NetXMS Support Forum

Спасибо Можно через PayPal или трансфер на банковский счет - пишите в приват, я пришлю детали. У нас раньше на сайте была кнопка donate, которая вела на PayPal, зря убрали похоже

Протестируйте с патчем из предыдущего поста - возможно эта проблема тоже исчезнет.

Оказалось, что это баг. В аттаче исправленный net.cpp из src/agent/subagents/portCheck - если заменить и пересобрать, то должно работать.

Так и знал что что-нибудь забуду Сделал вчера, скоро выложу обновление.

А как выглядит netstat -ant на хосте 192.168.0.126 после остановки апача?

Ona dolzna bit' propisana dlja vseh agentov, s kotorih proizvoditsja opros servisov. Server ispol'zuet lokal'nogo agenta dlja oprosa ob'ektov klassa "network service", esli ne ukazan drugoj poller node. Esli ze ispol'zuetsja DCI, to vse kak obichno - subagent dolzen bit' zagruzen na tom uzle, dlja kotorogo delaetsja DCI.

A portcheck.nsm zagruzen na sootvetstvujuschem agente?

Quote from: calin.sarac on January 14, 2011, 04:19:26 PM
Yes I tried, but nothing. Please take a look at attachements if you have time and tell me where I'm wrong.

In action configuration, backslash character (\) is an escape character, so if you need it in resulting string, you should use two backslash characters. So, in your case you must write


cmd /c "echo test > c:\\test.txt"


Best regards,
Victor

Hi!

The only way I could think of is to create dummy nodes (with IP address of 0.0.0.0) for each DCI group, create DCIs on these nodes instead of main node, and set "proxy node" for DCIs to original node. Then you will be able to give access rights to individual dummy nodes, without giving any access to main node. This solution has some drawbacks - you either should create two DCIs - on main node and on dummy node, or you will not have common list of parameters on single node - only spread across multiple nodes.

Best regards,
Victor

Quote from: Lexsus on January 22, 2011, 12:21:30 PM
Правда не вижу в Server Configuration вот этого параметра "AllowTrapVarbindsConversion"?
Параметр необходимо создать самому?

Да. По умолчанию используется значение 1, т.е. создавать его надо если необходимо получить поведение сервера как в предыдущих версиях.

Таблица MAC-адресов в ее нынешнем виде работает некорректно с некоторыми свичами. Сейчас я это меняю - сервер версии 1.0.9 уже умеет получать правильный FDB, но со стороны клиента пока используется старый вариант.

Hi!

Please post screenshots of your configuration (DCI configuration, thresholds, event processing policy) - it's hard to identify problem based on information you have provided.

Best regards,
Victor

Hi!

Personally I never try to monitor Windows servers via SNMP, but as far as I know, Windows SNMP agent does not provide performance data out of the box. You have to prepare it using Performance Monitor MIB Builder Tool (http://msdn.microsoft.com/en-us/library/cc723464.aspx#XSLTsection130121120120). I don't know if this still applies to Windows Server 2003.

Best regards,
Victor

Hi!

It's already implemented - if module's command handler returns NXMOD_COMMAND_ACCEPTED_ASYNC, main session processing thread will not delete message. Also, session's usage counter increased in this case, so it is guaranteed that session object passed to handler will be valid. When your module finishes async command processing, it must delete message and call method decRefCount() on session object. My code to start async command processing in the module looks like following:


typedef struct
{
void (*m_handler)(CSCPMessage *, ClientSession *);
CSCPMessage *m_msg;
ClientSession *m_session;
} PROCESSING_THREAD_DATA;

static THREAD_RESULT THREAD_CALL ProcessingThread(void *arg)
{
PROCESSING_THREAD_DATA *data = (PROCESSING_THREAD_DATA *)arg;

data->m_handler(data->m_msg, data->m_session);
delete data->m_msg;
data->m_session->decRefCount();
free(data);
return THREAD_OK;
}

static void StartCommandHandler(void (*pfHandler)(CSCPMessage *, ClientSession *),
  CSCPMessage *msg, ClientSession *session)
{
PROCESSING_THREAD_DATA *data;

data = (PROCESSING_THREAD_DATA *)malloc(sizeof(PROCESSING_THREAD_DATA));
data->m_handler = pfHandler;
data->m_msg = msg;
data->m_session = session;
ThreadCreate(ProcessingThread, 0, data);
}


and then somewhere in the command handler:


StartCommandHandler(GetSnapshot, msg, session);
return NXMOD_COMMAND_ACCEPTED_ASYNC;



Best regards,
Victor

Hi!

All DB functions located in libnxsrv in 1.0.x versions, and in separate library libnxdb in 1.1.x versions. They are defined in nxsrvapi.h (1.0.x) or nxdbapi.h (1.1.x). All functions require DB connection handle. You can either use global "default" handle (g_hCoreDB variable), or get handle from connection pool using function DBConnectionPoolAcquireConnection - this is the preferred method. Connections acquired from pool must be released with call to DBConnectionPoolReleaseConnection. Most of DB functions is straightforward. Two common patterns of usage are following:

1. Non-SELECT query: use DBQuery(handle, query); Function will return TRUE on success.

2. SELECT query: use DBSelect(handle, query); Function will return result set or NULL on failure. Then you can use DBGetNumRows(set) to get number of rows, DBGetValue(set, row, column, buffer, bufsize) (or any variations, like DBGetColumnLong) to retrieve data from set. When you finish, you must destroy result set by calling DBFreeResult(set);

In fact, there are audit log inside the server, it's just not accessible from management console. You can use server's function WriteAuditLog to add records to it.

Windows console is deprecated - I plan that 1.0.9 will be the last version with Windows management console as primary management tool. In 1.1.x branch, we are migrating to new cross-platform Java console. It is based on Eclipse, so it can be extended with plugins very easily. It's half-working already, and I hope to implement all functionality of legacy Windows console soon.

Best regards,
Victor