wiki:Internal/RunningInventory

Version 10 (modified by jkol, 9 years ago) ( diff )

added step to update inventory service cache before running topology scripts

How to Run Inventory

To run inventory preform the following:

From repository1
  1. Goto the tfptboot directory and make a symbolic link for the appropriate domain
    root@repository1:~# cd /tftpboot/pxelinux.cfg/
    root@repository1:/tftpboot/pxelinux.cfg# ln -s inventory-nb-grid 0A0A
    
HEX Domain
0A0A Grid
0A0B SB1
0A0C SB2
0A0D SB3
0A0E SB4
0A0F SB5
0A10 SB6
0A11 SB7
0A12 SB8
0A13 SB9
0A28 Outdoor
  1. goto the /export/nfsroot/etc/resolvconf/resolv.conf.d and make a symbolic link to head for the proper domain e.g. for grid:
    cd /export/nfsroot/etc/resolvconf/resolv.conf.d
    root@repository1:/export/nfsroot/etc/resolvconf/resolv.conf.d# ln -s head.grid head
    
    for sb1:
    cd /export/nfsroot/etc/resolvconf/resolv.conf.d
    root@repository1:/export/nfsroot/etc/resolvconf/resolv.conf.d# ln -s head.sb1 head
    
  2. If you are inventorying grid:

As root run the run_inventory script:

root@console.grid:~# cd /root/inventory/
root@console.grid:/root/inventory# ./run_inventory.rb

Starting Output will look like:

root@console.grid:/root/inventory# ./run_inventory.rb 
/usr/lib/ruby/vendor_ruby/restclient/response.rb:11: warning: method redefined; discarding old body
I, [2013-06-26T16:55:18.502861 #30052]  INFO -- : Database: Restfull DB connected to http://internal1.orbit-lab.org:5054/inventory/
=================== grid.orbit-lab.org  ====================
=================== First pass ====================
Turning off 400 nodes
Sleep for a minute
Turning on 400 nodes
Give it 900 to collect
.
.
.

This script will take 45 minutes to run.

If you are inventorying a sandbox:

  1. Turn on (or restart) all nodes, so the boot into the inventory pxe-image
  2. Check the inventory page for the sandbox you are inventorying, e.g. for sandbox9:
    http://internal1.orbit-lab.org:5054/inventory/attribute_list?set=node*.sb9.orbit-lab.org*
    
  3. Wait until all the nodes check in, then proceed to next step.
From repository1
  1. Break the link if this is not done the next person to use the grid will have a bad time.
    rm 0A0A
    
  2. Update the inventory service "domain topology" cache (which nodes belong to what testbeds) by visiting the following URL:
    http://internal1.orbit-lab.org:5054/inventory/update_cache
    
From internal1
  1. run the ruby scripts:
    root@internal1:~# cd /usr/share/omf-aggmgr-5.4/omf-aggmgr/ogs_inventory/
    root@internal1:/usr/share/omf-aggmgr-5.4/omf-aggmgr/ogs_inventory# ruby mktopo_cpu.rb
    root@internal1:/usr/share/omf-aggmgr-5.4/omf-aggmgr/ogs_inventory# ruby mktopo_dev.rb
    root@internal1:/usr/share/omf-aggmgr-5.4/omf-aggmgr/ogs_inventory# ruby mktopo_disk.rb
    
    Note: each ruby script takes some time to preform all the actions, and the omf-agg-manager restarts every even hour on the 11th minute. If the ruby process is running at that time it will fail.
Note: See TracWiki for help on using the wiki.