1 Reply Latest reply on Aug 26, 2016 1:40 PM by aLTeReGo

    SAM 6.3 RC (NPM 12.0.1 RC, etc) feedback and questions

    antonis.athanasiou

      Hi,

       

      We have set up a lab environment with SAM 6.3 RC, NPM 12.0.1 RC, NCM 7.5.1 RC and NTA 4.2.1 RC:

       

      we really like the HA and Linux agent features and after reviewing it with my colleagues we came up with the following questions:

       

      HA Feedback

      1. Is it possible to get different email alerting options on the HA events? we tend to use the ${DefaultEmailTo} etc on the Global settings and this may cause some issues with current configurations
        1. what about implementing HA alerts in the standard Web alerting engine?
      2. Will there be a NTA FSDB HA feature too or that has to reside on different methods (VMware etc)
      3. Which IP is being used to poll devices (SNMP/ICMP etc)? vIP or the dedicated per polling engine?
        1. I can see this troubling network engineers with Firewall rules, ACLs, etc
      4. how (will) we do upgrades? Is the HA engine capable of reducing the outage during an upgrade?
      5. After some tests, I can see the Pool on some different statuses with no further information. what is a "Partially Operational" pool, "Disabled", "Critical", etc ?
      6. can we have more than 2 servers in a Pool?

       

      Linux Agent feedback:

      1. Asset inventory does not exist on some distributions (CentOS, Ubuntu) while polling with the Agent. Is this by design or...?
      2. Routing table disappeared from the list resources after switching from SNMP to Agent, even when we are using the "include SNMP credentials" option
        1. What stats are collected when dual polling  is enabled (Agent and SNMP) from SNMP?
      3. Python 2.4.3 or newer required for Linux agent to run, but not required when installing the agent? we failed to connect the agent to a Fedora box:
        1. even after re-installing the plug-ins from the following agent page,

       

       

       

      PS: Really really like how the installation process for APEs has become, very good idea to download all needed files from the Orion server!

       

       

       

      Prosperon - UK SolarWinds Partners

      Installation | Consultancy | Training | Licenses

      facebook_icon.jpglinkedin.pngblogger.pngtwitter-icon.jpg

        • Re: SAM 6.3 RC (NPM 12.0.1 RC, etc) feedback and questions
          aLTeReGo

          HA Feedback

          1. Is it possible to get different email alerting options on the HA events? we tend to use the ${DefaultEmailTo} etc on the Global settings and this may cause some issues with current configurations
            1. what about implementing HA alerts in the standard Web alerting engine?

           

          In this release of High Availability catastrophic failure notifications only utilize the default email settings. Other HA related alerts may be configured through the Web Based Alert Manager and use any email settings you wish.

           

          Will there be a NTA FSDB HA feature too or that has to reside on different methods (VMware etc)

          The following section of the NTA Administrator's Guide outlines how to protect your FSDB in a High Availability configuration. Orion HA is designed to protect the Orion application modules and additional polling engines only, not the database storage. Clustering on AlwaysOn Availability groups are recommended for MSSQL and Windows Clustering is recommended for FSDB.

          Remote NTA Flow Storage Database covered by Microsoft Failover Cluster - SolarWinds Worldwide, LLC. Help and Support

           

          Which IP is being used to poll devices (SNMP/ICMP etc)? vIP or the dedicated per polling engine?

            1. I can see this troubling network engineers with Firewall rules, ACLs, etc

          The only right answer is 'it depends'. Outbound communication from the HA pool, such as WMI or SNMP polling requests, may be sent by the primary or secondary server's IP address or the VIP address. All inbound communication, unless restricted by IP address, goes to the VIP address.

          The active pool member has a minimum of two IP addresses available: the IP address of the server and the VIP address for the pool.

          Since there are multiple IP addresses being bound to a single NIC, Windows chooses which IP address is used as the originating IP address. The IP address with the most high order bits that match the destination of the next hop will be used as the source IP address for all outbound polling activity.

