Microsoft SCOM monitoring for PowerMax using Unisphere SNMP traps

Microsoft SCOM monitoring for PowerMax using Unisphere SNMP traps

I hope you liked my prior blog on NTFS allocation units and PowerMax storage provisioning and performance best practices. Looking forward to more comments and suggestions for future contents, so please keep them coming!!

Lately I have been hearing about Microsoft Systems Center Operations Manager (SCOM) based monitoring for PowerMax. And even though around for many years and integrated with many monitoring tools, SNMP traps based monitoring for PowerMax storage system using SCOM is fairly unexplored area so far. I have received some questions on various distributions recently about the topic, so I thought of trying that out. Let me share my experience based on the success of this deployment. It will certainly provide ease of use and control to the system administrators by incorporating PowerMax monitoring alongside other data center resources from single interface of SCOM. 

SCOM is around for many years as a cross-platform data center monitoring system for operating systems and hyper visors. SCOM monitoring console can provide single interface that shows alerts, health and performance information for diverse set of resources including servers, network switches, routers, among others. SCOM includes many management packs to provide standard set of monitoring for all those devices with pre-defined rules and views. SCOM uses SQL Server as the back end for collected information to provide historical charts, trends and analytics. SCOM works with Windows, UNIX and Linux servers and supports a variety of authentication mechanisms to provide single pane of glass management for large set of data center resources. Along with many different data collection mechanisms, SCOM also supports collection of SNMP traps from managed devices and processing them using various management packs and scripts for better analysis and reporting.

Unisphere for Dell EMC PowerMax (and its family of storage systems) offers full feature set for storage provisioning, performance and REST API based management for SAN and NAS based infrastructure. Unisphere for PowerMax also provides notifications and alerts via e-mail, sys log and SNMP trap mechanisms that can be leveraged by third party infrastructure monitoring products.

I will go over setting up data collection using Unisphere for PowerMax SNMP traps on Microsoft SCOM. SCOM by design does not allow collection of SNMP traps from Windows servers so Windows based Unisphere can’t be used for this purpose. Also using PowerMax eMgmt container is fairly locked down so it would not allow a key discovery piece needed for the setup on SCOM. There are some workarounds to avoid discovery but to make things simple, this time I will cover deployment using Linux based Unisphere for PowerMax.

High level overview of the steps needed for the SNMP trap based monitoring

Here are the steps needed on both Unisphere for PowerMax and SCOM for SNMP traps:

  • Allow communication on specific UDP ports between SCOM and Unisphere for PowerMax server 
  • Setup name or IP address resolution of Windows SCOM server by Unisphere server
  • Configure SNMP on Unisphere for PowerMax for SCOM server
  • Discover device on SCOM for Unisphere for PowerMax server
  • Create and install Unisphere management pack on SCOM using MIBs supplied by Unisphere
  • Configure rules and views for SNMP traps on SCOM
  • References and Further Readings

The following sections will describe all these steps in greater detail.

UDP ports used by SCOM and Unisphere for PowerMax server

UDP port 161 is used by SNMP service  on Unisphere server to send SNMP discovery information and UDP port 162 is used on SCOM server to receive SNMP trap information. These ports should be opened through firewall to ensure that SCOM can collect SNMP traps from Unisphere for PowerMax.

On SCOM: Allow ports 161 and 162 for inbound and outbound traffic using Windows firewall advanced settings.

On Linux server:


Name and/or IP address resolution on Unisphere for PowerMax server

Make sure that Unisphere server can reach SCOM server. Enter the host/IP address of SCOM server to /etc/hosts file on Unisphere server for proper name resolution if needed.

SNMP configuration on Unisphere for PowerMax for SCOM server

This step requires enabling SNMP configuration on Unisphere for PowerMax for the desired storage system and specifying the objects and conditions for monitoring and sending alerts.

(1)   SNMP configuration: On Unisphere for PowerMax select “Settings”, and under “Alerts” section select “Notification” to configure SNMP traps to send to SCOM.

Setup SNMP trap destination as shown below by giving SCOM IP address and port number. In this example, I am using SNMP version 1 but Unisphere also supports SNMP version 3 if needed for higher security. After entering IP and port, enter “OK”, close and re-open the screen, to “TEST” the connection.

Also make sure that desired PowerMax system has appropriate system level and performance level notifications enabled as shown below.

(2)   Select alert policies for the objects and conditions for sending alerts: Select “Alerts” and make sure that desired alerts have SNMP enabled for notification under “Alert Policies”.

Unisphere for PowerMax Server device discovery on SCOM

SCOM can collect SNMP traps from a device only after that device is first discovered on SCOM. Unisphere for PowerMax server does not really accept any discovery request from SCOM as it is designed to send just the traps to the specified destination.  To allow for discovery by SCOM, we first need to install a tool like Net-SNMP on Unisphere for PowerMax Linux server. Once the server is discovered by SCOM, we need to create a “Run as Account” to run Unisphere for PowerMax server discovery by providing the IP address and port number to use for the discovery.

On Linux server:

(1)   Install Net-SNMP:

(2)   Setup Net-SNMP in /etc/snmp/snmpd.conf file with proper community string. Typically, most SNMP devices use “public” as the community string but here we need to use the community string “SNMP_trap” as that is required by Unisphere for PowerMax.

