u "admin" -P "password" monitor monitor:Test=3403
Push failed at record #1 (#1 in batch): Invalid object ID.

Invalid object ID - это оно о чём?

Eto znachit, chto server ne nashel ob'ekt nodi s ukazannim imenem (monitor). Nado ukazivat' libo imja hosta tak, kak ono propisano v NetXMS'e, libo ego ID.

Quote from: Anth0ny on May 28, 2009, 11:35:50 AM

Quote from: Victor Kirhenshtein on May 28, 2009, 10:55:07 AM
A kakie znachenija na samom dele u etih counterov?

вот, я сделал небольшую урезанную выборку штатными средствами perfmon'а.


"(PDH-CSV 4.0) (Russian Daylight Time)(-240)","\\MYSERVER\PhysicalDisk(2 T:)\Avg. Disk Read Queue Length","\\MYSERVER\PhysicalDisk(2 T:)\Avg. Disk Write Queue Length"
"05/28/2009 12:09:24.908"," "," "
"05/28/2009 12:11:07.347","0","0"
"05/28/2009 12:11:08.348","0","0"
"05/28/2009 12:11:09.363","0.022350340104239228","0"
"05/28/2009 12:11:10.379","0","0"
"05/28/2009 12:11:11.379","0","0"
"05/28/2009 12:14:04.960","0","0"
"05/28/2009 12:14:15.960","0","0"
"05/28/2009 12:14:16.961","0.010499798403870646","0"
"05/28/2009 12:14:17.961","0","0"
"05/28/2009 12:14:18.976","0","0"
"05/28/2009 12:14:19.976","0","0.078398494748900815"
"05/28/2009 12:14:20.976","0","0"
"05/28/2009 12:14:21.976","0","0"
"05/28/2009 12:14:22.976","0","0"
"05/28/2009 12:14:23.976","0","0"
"05/28/2009 12:14:24.976","0","0.029299437450800946"
"05/28/2009 12:14:25.976","0","0"
"05/28/2009 12:14:26.976","0","0"
"05/28/2009 12:14:27.976","0","0"
"05/28/2009 12:14:28.976","0","0"
"05/28/2009 12:14:29.976","0","0.061698815382744653"
"05/28/2009 12:14:30.976","0","0"
"05/28/2009 12:14:31.976","0","0"
"05/28/2009 12:15:11.180","0","0"
"05/28/2009 12:15:12.180","0","0"

Teper' ponjatno. Esli v konfige zadano

AverageDiskWriteQueueLenght:"\PhysicalDisk(2 T:)\Avg. Disk Write Queue Length":60:A:INT:"Average Disk T: Write Queue Length"

to agent vozvraschaet srednee znachenie za minutu ukazannogo countera. Poskol'ku znachenija v osnovnom 0, inogda popadajutsja sotie doli, to srednee poluchaetsja ochen' blizko k 0. Vozmozno esche i okrugljaetsja do celogo (sudja po tipu INT v konfiguracii).

Quote from: Anth0ny on May 28, 2009, 10:56:16 AM
может чтото нужно особым образом настраивать в DCI (например Average Delta?)

я кстати по наивности думал что при добавлении своего канутера на Агенте его имя будет светиться в общем списке при нажатии на кнопку "Select" при создании DCI.

=))

Budet, no tol'ko posle sledujuschego configuration poll'a. Nado libo podozdat', ibo sdelat' configuration poll prinuditel'no.

Eto sdelano zatem, chto pri kazdom nazatii na knopku "Select" obraschatsja k agentu ne ochen' praktichno - on mozet bit' za medlennim linkom, etc. Poetomu server keshiruet informaciju o podderzivaemih parametrah, i obnovljaet vo vremja configuration poll'ov. Nu i parametri dobavljjutsja/udaljajutsja ne tak chasto, kak nazimaetsja knopka "Select".

A kakie znachenija na samom dele u etih counterov?