          1. Convert the IP addresses to binary.
          2. From left to right, compare how many bits in the IP addresses match to the default gateway's IP address.

          The IP address with the most consecutive, matching bits is used for the HA pool's source IP address.

          If you want outbound communication to come from the VIP address, choose an IP address close to the default gateway's IP address. You can also modify the converted bits to be the IP address with the longest match and convert it back to an IP address.

          Examples of matching the binary bits

          The following is an example where the VIP will be used as the outbound IP address.

          IP addressIP address converted to binary# of matching bits
          Default Gateway (first hop)10.199.15.100001010-11000111-00001111-00000001
          VIP10.199.15.2000001010-11000111-00001111-0001010027
          Primary pool member10.199.15.6100001010-11000111-00001111-0011110126
          Secondary pool member10.199.15.6200001010-11000111-00001111-0011110126

          The longest match in the example above is the VIP as it has 27 consecutive matching high order bits to that of the default gateway.

          The following is an example where pool members' IP addresses will be used as the outbound IP address.

          IP addressIP address converted to binary# of matching bits
          Default Gateway (first hop)10.199.15.100001010-11000111-00001111-00000001

          VIP

          10.199.15.8200001010-11000111-00001111-0101001025
          Primary pool member10.199.15.6100001010-11000111-00001111-0011110126
          Secondary pool member10.199.15.6200001010-11000111-00001111-0011111026

          In this example, the longest match is the pool members' IP addresses. When a failover occurs, the IP address of the active Orion server is used as the source IP for all polling requests. The VIP address is only used for inbound traffic, such as syslog, SNMP Traps, NetFlow, and accessing the Orion Web Console.

           

          how (will) we do upgrades? Is the HA engine capable of reducing the outage during an upgrade?

          HA will not provide zero-downtime upgrades.

           

          After some tests, I can see the Pool on some different statuses with no further information. what is a "Partially Operational" pool, "Disabled", "Critical", etc ?

          Each pool member has its own status based upon the health state of that member. These can be influenced by availability (up/down), if that member has access to the SQL database server, etc.. The Pool itself also has its own status which is calculated as a roll-up (combination) of both members states. For example a 'Partially Operational' status means one pool member is down/unavailable but the 'Active' member is fully operational.

           

          can we have more than 2 servers in a Pool?

          In this release, the maximum number of members in the same pool is two.

           

          Linux Agent feedback:

          1. Asset inventory does not exist on some distributions (CentOS, Ubuntu) while polling with the Agent. Is this by design or...?

          Asset Inventory, Hardware Health, and SAM's SNMP Monitor require that NET-SNMP be installed on the monitored host. The Linux Agent will enable and configure NET-SNMP if needed, but if it's not installed at all, then these three features will not be available. No other agent functions are dependant upon NET-SNMP.

           

          Routing table disappeared from the list resources after switching from SNMP to Agent, even when we are using the "include SNMP credentials" option

          1. What stats are collected when dual polling  is enabled (Agent and SNMP) from SNMP?

          Routing is a function of NPM and is not available with the Linux Agent. It's also typically not applicable either unless you're using a Linux machine as a router. When SNMP is available, Asset Inventory, Hardware Health, and SAM"s SNMP Monitor will operate no differently than if the node was managed via SNMP.

           

          Python 2.4.3 or newer required for Linux agent to run, but not required when installing the agent? we failed to connect the agent to a Fedora box:

            1. even after re-installing the plug-ins from the following agent page,

          The agent itself is written in C++ but the plugins which provide the polling functionality are written in Python. If Python is not available the Agent itself will still install and run, but no plug-ins will be deployed until Python 2.4.3 or later is installed.  Once the Python requirement has been satisfied you can manually force the plugins to redeploy by clicking the 'Re-install plugin' link in your screenshot above, or by restarting the Agent daemon on the server being monitored.

          3 of 3 people found this helpful