Print

Print


On 05/31/2013 04:35 PM, Winnie Lacesso wrote:

> If you would extract some general information in a timeslice, that would
> benefit myself&  anyone else who might in future go thru similar.

Here you are, some details may be missing but basically that's it.

---------------------------------------------------------------------------

Involved services :

     * xrootd manager and xrootd servers (~20)
     * Vmware virtual machines : BDII,VOBOX,CREAM(2),ARGUS server
       Torque server, DPM head+disk node
     * Worker nodes (~40)

Preparation

     A week before :
     * Notify the VOs
     * Declare GOC-DB Downtime from Day-1 17:00 to Day-2 17:00 (24h)
     * Notify NGI and all partners

     The day before Day-1 around 17:00 :
     * Drain CREAM-CEs (Alice jobs run for about 24h)
     * Prepare Quattor profiles
     * Prepare firewall filters for the new network segment
     * Create new virtual switch for the new network on VMware servers

Day-1 at 17:00

     Stop services
     * Stop xrootd services
     * Change IP of VMware virtual machines that support all services
       and stop them
     * Reconfigure the virtual network so that VMs are connected to the
       right network segment
     * Modify the DNS (modification will spread during the night)

Day-2 at 7:30

     Start services
     * Start xrootd servers, change IP, restart and start xrootd
       services (servers and manager)
     * Start service nodes : BDII,Torque server, CREAM, VOBOX, Argus,
     * Modify IPs of worker nodes and restart them (see comment below)
     Checks (many with the help of the NGI's nagios box)

Post-install

     * We had to give the new addresses of Torque server and CREAMs
       to the administrator of the MonBox since they filter the clients
       on the base of their IPs
     * Modify our local Nagios to take into account the new IPs

Comment : Here I do not give details on how to deal with an IP
modification for the Quattor managed worker nodes.

Comment : At the same time, we changed the NFS server used by worker nodes

---------------------------------------------------------------------------

So, you see, it was pretty straighforward for us...

JM


-- 
------------------------------------------------------------------------
Jean-michel BARBET                    | Tel: +33 (0)2 51 85 84 86
Laboratoire SUBATECH Nantes France    | Fax: +33 (0)2 51 85 84 79
CNRS-IN2P3/Ecole des Mines/Universite | E-Mail: [log in to unmask]
------------------------------------------------------------------------