(3)   Restart SNMPD after this

On SCOM server:

(1)   Create a “Network Device Discovery” rule in SCOM. For discovery, first create a  “Run As Account” using “SNMP_trap” as the community string. On discovery wizard, specify IP address and port # of Unisphere server. Choose explicit discovery and on “Advanced Discovery Settings” change defaults as specified to avoid timeouts for some long running operations.

(2)   Now discovery can be either scheduled or run manually. For our purpose, we will just choose “manual” discovery as we just want to run it once to capture some basic information from Unisphere server.

    (3)   Using tool like Wireshark, we can monitor the packets exchanged between SCOM and Unisphere server during the discovery. As we can see SCOM sends multiple SNMP “Get” requests and Unisphere server with Net-SNMP utility responds with various OIDs supported by the server. Output below also shows the version number and community string “SNMP_trap” sent by Unisphere.

    (4)   Once SCOM completes the discovery it will show various parameters for Unisphere servers. Note down SNMP version and System OID highlighted below as they will be used in subsequent steps for some changes that are needed for successful deployment.

 

Creation and installation of Unisphere management pack on SCOM using MIBs supplied by Unisphere

We now need to create a custom management pack with Unisphere specific MIBs so that SNMP traps received from Unisphere can be parsed and reported appropriately. This step requires a SNMP MP Generator tool from Microsoft that allows parsing of third party MIB and creating management pack for the device specified. This tool allows extending basic capabilities of SCOM to diverse set of devices with different capability profiles.

(1)   Install SNMP MP generator tool on SCOM server. This is a standalone tool and it can be installed on any other server as well. Download SCOM – Extensible Network Monitoring Management Pack Generator tool from: https://www.microsoft.com/en-us/download/details.aspx?id=54083

To use the tool, create an initial XML specifying the some of device discovery information from Unisphere server. As we can see below “Device Name” refers to the Unisphere server name, SysOID refers to SysOID found after Unisphere server discovery by SCOM. Also set the Id and Name to the values that you want to use for the management pack being created.

In SNMP MP generator console, open the existing project by selecting the above XML file. And choose the MIBs for Unisphere for PowerMax along with some standard MIBs specified below. Contact Dell EMC customer support to provide the MIBs needed for this step.

Once the MIBs are loaded SNMP MP generator tool populates the structure with all MIBs used by SNMP traps from Unisphere for PowerMax.

At this point a new Management Pack can be generated that uses the OID structure populated above. As we can see original sample XML file with just basic device specific information is now updated with rich set of configuration in the newly created management pack.

     (2)   Import the newly created management pack to SCOM


    (3)   SCOM will now show the management pack just imported

Configuration of rule for SNMP traps on SCOM

At this point we have SCOM updated with all the definitions needed to properly collect SNMP traps from Unisphere for PowerMax and process them. We will now create a rule to catch all SNMP trap based events.

Create a catch all SNMP trap rule with rule type of “SNMP Trap (Event)” that will collect all the OIDs from Unisphere for PowerMax. Use the management pack we created above to store the rule so that our management pack is augmented with new rule and associates all the properties to that rule.

This event collection rule should have the target type of “Node” which is the network device – Unisphere for PowerMax server that we have discovered.
Leave Object Identifier field as default that means rule will catch all the OIDs received from the SNMP trap.

Check the generated rule in SCOM. Keep the run as profile as default which means it will use “SNMP_trap” account we used during discovery.

Unisphere for PowerMax uses SNMP version 1 but as you may have noticed from device discovery screen, Net-SNMP used version 2c. SCOM by default only accepts the version number that matches with device discovery. So we need to update our management pack so that it can also allow collection of traps for version 1. To do that we need to first export our management pack as XML.

Open exported XML file and do a couple of changes as shown below – update the version number of the management pack and remove the “Version” specifier from the “Rules” section of the XML. Removing this line would allow SCOM to collect SNMP traps from version 1 as well.

Save the XML with these changes and re-import the management pack and make sure that the updated management pack version is now listed.

Configuration of Event View on SCOM to display the events collected from the SNMP traps

At this point SCOM has the rule to collect SNMP traps from Unisphere for PowerMax. Desired views can now be created to properly aggregate and group all the traps collected.

Once view is created SNMP traps from Unisphere for PowerMax will be populated in that view. If you don’t see any traps received run “Test” on SNMP configuration on Unisphere for PowerMax console and run Wireshark on SCOM to ensure proper connectivity and package delivery. As shown in the example below after successful deployment a device allocation related trap sent by Unisphere was collected by SCOM and reported in the event view.

Conclusion

Unisphere for PowerMax has nice SNMP trap based notification mechanism that can be integrated well with Microsoft Systems Center Operations Manager (SCOM). Such integration would allow centralized monitoring for PowerMax resources along with all other infrastructure services and components from single user interface of SCOM.

References and further readings

(1)    Microsoft Systems Center Operations Manager (SCOM): https://docs.microsoft.com/en-us/system-center/scom/welcome?view=sc-om-2019

(2)    Microsoft SNMP MP Generator Toor: https://www.microsoft.com/en-us/download/details.aspx?id=54083

 

 

 


Comments

Popular posts from this blog

NTFS allocation units and PowerMax storage provisioning and performance considerations