NetXMS Support Forum

Hi,

I've installed NetXMS 2.0-M4 on CentOS 6.6 and have some feedback / issues - hope this helps.

Compared to release 1.2.17 I find it not so user-friendly to have to create nodes in the "Infrastructure Services" tree to force nodes discovered. That's currently the only way I found to force the discovery of some nodes. But then I cannot delete the node from the "Infrastructure Services" otherwise it disappears from the global "Network". Is there a way to proceed? I think there was a way to do it in a user-friendly way in 1.2.X. "Create Node", define IP and something like "Ping" or "Resolve". Why not available anymore?

Second, Many of my nodes have several IP addresses on a single NIC (alias / virtual IP). NetXMS 2.0-M4 does not seems to manage this well. Sometimes, it finds the physical IP, sometimes the virtual and sometimes the node hostname. Whatever it takes, it does not seem to be possible to harmonize the display and get hostname displayed.

Also, in 1.2.x the cluster related features were not optimal but seemed to work. Here as the alias / virtual IP are not well handled, it is not usable.

Also, when defining the "networks" of the cluster, if you enter an ip address that you then change, your network is automatically increased by one!!!

I regularly get put up with "An internal error occurred during "Get DCI values for node <IP Addr>", java.lang.NullPointerException.

In Management Console (not the Web based), the "Cluster" panel in "Object Details", the window showing the schematic of the cluster infrastructure is not properly sized. The bottom info about resources ownership is not displayed entirely and there is no roll bar. If you resize window then things appear.

For some strange reason, I had the "MTU" column filled in in the "Interfaces" panel of "Object details" for some Agent/SNMP enabled nodes but not all. And for those, the aliases where found. But that was under NetXMS 2.0-M2. The other columns seem to be correctly filled up.

Also in "Entire Network" nodes are discovered and labeled with an IP address that is not related to the network under which they are reported. Those nodes have an IP address in this network. But one would expect to get the node name as label or the IP address related to the network under which it is listed.
For example, a node has an IP address 172.10.25.11 and the virtual IP of that interface (cluster resource) is 172.10.25.10. That same node as another interface on 10.10.0.11 and the virtual IP for that side is 10.10.0.10. Well the node is then discovered as 172.10.25.10 under the network 10.10.0.0/24 and the Interfaces under "Object Details" show only the IP 172.10.25.11 and 10.10.0.11 not the virtual ones while they are used somehow for the labeling.

Do you have a clean way to empty the DB so that I can restart discovery from scratch and see if it helps?
That does not seem to be feasible through nxdbmgr tool.

Many thanks
Best regards,

Jdamien

Hi,

Is there any function similar to setDCIValue?

That could be useful while executing a script attached to an event policy to define the next value of a DCI based on event parameters and/or last DCI value (or some other DCI values) for example. Unless this can be done differently.

Br,
Jdamien

Hi,

Is there a way to define a data collection on a node that would collect data from another node?
This can be interesting when willing to poll a node for a data that is already available on another node (maybe becasue there is already a specific monitoring ongoing) which is much simpler / cheaper to poll than this other one (possibly because the link is congested or resources shall be reserved for service and not monitoring)

Many thanks
Jdamien

Hi,

I've installed new release 1.2.7 and agent as well on Linux CentOS.
Both agent and server logs do not show any error. When willing to define a collection data item, the list of possible parameters when "NXagent" is selected in "Origin" drop-down box is emtpy. Event those defined per default and those I defined while using previous releases are not shown! It seems the collection of older defined dat acollection item is still performed though.

Please advise on how to solve htis issue.

Many thanks
Best regards,


Jdamien

Hi,

We are currently trying to setup a HTTPS Reverse Proxy for accessing the Dashboard.
A server placed on a private network hosts NetXMS 1.2.5 and Dashboard. A Reverse Proxy in the DMZ is accessible via HTTPS only.

Currently, we see the client request on https://<public_IP>/nxmc/ forwarded to the internal server correctly as http://<private_IP>:8080/nxmc/. NetXMS internal server replies with a 302 (Page moved) with appropriate URL (http://<public_IP>/nxmc/nxmc?start=org.netxms.webui.core.nxmc) except that it is HTTP and not HTTPS. Do you have an idea how we could have HTTPS instead of HTTP?

FYI: when we provide directly the final URL (https://<public_IP>/nxmc/nxmc?start=org.netxms.webui.core.nxmc - note: HTTPS), everything works fine.

