NetXMS Forum

Hi!

After changing the config fiel in netxms\etc to your specified settings the crashing problem has been fixed

What does this mean?

Hi

Here is your requested log file, I left a comment for you at the bottom

**************************************************************
[10-Jun-2008 11:28:08] Log file opened
[10-Jun-2008 11:28:08] Debug: Log file opened
[10-Jun-2008 11:28:08] Debug: Subagent API initialized
[10-Jun-2008 11:28:09] Subagent "WINNT.NSM" loaded successfully
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \System\Threads ==> \Systeem\Threads
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \System\System Up Time ==> \Systeem\Systeem ingeschakeld
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \System\Processor Queue Length ==> \Systeem\Lengte van de processorwachtrij
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \System\Processor Queue Length ==> \Systeem\Lengte van de processorwachtrij
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \System\Processor Queue Length ==> \Systeem\Lengte van de processorwachtrij
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(_Total)\% Processor Time ==> \Processor(_Total)\Percentage processortijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(_Total)\% Processor Time ==> \Processor(_Total)\Percentage processortijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(_Total)\% Processor Time ==> \Processor(_Total)\Percentage processortijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \PhysicalDisk(_Total)\Avg. Disk Queue Length ==> \Fysieke schijf(_Total)\Gemiddelde wachtrijlengte voor schijf
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \PhysicalDisk(_Total)\% Disk Time ==> \Fysieke schijf(_Total)\Percentage schijftijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(0)\% Processor Time ==> \Processor(0)\Percentage processortijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(0)\% Processor Time ==> \Processor(0)\Percentage processortijd
[10-jun-2008 11:28:19] Debug: WINPERF: Counter translated: \Processor(0)\% Processor Time ==> \Processor(0)\Percentage processortijd
[10-jun-2008 11:28:19] Subagent "winperf.nsm" loaded successfully
[10-jun-2008 11:28:19] Counter set B is empty, collector thread for that set will not start
[10-jun-2008 11:28:19] Counter set C is empty, collector thread for that set will not start
[10-jun-2008 11:28:20] Unable to bind socket: Elk socketadres (protocol/netwerkadres/poort) kan normaal slechts één keer worden gebruikt.


===========	Victor: Bottom line of this log says:   Every socket address (protocol/networkadress/port) can only be used once. ===========

I am rebooting right now to test the default counter config setting

@victor said in Agent service crashes, delaying boot times, BIG problem:

Sorry for delay, I was quite busy Could you please do some additional tests?

1. Comment out winperf.nsm subagent - I suspect that it can be the cause of crash.
2. If agent service crashes without winperf.nsm, run nxagentd.exe from command line with -D switch - it will cause agent to log additional debug information.

Best regards,
Victor

Here are some tests I ran for you:

Config file change:

#
# NetXMS agent configuration file
# Created by agent installer at Tue May 13 14:14:16 2008
#

MasterServers = 192.168.150.9
LogFile = {syslog}
FileStore = C:\NetXMS\var
# SubAgent = winperf.nsm

Time from login screen to usable system: 2.40 Minutes 

Agent completely disabled, time: 2.05 Minutes

Standard settings: 3 minutes

Winperf renamed and taken out of config: 2.30 minutes

Placed config in root to test nxagentd -D : 3.4 minutes

System.uptime: Gives value of 574  (on for 5 minutes ~ )
system.threads: gives value of 386

NetXMS agent crash is gone after I make sure winperf doesn't start but slow start up isn't fixed by this.

Hope this gives you enough info!

Hello!

I would really appreciate any updates 

Hi, I use this feature as well so maybe I can be of some assistance.

1. Download "NetXMS agent installer" from https://www.netxms.org/download/
2. Download the appropriate NPI file for your agent installer
3. Go to Control Panel - "agent packages"
4. Right click, choose "install"
5. Browse for appropriate NPI file
6. Agent package should be added to package manager now
7. Right click on the package you want to distribute
8. choose "deploy"
9. Select the nodes you want to upgrade by holding CTRL key

That should do the trick 

Yes, that works fine.

Two DCI'S

net.interface.bytesin(65539)  :  Incoming traffic , this value is in bytes
net.interface.speed(LAN)        :  Total interface speed,  this number is in bytes (converted it from bits to bytes successfully)

I want the incoming traffic DCI to display the network load procentually instead of showing the bytes.

The transformation on incoming traffic DCI is as followed:
$1 / GetDCIValue($node, FindDCIByName($node, "Net.Interface.Speed(LAN)")) * 100

This transformation doesn't work, it just gives me an output of "0"

What am I doing wrong?

Hi, any updates you can give me on this problem are very welcome 

Allright

On all the workstations I am monitoring this parameter is being read out properly, however on two of my servers this parameter just wont work properly.

I'm sure I have the right HDD ID but it never polls (timestamp stays at 1970)

This is happening on both an x86 and an x64 windows server 2003 server.

Thank you!

I use the IP address, it's impossible to use the host name in my setup (Set the firewall to block DNS to NetXMS section)

Bump, any idea?

On all the machines that I am monitoring, startup times are quite long

Yesterday I found out that the NetXMS agent is causing this problem, basically the agent service is started but it immediately crashes, after some time it starts up again and everything is fine. The readouts are also fine.

