NetXMS Support Forum

Quote from: deepjohnsea on November 13, 2019, 03:08:55 PM
1. Ваша система все больше и больше нравиться. Заменила многие отдельные приблуды и тулзы.
Но вот в процессе наведения порядка в сети потребовалось менять огромное количество описаний портов (они же Alias) и самих железок.
Есть ли какой либо легальны способ это сделать прямо из NetXMS.

Сейчас только из скриптов, используя функцию SNMPSet. По идее можно сделать object tool который запрашивает интерфейс и значение для alias и вызывает SNMPSet.

Quote from: deepjohnsea on November 13, 2019, 03:08:55 PM
2. Также почему то не на всем оборудовании отрабатывает получение списка MAC адресов (FDB). Абсолютно одинаковое оборудование (Cisco 2960) с одинаковой IOS, но на одном все великолепно на другом (Cannot get switch forwarding database). Может есть какие то условия для работы?

Сервер читает FDB используя .1.3.6.1.2.1.17.1.4.1.2, .1.3.6.1.2.1.17.7.1.2.2.1, и .1.3.6.1.2.1.17.4.3.1 - проверьте, что проблемное устройство возвращает на SNMP walk по этим OID.

Quote from: deepjohnsea on November 13, 2019, 03:08:55 PM
3. Нашли ошибку VLAN View. У наc есть две Cisco 3750 в кластере. Одна с Gi интерфейсами, а другая с Fa. Почему так - не спрашивайте.
Короче подсвечивает неправильно порты. На картинке, строчка выделена, и видно что нет порта Fa4/0/4/

Можете прислать скриншот закладки "Interfaces" для этого устройства?

А везде это где? У разных компонентов последнее число может различаться - например текущая версия консоли это 3.0.2329, сервера - 3.0.2357, а AIX агента - 3.0.2354. Начиная с версии 3.0 первые два числа обозначают версию продукта, а третье - patch release. Все компоненты с одинаковыми первыми двумя числами совместимы между собой. Если мы делаем патч, то обновляем только те компоненты, которые затронуты изменениями - и у них меняется третье число в номере версии.

There is no standard way of getting this information. If device supports entity MIB then most likely you can get P/N and S/N from there. If not you should looks at vendor specific MIBs - many vendors do provide that information in their custom MIBs.

Best regards,
Victor

Please share your config.log.

Best regards,
Victor

Hi,

problem is that name of those nodes does not starts with "cab", it is in the middle. If you want to match any node that contains cab it will be different:


return $node->name imatch ".*cab.*";


This will match any name with letter combination cab at any position using regular expression. If you want something more complex you can use more complex regular expression. Site https://regex101.com/ is a good resource for testing your expressions.

Best regards,
Victor

Yes, TimescaleDB could increase graphing speed. You have to experiment. If you have enough disk space you can do migration from one to another quite fast and check the difference.

Best regards,
Victor

Hi,

you can try this script:


p = substr($node->name, 1, 3);
return (p == "cab") || (p == "CAB");


This will add nodes with names starting with CAB or cab (but not Cab or caB). If case is actually not important at all you can use simply


return $node->name ilike "CAB*";


Best regards,
Victor

Hi,

recommended database engines are PostgreSQL and TimescaleDB, depending on deployment size. For small deployments (few thousands of DCIs) you most likely won't get any significant performance gains. The larger your deployment is the more performance improvement you'll see.

Support for TimescaleDB in 2.2.17 was still experimental, and I would not recommend to do migration to TimescaleDB using 2.2.17. I would do as following:
1. Install new server and initialize TimescaleDB database
2. Upgrade old database to 3.0 using new nxdbmgr
3. Migrate old database to new one

We are planning to add in-place conversion from plain PostgreSQL to TimescaleDB but I'm not sure when it will be ready.

Best regards,
Victor

Sorry, I miss that first part with file names. Will check static builds, looks like this is issue specific to either static linking or non-UNICODE build. Usually you do not need static agents though - generic Linux agent (nxagent-3.0.2357-linux-x86_64.tar.gz) should work as well.

Best regards,
Victor

Please provide agent configuration file and configure options used for build.

Best regards,
Victor

You can mark interfaces em0.0 as "exclude from topology" if they does not have any valid addresses. I believe those are for stacking/clustering and not visible from outside anyway.

Best regards,
Victor

Currently it is not possible. If you don't mind please create feature request in our bug tracker for that.

Best regards,
Victor

So I see two unrelated problems here:

1. Duplicated 10.0.0.0/8 subnets - definitely a bug in the server. We have similar reports for duplicate subnets and will investigate that.

2. Node is placed into 10.0.0.0/8 subnet because address 10.100.7.1 is within that subnet. The question is why you have 10.0.0.0/8 in the first place - if it was created by server that means that at least one node has IP address 10.x.x.x with network mask /8 (or had it before). That looks like misconfiguration to me. I suggest to check all nodes within 10.0.0.0/8 for addresses with /8 mask, and if there is none just delete 10.0.0.0/8 subnet object (making sure that nodes are bound to some container or other subnets so they won't be deleted) and run forced configuration poll on them - it will create correct subnets.

Best regards,
Victor

Server generates event SYS_SCRIPT_ERROR for runtime errors in scripts, including instance discovery filter scripts, and also generates "warning" record in server's log. First parameter contains script identification string and second parameter contains error message. Unfortunately event is not generated if there is compilation error like in your case (although warning message in log still generated). I've fixed that in development branch, starting with next release you'll also get SYS_SCRIPT_ERROR event for instance discovery filter compilation errors.

On a side note, script syntax didn't change in that part - expression like

if (a < b) && (c < d) { }

was always a syntax error. Whole "if" expression should always be enclosed in parenthesis.

Best regards,
Victor

Hi,

I was able to reproduce this on my system. It's a bug in server. It is already fixed in development code and will be included into next patch release.

Best regards,
Victor