DX Unified Infrastructure Management

Have you ever wanted to make a change to one setting on all disk configs but don't want to delete the existing config and re-apply some new default because you have a lot of minor customisations out there. Some drives have been deactivated, many thresholds have been tweaked etc.

Here is a script that can change just one setting in your deployed cdms without altering anything else. I ran it direct from the nas lua script editor. It did give a timeout in the GUI but the thread kept on running and the job got done. It could be run from the scheduler so it wrote to nas.log, or with a little work run under sdk/nsa.

Sorry about the indentation. the cut and paste lost all my leading spaces, yuck My re-format may have introduced minor errors, caveat emptor. 

-- change some disk settings without re-discovering and losing existing settings

-- put the setting you wish to change in here and the desired value
-- this example sets disk usage in MB active
key = "qos_disk_usage"
value = "yes"

-- set the param pds for getting cdm configs
local getarg = pds.create()
pds.putString(getarg, "name", "cdm")

-- Find hubs
local list = nimbus.request("hub", "gethubs")
for _,hub in pairs (list.hublist) do
   print ("Finding Robots on "..hub.name)
   local list,rc = nimbus.request(hub.addr, "getrobots")
   if rc == 0 then
     for _,robot in pairs (list.robotlist) do
       print (" ",robot.name)
       local cdm, rc = nimbus.request(robot.addr.."/controller", "probe_config_get", getarg)
       if rc == 0 then

         print (" Got cdm config from ",robot.name)

         for section,rec in pairs (cdm) do
            local changes = 0
            if rec.disk ~= nil then -- is this a disk 'root' section?
               if rec.qos_disk_usage == value then
                  print (section, " key: ",key," already set to ",value)
               else
                  local setargs = pds.create()
                  pds.putString (setargs, "name", "cdm")
                  pds.putString (setargs, "section", section)
                  pds.putString (setargs, "key", key)
                  pds.putString (setargs, "value", value)
                  _, rc = nimbus.request(robot.addr.."/controller", "probe_config_set", setargs)
               if rc == 0 then
                  print (section," changed OK")
                  changes = changes + 1
               else
                  print (section," change failed")
               end
            end
         end
         if changes ~= 0 then -- restart the probe
            sleep (500) -- give it a moment to write
            nimbus.request(robot.addr.."/cdm", "_restart")
         end
        end
      else
         print (" Failure getting cdm config from ",robot.name)
      end

    end

  else
     print (" No response or bad response from this hub. rc = ",rc)
  end
end

How would I modify this such that I can change the disk thresholds for currently monitored disks?  For example:

Based on your script, it's just looking for the key of qos_disk_usage... In my case the key of active=yes is almost everywhere within the probe.  I just want to change it under the section for the drive letter and the threshold under that drive letter section.

<disk>

   interval = 15 min

   samples = 4

   <alarm>

      active = yes

      <fixed>

         <E:\>

            active = yes

            description = File system E:\

            disk = \Device\HarddiskVolume3

            percent = yes

            qos_disk_usage = no

            qos_disk_usage_perc = no

            inode_percent = yes

            qos_inode_usage = no

            qos_inode_usage_perc = no

            delta_percent = no

            delta_calculate_all = yes

            delta_type = both

            qos_disk_delta = no

            <error>

              active = yes

               threshold = 10

               message = DiskError

            </error>

            <warning>

               active = no

               threshold = 20

               message = DiskWarning

            </warning>

            <inode_error>

               active = no

               threshold = 10

               message = InodeError

            </inode_error>

            <inode_warning>

               active = no

               threshold = 20

               message = InodeWarning

            </inode_warning>

            <missing>

               active = yes

               message = DiskMissing

            </missing>

            <delta_error>

               active = no

               threshold = 10

               message = DeltaError

            </delta_error>

            <delta_warning>

               active = no

               threshold = 8

               message = DeltaWarning

            </delta_warning>

         </E:\>

What prevents you from doing this with a CFX? I believe with the "edit" flag/verb you should be alright.

-jon

For existing file systems, you need to know the name of the drive path. The script does it for all known file systems in the existing config it appears.

I do something similar myself. On top of this we use user_tag_1 to define the purpose of the server. That way one can easily identify the list of servers to do a task like "change disk HWM to 97% for all SQL servers"

-Garin

Yeah true. I guess for windows "simpe" disks like C: D: etc cfx is alright. But if you have mountpoints etc, it'll be harder to include all disks.

Right, nothing prevents me from using a cfx.. The issue arises where I have a system with 3 monitored disks(where there may be 5 on the actual box)  Deploying the cfx to clear the fixed section forces a re-discovery and will then add ALL disks into monitoring.  So I want to basically say for anything that is currently monitored, ensure there is a high and low threshold set for the disk at specific thresholds, and then thus that requires a special cfx for each box (given the monitored drives) so it knows which disks to modify.