NetXMS Support Forum

Hi!

Server considers node as down if all it's interfaces is not reachable and node cannot be contacted by SNMP or via agent. How long a reboot was taken? If it was relatively fast, then it is possible that moment when node was completely unreachable fall between two status polls (each 60 seconds by default), and server never seen these two nodes as completely down.

If this is the case, solution depends on your goal - for example, if you wish to detect that node was rebooted, you can add monitoring of system uptime, and if it below some small value, like 100 seconds, that means that node was just rebooted and you can generate appropriate event.

Best regards,
Victor

Hi!

Did you turn off automatic layout (menu Map -> Enable automatic layout)?

Best regards,
Victor

Because you are monitoring services using DCIs, you will not get service down events in case of entire node down, because there will be no new values for DCI, and therefore no events will be triggered. Btw, it is not recommened to use SYS_xxx events in thresholds - they are generated by server in various situations, and you can get unexpected results. SYS_SERVICE_DOWN and SYS_SERVICE_UP generated by server during status poll of network services. Until you are not using network service objects, you are fine, but if you eventually configure some, you will get unexpected events. Better approach is to create your own events.

Best regards,
Victor

Hi!

Try to use GetSituationAttrubute(s, "status") instead of s->status - there was a bug report that latter method of accessing situation's attributes still not working, although I was thinking that I have fixed it for 1.0.8 - I still didn't check that myself (see this: https://www.netxms.org/forum/configuration/logwatch-function/msg4934/#msg4934).

Best regards,
Victor

Did you try console programs as well (something very simple like cmd /c echo test > c:\test.txt)?

As you are running Windows 2008 R2, it is 64 bit. Are you running 1.0.4 agent? If yes, it is 32 bit, and it may cause some problem (although it shouldn't). In this case, could you please try to replace agent with x64 version (it will be agent version 1.1.0, but it is compatible with 1.0.x servers).

Best regards,
Victor

Hi!

How many nodes you are monitoring? Could you please check status of pollers and queue sizes at the moment when node in question is down, but not detected yet (by running nxadm -i and then giving commands "show pollers" and "show queues")?

Best regards,
Victor

Quote from: EvilClown on January 11, 2011, 03:07:34 PM
1. Is it possible to manually connect servers in the map (Layer 2 and 3)?

Yes, if you are talking about custom maps. You can select two objects, and press Ctrl+Alt+C (or select Map -> Create link between objects menu) - this will create labeled link.

Quote from: EvilClown on January 11, 2011, 03:07:34 PM
2. How to integrate external service checks?

By using agent's external paremets. One example is here: https://www.netxms.org/forum/configuration/how-to-countgraphcheck-tcp-connection-states-table-with-netxms/. But this is only for creating new DCIs,  it has no relation to network service objects.

Best regards,
Victor

Teper' ponjal. S temi object toolami, kotorie est' sejchas, sdelat' normal'no ne poluchitsja - tol'ko esli sdelat' tool tipa "command", kotorij vizivaet nxevent dlja generacii special'nogo eventa, kotorij triggerit server action. Ja v sledujuschem relize dobavlju esche odin tip toola, eto neslozno.

Eto znachit chto v event processing policy ostalas' ssilka na ob'ekt, kotorogo bol'she ne suschestvuet. ID etogo ob'ekta 36. Poprobujte zapustit' nxdbmgr check pri ostanovlennom servere - on dolzen ubrat' reference na ob'ekt, a potom prover'te event processing policy - v kakom-to pravile budet nehvatat' source ob'ekta skoree vsego.

Hi!

Agent uses Windows API call GetTempPath to obtain directory for temporary files. It works as following (quote from MSDN):

The GetTempPath function checks for the existence of environment variables in the following order and uses the first path found:

   1. The path specified by the TMP environment variable.
   2. The path specified by the TEMP environment variable.
   3. The path specified by the USERPROFILE environment variable.
   4. The Windows directory.

Note that the function does not verify that the path exists, nor does it test to see if the current process has any kind of access rights to the path.

Could you please check values of mentioned environment variables and access rights for appropriate directories? Also, do you run agent under local system account?

Best regards,
Victor

Quote from: calin.sarac on January 08, 2011, 09:44:48 AM
One question please, as long as  I was not able to execute comands or scripts from Actions section in your program could be there any connection with the present situation ?  I tried a simple command like "cmd.exe /C "mplay32"" and even it seems that is fine is not working.I tried to put the same command in a bat file and add that bat file in actions and still not working even if everything is fine (I read all your post on the forum regarding this subject and follow the recomandations).

If you are running NetXMS server as a Windows service, you will not see any GUI for the programs started by the server. If you really wish to see GUI program started by the server, you should turn on "Allow service to interact with desktop" in NetXMS Core service properties. For simple tests, I recommend you to start simple program or bat file which writes something to some file, and check content of that file.

Best regards,
Victor

Hi!

As I can see from your config, you extract only one varbind from SNMP trap - with OID .1.3.6.1.4.1.2509.8.21.2.2.1.7. Value of this varbind will become event's parameter number 2, as you can see from column "No." in trap configuration's "Parameter buindings" list. But, in your event configuration and event processing policy you refer to parameter number 3 (by using macro %3 and variable $3 in script), which will always be empty string, because you don't extract anything else from trap. If you change $3 in your script to $2, it should start working.

Also, why you are using syslog parser? Do you receive anything via syslog as well?

Best regards,
Victor

Hi!

I don't understand what is the problem with Config::print. It uses wide character version of printf, so UNICODE build should print all characters correctly. Or you have problems with ANSI build?

Function StrStrip removes leading and trailing spaces and tabs from given string. It works in both ANSI and UNICODE builds. What exact problem you have?

Best regards,
Victor

Could you please send me screenshots of your trap configuration, event configuration, rule in event processing policy, and record in SNMP trap log?

Best regards,
Victor