This problem is causing a two minute delay (or longer) on boot times for some of my more slower machines 

I am using version 0.2.21, I haven't yet tested on X64 machines but most likely the problem occurs there as well. So far I have only tested windows XP machines for this problem, I will test vista machines as well if you need that infromation.

Config file:

#
# NetXMS agent configuration file
# Created by agent installer at Tue May 13 14:14:16 2008
#

MasterServers = [censored]
LogFile = {syslog}
FileStore = C:\NetXMS\var
SubAgent = winperf.nsm

Here are all the syslog reports (not running an english OS so I translated some things)

1st one

Type of event:	Information
Source of event:	NetXMS Win32 Agent
Category of event:	None
Event-ID:	3
Date		28-5-2008
Time:		9:04:57
User:		none
Computer:	Censored
Description:
Subagent "WINNT.NSM" loaded successfully

2nd

Type of event:	error
Source of event:	Service Control Manager
Category of event:	None
event-ID:	7022
Date:		28-5-2008
Time:		9:06:32
User:		none
Computer:	[censored]
Description:

The NeTXMS agent-service has crashed at startup.

20 Second after that 5 other syslog events regarding the agent are created with the following descriptions:

3d:

Counter set B is empty, collector thread for that set will not start

4th:

Subagent "winperf.nsm" loaded successfully

5th:

Counter set C is empty, collector thread for that set will not start

6th:

Listening on socket 0.0.0.0:4700

7th (23 seconds after error or so):

NetXMS Agent started

This is a very big problem as you can understand, if you need any other info please ask.

One of the groups in my object viewer has the "unknown" status symbol, I can't seem to figure out why this is.

It's reachable by NetXMS, status polling is enabled, none of the checks are giving me any problems.

Polling the status manually doesn't change the status either

Your third option sounds the best Giving the specific interfaces that you do not want to monitor on nodes an attribute which would stop their "network range" from being created

--------------------------------------------------------------------------------------------------
This is an example of what would work nicely:

I add a router with two ports, one LAN port and one DMZ port, the LAN port runs at 192.168.130.* and the DMZ runs at 192.168.210.*

Now after the router has been added, both ranges are created in the object viewer.

I select the DMZ port of the router and select the option "Ignore for topology" , now I delete the 192.168.210.* range from NetXMS, I do a configuration poll on the router and the DMZ port is not added again.

Allright, here goes

--------------------------------------------------------------------------------------------
DCI 1: BLACK TONER (this one works)

TYPE: SNMP
OID  : .1.3.6.1.4.1.2001.1.1.1.1.100.3.1.1.3.1
DATATYPE: Integer
INTERVAL: 3600

THRESHOLD:

LAST POLLED VALUE will be LESS OR EQUAL than 10
If true execute: X_TONER
If false exectue: X_TONER_NORMAL

GRAPH (it gets turned off a lot so I selected the data it did collect):

--------------------------------------------------------------------------------------------

DCI 2: CYAN TONER (this one works as well)

TYPE: SNMP
OID  : .1.3.6.1.4.1.2001.1.1.1.1.100.3.1.1.3.2
DATATYPE: Integer
INTERVAL: 3600

THRESHOLD:

LAST POLLED VALUE will be LESS OR EQUAL than 10
If true execute: X_TONER
If false exectue: X_TONER_NORMAL

GRAPH (it gets turned off a lot so I selected the data it did collect):

--------------------------------------------------------------------------------------------

DCI 3: MAGENTA TONER (This one does NOT workl)

TYPE: SNMP
OID  : .1.3.6.1.4.1.2001.1.1.1.1.100.3.1.1.3.3
DATATYPE: Integer
INTERVAL: 3600

THRESHOLD:

LAST POLLED VALUE will be LESS OR EQUAL than 10
If true execute: X_TONER
If false exectue: X_TONER_NORMAL

GRAPH (it gets turned off a lot so I selected the data it did collect):

--------------------------------------------------------------------------------------------
DCI 3: YELLOW TONER (This one does NOT workl)

TYPE: SNMP
OID  : .1.3.6.1.4.1.2001.1.1.1.1.100.3.1.1.3.4
DATATYPE: Integer
INTERVAL: 3600

THRESHOLD:

LAST POLLED VALUE will be LESS OR EQUAL than 10
If true execute: X_TONER
If false exectue: X_TONER_NORMAL

GRAPH (it gets turned off a lot so I selected the data it did collect):

--------------------------------------------------------------------------------------------

I have set the threshold really high on the DCI'S, I was testing to make sure that an alert would be triggered, I tried putting them on an interval of 10 seconds, I disabled all DCI'S but the ones I was testing.

I hope you see something wrong because I can't find the problem 

Great, thank you

Hi Victor,

I recently read one of your posts saying that this method is not the correct way to deal with this problem, do you have any other suggestions on how to fix this? Unfortunately I'm going to encounter this problem a lot so I would like to be sure that the solution is stable.

Wow this is getting really weird.

Four different DCI'S, for four different toner statistics, all of them are integers, all of them have the same threshold, and all of them trigger the same events

But only two of them, actually trigger the event when a treshold is reached, I tested this by disabling the toner DCI'S that I wasn't testing at that time.

Do you have any idea what might be causing this?

Also, what is the best database software to run NetXMS server on?

I've been using MySQL for quite some time now but would MSSQL perform even better or.. ?