Many thanks in advance.
Best regards,

JDamien

As it is soon XMas, here are some requests for nice features:

• Get possibility to insert GPS coordinates as for a DCI (external script pushing the value in a specific formatted string) - is this already available?
• Possibility to zoom in/out in a Geo Map in the dashboard (possibly using a scale on the side as for GoogleMap / GoogleEarth...
• Ability to configure the line chart as fro the graphs produced in NetXMS console (right clik on the graph opens a drop down menu -> "Properties..."
• Especially the ability to zoom on an interval in the line chart would be very interesting
• Ability to define font size in label element in Dashboard
• IMPORTANT: Ability to remove the menus that are not accessible as per the access rights defined per user
• IMPORTANT: Ability to remove any information or link that is not accessible as per the access rights defined per user (Example: in Dashboard, in a node context, the command panel shall not be visible if user has only "Read" access)
• IMPORTANT: When providing access to a limited number of items (nodes, dashboards, services, etc) the Web GUI shall provide a structured tree of all those elements (Services, Dashboard, Nodes...) and not have them just listed one after the other in alphabetic order

Br,
Jdamien

hi,

Please, find here below some minor issues we found while using the management console / dashboard / Web GUI:

• Web GUI - Login window: need to type twice "tab" key to move from "login" field to "password" field
• Management console - DCI definition: it seems that the field "Description" is erased when selecting the "Parameter". So one shall FIRST select the "Parameter" then define the "description"
• Web GUI - Dashboard: pie chart title does not seem to be displayed even though the relevant check box is ticked in the config panel

Hi all,

I have to monitor some service / statistics (DCI) on a cluster which is composed of at least 2 nodes. In the Management console, it is clear how to create the cluster and the two (or more) nodes. The I have created the DCI on the cluster object (not the node) and assigned them to the virtual resource (active on the master). I see all DCI being replicated on each node with the ones on the master node being the only ones active and collected. So far, so good.

Now, in the Dashboard, I want to graph (line chart) the DCI related to the service. How do we do this? When i'm at the point where I have to select the data source for the graph, the cluster object does not provide any DCI, only the child nodes do. Do we have to select the DCI(s) on all nodes composing the cluster? In this case the legend shows twice each DCI...

I was expecting the defined DCI would be listed under the Cluster object and would report - when selected - the value collected by the node owning the virtual instance.

Could you, please, clarify where am I wrong? How shall I configure my system to perform as expected?

Currently, the DCI script reported '0' when running on the slave instance (where the service does not run) and 'real value' when running on the master instance (where service is active). When a fail-over takes place, I expect the 'real value' to be the one on the new master (former slave)...

Br,
Jdamien

Hi there,

When building a dashboard and defining the access rights for the different users, is it somehow possible to configure which menu would be visible depending on the user/login?
For example, for a dummy user, I would like him to have only access to the dashboard view / i.e. panel -> no menu, no side panel, nothing else than the defined dashboard.

Is there a way to do this?

BTW, some feedback on dashboard:

• It would be nice to have the possibility to set font parameters for the label that can be added in the dashboard.
• Also, the pie chart title does not seem to be displayed even though the relevant check box is ticked in the config panel.

Many thanks in advance

Best regards,
Jdamien

Hi,

I tried to install nxgentd package 1.2.4 on a CentOS 6.0 and I got the following errors:

line 84: [: too many arguments
line 93: [: too many arguments

and failed. The system is currently running nxagentd 1.1.10 installed using the package as well. The agent is stopped while upgrade is attempted.

Any idea?

thanks
Jdamien

Hi,

would you be so kind to give us some information about how the communication is setup between any Web GUI (remote laptop for example) and the NetXMS server?

Which port is effectively used? Wireshark shows only traffic to server port 4701. No traffic seems to be going on 8787.

How secure is the link between the remote laptop running the Web GUI dashboard and the NetXMS server?

We try to understand how likely it is that comm between remote dashboard and server will cross firewalls and how secure this comm can be.

Many thanks in advance.

Best regards,
Jdamien

Hi,

when you will find some time for "non critical" items, maybe you want to improve this small one:

When you use the MIB Browser and select a node to get a value, if you want to change node you have to quit MIB Browser and restart it.
Adding some flexibility by allowing to change all involved parameters (a.o. the IP address of the polled node) would be more user-friendly 


Best regards,

JDamien

Hi,

Using the new PostEvent functionality in a NXSL script, I managed to redirect alarms (warning and more severe msg) to appropriated nodes (kind of proxy). These messages are extracted from an application log file on the application server and tell whether a node has an issue or not w.r.t. this application.

Now I was wondering why (and if this is the expected behaviour) all messages appear correctly in the main "event log" opening from the icon tool bar menu and were not displayed in the "Alarms" tab of the specific nodes in the "Object browser"? I would have expected that I can have those messages (or at least the alarms) displayed in this tab.

Am I right or is there a difference between the "Event Log" window and the node specific "Alarms" tab?

Many thansk in advance
Best regards,

Jdamien

Hi there,

This is a low priority feedback.
When graphing a stat and zooimng in using the mouse to select an area, while x-axis adjusts accordingly, y-axis and related scale stays as it was. Would it be possible to have Y-axis dynamically adapted to the selected area to be zoomed.

If you have values from 1 to 100 and zoom only on an area where only values between 50 and 60 are present, you expect the y-axis to be tuned to 40-50-60-70, for example. Currently it stays to 0 - 100. Which may become very unconvenient when values are high.

Best regards,

Jdamien

Hi,

I'm getting closer and closer to what I want but still there sth missing  
Would be glad if you help me on this one.

So I come back to my story...
I have one service server which host a logfile and an NXAgentd. There is a LOGWATCH running for this logfile. Parsing each alarm, it is possible to determine which node is impacted by the error - not the server, some other node which is not hosting any NXAgentd.

