2 Replies Latest reply on Mar 11, 2014 5:04 PM by visolara

    Monitoring Linux hosts - Disk, Memory, CPU - Help

    visolara

      I've spent the better part of this week learning, and researching how to configure my environment (RedHat/CentOS) in SolarWinds Orion.  There are a few things I can't seem to figure out.  All my work has been through the browser based management portal (host/Orion/SummaryView.aspx).  I do not have access to the actual windows based management application yet, which may be the root of my problem.

       

      First, let me clear some basic information;

      snmp is configured and working on my linux hosts, snmpwalk is successful, snmpdf from a remote host (linux to linux) is successful, the host is configured and shown in the Orion browser and self tests are successful.  From this I can confirm snmp communication between Orion and the host is working.

       

      Examples:

      (local test)

      [yaddayadda@hostname-02 snmp]$ snmpwalk -v1 -cpublickey localhost 1.3.6.1.4.1.2021.13.15.1

      UCD-DISKIO-MIB::diskIOIndex.1 = INTEGER: 1

      UCD-DISKIO-MIB::diskIOIndex.2 = INTEGER: 2

      UCD-DISKIO-MIB::diskIOIndex.3 = INTEGER: 3

      UCD-DISKIO-MIB::diskIOIndex.4 = INTEGER: 4

      UCD-DISKIO-MIB::diskIOIndex.5 = INTEGER: 5

      UCD-DISKIO-MIB::diskIOIndex.6 = INTEGER: 6

      UCD-DISKIO-MIB::diskIOIndex.7 = INTEGER: 7

      UCD-DISKIO-MIB::diskIOIndex.8 = INTEGER: 8

      (lots of additional omitted output) ...

       

      (server to server):

      [yaddayadda@hostname-01 ~]$ snmpdf -v 1 -cpublickey -Cu 128.12.8.128

      Description              size (kB)            Used       Available Used%

      /                          9296968         2578672         6718296   27%

      /home                      4128448          139364         3989084    3%

      /var                       4128448          274556         3853892    6%

      /tmp                        516040           16820          499220    3%

      [yaddayadda@hostname-01 ~]$

       

       

      Lets just focus on disk space monitoring, if I can get that working, the rest will follow.  If I go into the browser management tool, navigate to "Manage nodes", I can click on "net-snmp" and see my linux hosts.  By opening the "+" sign next to one of the hosts listed, I can see the file systems specific to this host it is monitoring.  This is all good, except it does not allow me to uniquely configure them beyond showing they are being monitored.  NOTE:  During the initial "Add Node" steps, it displays more file system options, I selected the meaningful ones and they are what I see being monitored.

       

      What I need is;

      Disk Space monitoring specific to the configured host.  This goes into thresholds, who receives the email alerting, etc.  One host may require critical alerting at 95% capacity, while another at 85%, and respectively email different groups.  From the browser based tool, I can not seem to figure out where to configure this.

       

      I have spent a good deal of time with the search function, and have found some helpful tidbits, but nothing that really helped me get the results I am looking for.

       

      Can anyone help me with this?

       

      Thank you!

        • Re: Monitoring Linux hosts - Disk, Memory, CPU - Help
          Leon Adato

          You have three concerns:

          1. thresholds
          2. alerting
          3. notifications

           

          For any other element (node, interface) I would tell you that the new NPM 10.7 custom threshold features is your friend. But for some reason disks (volumes) continue to be the red-headed step child of NPM (maybe rob.hock can explain why. I sure can't. )

           

          However, both issues can be solved with custom properties.

          Thresholds

          • Go to the custom property editor and get into the volumes section.
          • Create a custom property called "Disk_Crit". Make it a floating point type property.
          • For each disk that you want to have alert on something OTHER than the standard threshold, fill in this custom field with a number.

           

          Notifications

          • Still in the Volumes area of the custom property editor, create a custom property called "email" (or email_group, or whatever)
          • For each disk where the notification has to go to a different group enter the email address. If you need multiple addresses, just separate them by commas.

           

          Alerting

          NOW... on the polling engine, in the Advanced Alerts utility, you will create your alert as follows:

          • In the alert trigger, your logic will look something like this:

          Where ANY of the following are true

               Where ALL of the following are true

                    Disk_Crit is empty

                    Percent Diskspace Used is greater than or equal to 90

               Where ALL of the following are true

                    Disk_Crit is NOT empty

                    field Percent Diskspace used is greater than or equal to field Disk_Crit

           

          • NEXT... in the Alert Trigger action, create an email action
          • In the "Send To" box, insert the field ${email} (or whatever you called it

           

          Now you have a single alert that will obey both default settings (if disk_crit is blank) or a custom setting on a per-disk basis. AND it will notify whatever group you indicate in the ${Email} field.

           

          For more info on this subject, check out my talk at last years' Thwack Camp (Day 2, "Leveraging the Power of Custom Properties")

          http://thwack.solarwinds.com/groups/thwackcamp-2013