This page is not yet up-to-date...

Each farm needs to have a Tao NameServer running on the oprserv machine. The LoggingManager (LM) registers itself with this NameServer.

If you're asking, "How do I know when I have to reset the name server?", believe me, you'll know.
You'll get bombarded by emails. If you start the jobs on a farm without NameServer running, the farm will
get stuck at the state of "GetOprOutputSocket", then the farm comes to FatalError.

Start the NameServer (SLAC)

1. Log on to the OPR server machine using the group account.
2. Set the root path:
   setenv TAO_NS_ROOT /u1/TAO
   (you will probably find this variable already defined, please check). If this directory does not exist, create it.
3. Start the nameserver:
   go to the production area prod#
   do an srtpath
   startNamingService
4. Check that it has started:
   look for the process 'Naming_Service' with top, to see if there is a process called "Naming_S"if there are already process running on the machine, usually, you will have a hard time to see if the process is there or not,
   you can use the command:

   ../bin/$BFARCH/OprCmd.pl -iUser,Farm -noprserv0X ps [PC|ER]X

   to check if the process is there or not.

   check the timestamp on the TAO_NS.ior file in /u1/TAO.

TAO Commands

You can look at what is registered in the NameServer and kill the LM by disconnecting it from the NameServer.

• Setup everything to see the right NameServer:

setenv NameServiceIOR `cat /nfs/oprserv0x/u1/TAO/TAO_NS.ior`
or use the alias ns#
srtpath <release> (where release is the current production release)
• Check what is registered:
  TaoNSDumper
• Disconnect a registered LM (causes the LM to crash):
  OprLMInterrupter XX#/LM/<run#>-<cb>
  (e.g. ER5/LM/11334-1)
• Remove an entry from a dead LM:
  TaoNSUnbinder XX# LM/<run#>-<cb>
  (e.g. ER5 LM/11334-1)