I want to have a specific DCI defined and updated according to data in the error message for each particular node.
Using NXSL I can determine the node name and use the NXPUSH command. But I'm block at the time I need to identify the DCI and get its latest value (GetDCIValue + FindDCIValue). Why? Because in the NXSL script those two functions require the first argument (node) to be an object! In my context, the $node object refers to the service server and not to the node which name can be extracted from the error message.

Is there no way to get around this problem?

One may want to consult a DCI from a node when handling events related to other nodes.

I understand from other posts that one could use the FindNodeObject function to get the node object" once we know for example its name. But I persistently get error 14, saying that my left side (objDciNode->name) of the following expression is not an object:

objDciNode = FindNodeObject($node, vDciNode);
// Some debugging trace
trace(3, "Node name = " . objDciNode->name);

Is there sth I forgot?

typeof(objDciNode) == NULL but my node exists well!

Many thanks in advance for your help
Best regards,

JDamien

Hi,

It seems I have some problems to get my "Actions" executed.
I'm currently running server 1.0.6 with agent 1.0.6 on this server. Another node is running agent 1.0.4.

A simple action is defined in both nodes config file:
Server node:
SubAgent = /usr/local/netxms/lib/libnsm_linux.so
Action = testjd:touch jdtest.txt

Agent node:
SubAgent = /opt/netxms/lib/libnsm_linux.so
Action = testjd:touch jdtest.txt

SubAgent = /opt/netxms/lib/libnsm_logwatch.so
*LOGWATCH
Parser = /home/netxms/Parsers/Parsing_my_log.xml

Both agents show in logs that action is expanded and executed. But when listing the files in directory there is simply no file.
Action is triggered from an event which is triggered by a valid parsing of some text in the log file.

Action is defined to be executed on remote agent with name of remote agent given.
Even when I exchange the action config for executing it on server or remote agent with server name inserted, nothing happens.

There is no error in log files except the one related to DRBD... :
...
[09-Nov-2010 11:51:12] ParseIoStat(): new device added (name=dm-1 isRealDevice=1)
[09-Nov-2010 11:51:12] Unable to parse /proc/drbd, DRBD data collector will not start
[09-Nov-2010 11:51:12] Subagent "/opt/netxms/lib/libnsm_linux.so" loaded successfully
[09-Nov-2010 11:51:12] Subagent "Linux" already loaded from module "/opt/netxms/lib/libnsm_linux.so"
[09-Nov-2010 11:51:12] LogWatch: registered parser for file /home/daks/n2p_out.txt, trace level set to 0
[09-Nov-2010 11:51:12] LogParser: parser thread for file "/home/daks/n2p_out.txt" started
[09-Nov-2010 11:51:12] LogParser: file "/home/daks/n2p_out.txt" successfully opened
[09-Nov-2010 11:51:12] Subagent "/opt/netxms/lib/libnsm_logwatch.so" loaded successfully
[09-Nov-2010 11:51:13] Trying to bind on 0.0.0.0:4700
[09-Nov-2010 11:51:13] Listening on socket 0.0.0.0:4700

Thanks for any support.
Br,

Jdamien

Hi,


Is there a way to change the event source?
Here is the purpose of this request...

Let's assume a server for a particular service piles up information on different entities in some log file. You are only able to install NetXMS agent on this server, not on other entities. But through the log file you can identify some status information for each entity. Now I want to parse this log file, extract the alarms and depending on the entity identified through the parsed text, I want to generate an alarm for that particular entity. So if alarm is critical, the event shall popo up in the entity alarm list and not in the server (where the log file is located) event/alarm list.

also in the event log, the node name/address shall be the one of the entity and not the one of the server.

Is there an $event->source attribute or sth similar?

BTW.: I would appreciate if you could post the list of attributes for event and node if it has changed since last time. Any other objects that event and node? What attribute?

Many thanks in advance
Best regards,

Jdamien

Hi,

One small feature proposal, when working on the event list (for troubleshootign for exmaple or understanding the monitoring...) it would be quite nice to have the ability to stop the automatic scrolling. Indeed, whenever a new event pops up in the list, your cursor is automatically moved back to latest event. Then you have to scroll back to the events you were working on...

So stopping this automatic scrolling would prevent being push back to latest event when working in the history.

Btw, is there a way to remove events from the list? Export for backup and archiving, for example?
Automatic export in txt/CSV file every TBD (param) days - export performed at TBD (param) time of the day of any event older than TBD (param) days.

Best regards,

Jdamien

Hi there,

Simple question: I noticed while installing r1.0.3 that the DB stays with release 1.0.2 - using nxdbmgr.
Is that supposed to be the case? Shouldn't it be upgradd to 1.0.3?

nxdbmgr upgrade does not do anything.

Regards,

Jdamien

Hi there,

I've just installed your new release 1.0.2 and was looking at the new "Windows console: implemented custom graphs on Performance tab in Object Browser"
What exactly is coming in 1.0.2 for this? I can see perf graphs are presented a bit more esthetic... but what can you exactly customized? And how?

Many thanks in advance for this new release.

Rgds,

JDamien