Installing replication features in DSL pool
			-------------------------------------------

1. Check out the latest version from CVS's V6_7-Technion_ha5-branch

2. Configure it, compile the version and put the binaries inside /usr/local/condor/glibc23/condor-6.7.14

3. On both ds-ibm1 and picasso machines:
	* In /etc/init.d/condor comment out the below line
	CONDOR_VERSION=6.7.13 
	and add the following line
	CONDOR_VERSION=6.7.14

	* In /etc/profile.d/condor.sh comment out the below line
	CONDOR_VER=6.7.13
	and add the following line
	CONDOR_VER=6.7.14

	* In /etc/profile.d/condor.csh comment out the below line
	set CONDOR_VER="6.7.13"
	and add the following line
	set CONDOR_VER="6.7.14"

these are necessary for startup scripts/various profile definitions of sysadmins

4. In /usr/local/condor/etc/{ds-ibm1,picasso}.local files inside ds-ibm1 
add the below lines at the end of the file:

	RELEASE_DIR = /usr/local/condor/glibc23/condor-6.7.14

	REPLICATION = $(SBIN)/condor_replication

	REPLICATION_ARGS = -p 61450

	HAD_USE_REPLICATION = true
	MASTER_NEGOTIATOR_CONTROLLER = HAD
	
	REPLICATION_LIST=picasso.cs.technion.ac.il:61450,ds-ibm1.cs.technion.ac.il:61450
	NEGOTIATOR_STATE_FILE=$(SPOOL)/Accountantnew.log
	REPLICATION_INTERVAL=25
	HAD_ALIVE_TOLERANCE=150
	MAX_TRANSFER_LIFETIME=10
	NEWLY_JOINED_WAITING_VERSION_INTERVAL=5
	HAD_UPDATE_INTERVAL=300

	REPLICATION_LOG      = $(LOG)/ReplicationLog
	TRANSFERER_LOG       = $(LOG)/TransfererLog

	MAX_REPLICATION_LOG = 64000000
	MAX_TRANSFERER_LOG  = 64000000

	REPLICATION_DEBUG    = D_FULLDEBUG
	TRANSFERER_DEBUG     = D_FULLDEBUG

	DC_DAEMON_LIST = MASTER, STARTD, SCHEDD, KBDD, COLLECTOR, NEGOTIATOR, EVENTD, HAD, REPLICATION
	DAEMON_LIST = MASTER, COLLECTOR, NEGOTIATOR, HAD, REPLICATION



			Problems during installation
			----------------------------

There was a problem of creating a transferer process. The error that appeared in the replication daemon log
was as follows:
	* Create_Process: child failed because PRIV_UNKNOWN process was still root before exec()

The problem appeared because Condor does not allow a process be created by user, the real uid of which
is the root's one (i.e. Condor does not use 'geteuid' function, it uses 'getuid' function instead)
This is exactly what happened in DSL pool, because Condor processes ran under effective user, 
called 'condor', whereas the real user was 'root'.



			Solution
			--------
1. Add NIS group "condor" with GID='id -n condor'
2. Change condor's default group to "condor"
3. Rebuild NIS database.
4. Chown -R condor:condor /home/condor
5. Change condor init script to run the Condor daemons with 'condor' user by

	 su - condor -c /path/to/condor_master