What is this?

This knowledgebase contains questions and answers about PRTG Network Monitor and network monitoring in general. You are invited to get involved by asking and answering questions!

Learn more

PRTG Network Monitor

Intuitive to Use. Easy to manage.
More than 500,000 users rely on Paessler PRTG every day. Find out how you can reduce cost, increase QoS and ease planning, as well.

Free Download

Top Tags


View all Tags

Sensors in Windows failover cluster environment

Votes:

0

Your Vote:

Up

Down

Hi, there seems to be a problem within sensors in windows failover cluster environments. I configured some sensors (services, harddisk usage,...) on the cluster ip address (of the specified cluster services, NOT the physical node!) when i perform a failover to the 2nd node all sensors show failures as they cannot find the specified drives and services anymore. any ideas?

cluster sensors windows

Created on Jan 17, 2013 1:53:52 PM by  Matthias Jahn (60) 1



Best Answer

Accepted Answer

Votes:

0

Your Vote:

Up

Down

The logic and behavior of WMI sensors was massively changed with the PRTG Versions 17.1.28 and later:

Essentially, WMI connections (Which are DCOM based) are kept alive for as long as possible, preferably forever. A new connections (which will then re-resolve the virtual ip) will only happen if the connection is broken.

Created on Mar 28, 2017 8:38:08 AM by  Luciano Lingnau [Paessler]

Last change on Mar 28, 2017 8:39:30 AM by  Luciano Lingnau [Paessler]



10 Replies

Votes:

0

Your Vote:

Up

Down

Hello,

I'm afraid that can't be changed, due to the way the inner workings of WMI are implemented. The reason is, if the cluster switches, then even though the IP stays the same, the MAC address changes, and unfortunately WMI connections are linked to the MAC-Address. PRTG establishes a new connection every 60 minutes, and this is probably the time you had to wait. But this time will most likely not be changed, as establishing a WMI Connection is unfortunately horribly slow (over 20s), so we can't do that before every sensor scan.
We would recommend to monitor with "hardware-near" sensors directly on the physical nodes (not via the cluster-IP), and then use service-sensors (HTTP, SMB-Sensors, etc..) to check the clustered service.

Best regards.

Created on Jan 17, 2013 6:31:50 PM by  Torsten Lindner [Paessler Support]



Votes:

0

Your Vote:

Up

Down

Hello Torsten, i unfortunately can´t do that. Means I can not monitor the hardware itself because there are a couple of sql server instances and belonging windows services running. Each SQL server instance has an own ip address and an own hostname of course. So when the mentioned services are stopped on the first node and started on the second then ia have to monitor these processses on the second. and i can do this only by monitoring the ip address of the cluster services.

if i would use sensors to monitor these services on the machine itself i would get failures on that machine that is currently NOT hosting that services.

Created on Jan 18, 2013 7:39:23 AM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

@Torsten:

FYI: i just performed another failover and see that all services are monitored correctly but my "hardisk-usage-sensor" is still erroneous.

Created on Jan 18, 2013 9:16:28 AM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

The suggestion was, to monitor services, like the SQL via the Cluster IP, with SQL-Sensors (not WMI SQL Sensors, because they will face the same issue), and then all hardware sensors, like harddisk-usage on the physical nodes. With WMI, there is no other way, sorry. But that is not PRTGs fault.

Created on Jan 18, 2013 10:55:54 AM by  Torsten Lindner [Paessler Support]



Votes:

0

Your Vote:

Up

Down

I can not cover the harddisk-usage on the physikal disk because that disk is not part of that node anymore...

Created on Jan 18, 2013 11:08:45 AM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

as you wrote PRTG establishes a new connection every 60 minutes. i was waiting almost 120 minutes but the status of that sensor ist sill "erroneous"

any ideas? Thank you!

Created on Jan 18, 2013 11:13:47 AM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

the funny thing is when i replace the configured hostname with the cluster ip address all service states show errors again. placing the hostname again everything is green but the harddisk-sensor...

Created on Jan 18, 2013 12:42:37 PM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

OK, i got it. As mentioned in PRTG i had to use the drivletter and not the ID. Everything is working fine now :-)

Created on Jan 18, 2013 12:49:11 PM by  Matthias Jahn (60) 1



Votes:

0

Your Vote:

Up

Down

Changing the Hostname/IP actually also creates a new WMI Connection.

Created on Jan 18, 2013 1:00:47 PM by  Torsten Lindner [Paessler Support]



Accepted Answer

Votes:

0

Your Vote:

Up

Down

The logic and behavior of WMI sensors was massively changed with the PRTG Versions 17.1.28 and later:

Essentially, WMI connections (Which are DCOM based) are kept alive for as long as possible, preferably forever. A new connections (which will then re-resolve the virtual ip) will only happen if the connection is broken.

Created on Mar 28, 2017 8:38:08 AM by  Luciano Lingnau [Paessler]

Last change on Mar 28, 2017 8:39:30 AM by  Luciano Lingnau [Paessler]



Please log in or register to enter your reply.


Disclaimer: The information in the Paessler Knowledge Base comes without warranty of any kind. Use at your own risk. Before applying any instructions please exercise proper system administrator housekeeping. You must make sure that a proper backup of all your data is available.