LXSERV Migration Status
=======================

ELFms Meeting, 23.08.2005

Front end machines (lxserv...):
-------------------------------

- All machines have been upgraded to SLC3 (LXSER01 - LXSERV06 (07))

1) Software:

   Web-Proxy:
 - SQID			German & Marco Emilio

   Accounting:
 - System-Accounting	Vlado

 ? Status of configuration


Back end machines (LXSERVB..):
------------------------------

LXSERVB02:

0) General:

 - Moved to building 613
 - Reinstalled with SLC3
 - Partition table:
   
   [root@lxservb02 root]# df
   Filesystem           1K-blocks      Used Available Use% Mounted on
   /dev/sda1             25806268   2309352  22186016  10% /
   none                   2016852         0   2016852   0% /dev/shm
   /dev/sda2              2063536     33184   1925528   2% /tmp
   /dev/sda9              1035660     26996    956056   3% /usr/vice/cache
   /dev/sda3            158855336    188700 150597188   1% /var
   AFS                    9000000         0   9000000   0% /afs

 ? Is this o.k? All data goes to /var 

1) Software:

   CDB:
 - CDB:			German (&Thorsten)
 - cdbop:		German
 - swrep:		German

   CDBSQL:
 -			German

   3ware:
 - CERN-CC-3dmd:	Jan (BIOS upgrade necessary, done for LXSERVB02)

   Backup:
 - TIVsm-BA		Thorsten (Harry)

   Load Balancing:
 -                      Vlado => done

   Notification Server:
 - 			Thorsten => done

   Registration Server:
 - 			Thorsten => done

   GPG Framework (to be replaced by SINDES):
 - SINDES		Vlado (with MARC) 

   Zephyr:
 - 			Vlado => done
 
   Power Queries:
 - 			Jan taking over from Tony C. ?

   Feedback:
 - 			Vlado   

   CERN-CC-lxserv_addon:
 - 			Jan: Still needed (Do we still use CCConfig)?


 ? status of configuration
 
Redundandencies:
----------------

How to implement h/w redundancy in the system:

a) For the front end nodes:

   - Web proxy:

     + SWREP


     + cdb-configuration



   - system accounting:



b) For the back end nodes:

   - CDB:



   - CDBSQL:



   - Load Balancing:



   - Notification Server:
     * Can run on both machines in parallel, uses ORACLE as a backend
     * The cleanup script should run on at most one machine
       (can be configured in /etc/sysconfig/notd_Srv)

     
   - Registration Server:
     * Can run on both machines in parallel, uses the name alias 'regis'
       to resolve if it is the master or the standby backup, in which case
       it will only update the web pages.

  
   - GPG/SINDES framework:



   - Zephyr:



   - Feedback: