IBM XIV Storage

[fore]

Есть вышеназванный сторадж, который хочется мониторить. Вариантов немного, а, если исключть трапы, то всего 1 - xcli.
Xcli - скрипт, где сплошная java. Я могу запустить скрипт с нужным IP + параметр и отгрепать нужное из результата.

Вопрос: как сказать zabbix'у чтоб запускал этот скрипт (на машине zabbix_server) и передавал ему нужный параметр, собирая назад результат? Спасибо.

[dima_dm]

Ïðèìåð ðåàëèçàöèè

http://www.zabbix.com/forum/showthread.php?t=14828&highlight=wbem

[fore]

Большое спасибо! А для менее продвинутых пример опроса можно? Я, например, должен для статистики запустить:
./xcli -s -u user_name -p user_password statistics_get count=1 interval=1 resolution_unit=minute local_fc_port=1:FC_Port:5:1 end=2010-01-13.14:15:00 -m IP_ADDRESS

чтобы получить (csv):

"Time","Read Hit Very large - IOps","Read Hit Very large - Latency","Read Hit Very large - Throughput","Read Hit Large - IOps","Read Hit Large - Latency","Read Hit Large - Throughput","Read Hit Medium - IOps","Read Hit Medium - Latency","Read Hit Medium - Throughput","Read Hit Small - IOps","Read Hit Small - Latency","Read Hit Small - Throughput","Read Miss Very large - IOps","Read Miss Very large - Latency","Read Miss Very large - Throughput","Read Miss Large - IOps","Read Miss Large - Latency","Read Miss Large - Throughput","Read Miss Medium - IOps","Read Miss Medium - Latency","Read Miss Medium - Throughput","Read Miss Small - IOps","Read Miss Small - Latency","Read Miss Small - Throughput","Write Hit Very large - IOps","Write Hit Very large - Latency","Write Hit Very large - Throughput","Write Hit Large - IOps","Write Hit Large - Latency","Write Hit Large - Throughput","Write Hit Medium - IOps","Write Hit Medium - Latency","Write Hit Medium - Throughput","Write Hit Small - IOps","Write Hit Small - Latency","Write Hit Small - Throughput","Write Miss Very large - IOps","Write Miss Very large - Latency","Write Miss Very large - Throughput","Write Miss Large - IOps","Write Miss Large - Latency","Write Miss Large - Throughput","Write Miss Medium - IOps","Write Miss Medium - Latency","Write Miss Medium - Throughput","Write Miss Small - IOps","Write Miss Small - Latency","Write Miss Small - Throughput","Time (s)"
"2010-01-13 14:14:00","0","0","0","43","2204","5110","32","948 ","815","136","491","790","0","0","0","1","9825"," 200","13","9021","424","105","11255","515","0","0" ,"0","1","5166","177","60","3253","2120","538","87 2","2823","0","0","0","3","3273","387","74","1402" ,"1719","219","1093","825","1263388440"

И результат только через 12-16 секунд приходит (zabbix'у timeout ещё учитывать)...

Т.е. пока, кроме запуска по крону с подстановкой времени на 10 минут раньше (во избежание разницы во времени, что бы не запросить "будущее") с последующим парсингом и отправкой через zabbix_sender - ничего придумано не было

[dima_dm]

В Zabbix agent можно поставить Timeout 30 cek. Нужно распарсить результат и записать в файл в формате (через ограничитель ключа, в примере символ двоеточие и разделитель пробел или табуляция )
Ключ: Значение
А дальше получить нужные значения другими вызовами.
P.S. ограничитель ключа нужен, чтобы отличать ключи, когда имя ключа полностью содержится в другом ключе при моём способе извлечения данных.
Пример:
SMX_NumericSensor_18: 2
SMX_NumericSensor_1: 2
Если бы не было ограничителя ключа, то на запрос SMX_NumericSensor_1 можно получить некорректные данные.