Yes, you can use extended form of PDH.CounterValue:

PDH.CounterValue("counter name",1)

If second argument is present in PDH.CounterValue and set to non-zero, then agent will get two samples with 1 second interval, and use them to calculate resulting value.

V techenii leta Tochnee poka skazat' ne mogu, ochen' mnogo raboti. Iz polozitel'nih momentov - chast' etoj raboti svjazana s razvitiem NetXMS.

Eto srednee znachenie countera \Physical Disk(_Total)\Current Disk Queue Length za minutu.

A esli napisat'

AverageDiskWriteQueueLenght:"\PhysicalDisk(2 T:)\Avg. Disk Write Queue Length":60:A:INT:"Average Disk T: Write Queue Length"

?

Svjazat' s tem ob'ektom, kotorij pinguetsja, mozno - nado DCI delat' na toj node, kotoruju pinguem, no v nastrojkah ustanovit' "proxy node" v tu nodu, kotoraja delaet ping - togda dannie budut real'no bratsja s proxy node, no DCI (a takze sobitija i alarmi, vizvannie thresholdami) budut privjazani k toj node, kotoruju pinguem.
S ostal'nim - soglasen, budu dumat' kak eto po udiobnej sdelat'.

Bez peredelki servera etu problemu normal'no ne reshit'. Ja dumaju sdelat' sledujuschie izmenenija - dobavit' flazok dlja DCI, chto esli OID ne suschestvuet, to generirovat' oshibku sbora dannih, a ne perevodit' v sostojanie not supported. Nu a na oshibku sbora dannih mozno postavit' threshold i dal'she obrabativat' kak obichno.

Da, eto izvestnij gluk. No on vrode bil svjazan s versiej internet explorer'a.

Ne objazatel'no pingom. Vo vremja kazdogo status poll server proverjaet, otzivaetsja li NetXMS agent, SNMP agent, i tol'ko potom, esli nikto ne otozvalsja, delaet ping. Prichem ping delaetsja na kazdij interfeis, i esli otveta ne prihodit voobsche, to interfejs perehodit v status critical. Esli vse interfeisi stali criticval, togda i ves' host perehodit v status critical.

Esli est' neobhodimost' sohranjat' vremja otklika ping'a, to nado delat' dopolnitel'nij DCI. Na mashine, s kotoroj delaetsja PING, dolzen bit' zagruzen PING subagent. Vot neskol'ko obsuzdenij pro ping:

https://www.netxms.org/forum/index.php/topic,491.0.html
https://www.netxms.org/forum/index.php/topic,456.0.html
https://www.netxms.org/forum/index.php/topic,553.0.html

Nu i navernoe esche mozno informaciju na forume najti.

V etom sluchae pridetsja delat' dva pravila s filter-scriptami. Obschaja posledovatel'nost' shagov:

1. V nastrojkah trapov vitaskivaem varbind .1.2.3.4.5 v parameter eventa (pust' budet parametr nomer 2)
2. V event policy delaem pravilo dlja generacii alarma: event, naznachennij trapu, i filtering script


$2 == 1


kljuch mozno sdelat' na osnove kakogo-nibud' teksta i _%i - chtobi razlichat' alarmi s raznih hostov.

3. V event processing policy delaem vtoroe pravilo - dlja terminate alarma. Event i kljuch alarma ukazivaem te-ze, a filtering script delaem kak


$2 == 0


teper' pervoe pravilo budet srabativat', esli nash varbind raven 1, a vtoroe - esli on raven 0.

It's a bug in MIB compiler - in lines 18622 and 18624 there are tab between words OBJECT and IDENTIFIER instead of space. As workaround, replace these tabs with space character and MIB should compile without error.

Best regards,
Victor

Quote from: Ethril on May 18, 2009, 06:03:53 PM
P.S. И ещё вопросик вдогонку: не планируется ли доработать в будущем модуль для update/deploy клиентов, чтобы заодно распространять на них файлики с парсерами? Чтобы не копировать ручками .xml-файлик на 20 серверов.

Планируется - будет возможность централизованно распространять конфиги, скрипты, и т.д.