
PRE-INSTALL REQUIREMENT:
	You need to have GNU tar in your PATH environment. This may be difficult
	for SUN users. If you have no way get a GNU tar, please go to:
	http://sunfreeware.com to download a copy.

HOW TO INSTALL:
1. Do "tar xvzf xxx.tgz" to unpack the distribution. It will create a 
   directory nameed "dist" in your current dir. 

2. cd to the "dist" directory and type "./setup".

3. "setup" will extract all the program files to the directory you specified.
   The software is self contained in the directory you specified during install,
   so that when you remove the software, it's very simple and clean.

4. Read the qstart.txt file in the distribution.

5. If you are to configure cluster with different OSes and CPUs, you need 
   to download each corresponding tar file onto each architecture, start the 
   Cluster Slave Daemon (Csd) on each cluster node. and start Clmd, the 
   Cluster Master Daemon on two servers--The second one is for redundancy. 

CONFIGURATION:
1. Read the qstart.txt file in the distribution.

2. The Cluster Master(CLMD) and the Cluster slave(CSD) can be setup in either
   Internet/Intranet environment.

3. To configure the software, you need basic DNS knowledge. There are 2 way you
   can setup the eCluster software. You can run it as a standalone DNS server or
   use the delegation method. To do the delegation setup, you need to create
   a NS record in your primary DNS server(either registered with 
   www.networksoluions.com if you want the cluster accessible from Internet 
   or your non-registered primary DNS server for Intranet use only) to point
   to the server running CLMD(This will be your Cluster Master Server). 
   If you don't host your own DNS, instead, via your ISP, you need to 
   contact your ISP to create the following DNS NS record entries to point to
   your Cluster Master. 

   If you choose delegation setup, in option part of your primary(Official) DNS
   named.conf, specify: 
		"rrset-order { 
			class IN type A name "www.yours.com" order fixed; 
			}; ".  
   This is for bind8.2.4.  For older version with ROUND_ROBIN defined in the 
   code, you will have to #undef that and recompile. for version older and 
   older, you won't worry this #undef thing.  Early version of bind 9 still 
   have rrset-order syntax not implemented, don't use these bind 9 yet.

	If the virtual group name in gpfile is www, then you will: 

	a). Create the NS record in your official DNS zone file on your primary
	    DNS Server as: 

		www IN NS eCluster_1 

   	b). The eCluster_1 should have an A record as well, as:

		eCluster_1 IN A 192.168.1.1

	c). Modify and add the zone in the /install_path/etc/ns.conf file.

	d). You will then create and scrach the zone file in the 
	    /install_path/etc/nsdb dir in the cluster master server
            running CLMD. In the file, remember to add a line at the bottom:

		$INCLUDE "cn/yourcompany.com.cndb"

	    CLMD will create the yourcompany.com.cndb file in the 
	    "/path/etc/nsdb/cn" directory. If you don't have this $INCLUDE, 
	    your cluster won't work.

	e). Specify the same key(password) in the clmd's configuration file, 
	    ("gpfile") the same as to the CSD's key. Make sure the port 
	    number in "gpfile" and CSD is using the same port number so that 
	    the two may communicate. CSD, by default, listens to port 10000.
	    To change it, do: /path/esdx start csd "-p port".

	f). To start csd, you need to type in a key in the /install_path/etc/key
	    file. csd will communicate with clmd and/or cfsd only when they're
	    using the same key and port.

	g). Start the server(refer to qstart.txt).

	TESTING:
	a). Do: nslookup www.yourcompany.com localhost on the CLMD Server. 
	    You should see 2 or more ip addresses bound to the domain 
	    if you have 2 or more node/IP listed in gpfile.

	b). Type: "/install_path/bin/snap" to view and check the cluster stats.

	c). Use" ps" to find out if the correct number of daemons and a single
	    nsd daemon are running. For CLMD, CFSD, NISD, the correct number of
	    daemons after started are N Node + 2(if the node is resolvable).
	    For csd, the default is to fork 5 children.

4. The /install_path/etc/ports/fallback.cfg file contains the ports that you 
   want to load balance and failsafe. The typical ports are HTTP(80) and 
   ftp(21) ports. fallback.cfg is the site port configuration file. You 
   may create file with a name as vgroup.domain.com.cfg for per group port 
   configuration. The file will be the same format as fallback.cfg. For 
   example, the Cluster www.company.com's port configuration file 
   will be www.company.com.cfg.

5. If you don't know DNS, please download and read the DNS (BIND) manual.
   The files you need to download are in the same directory that this file
   is located(ftp.xgforce.com/pub/eCluster/2X4). The 2 files are
   dns.ps and dns.me. The dns.ps is a Postscript version of the dns manual. To
   print it out, in DOS, "copy dns.ps prn" to a Postscript capable printer.
   In UNIX, print it directly to your Postscript capable printer. It's about
   46 pages. You may also use Ghostscript program to view the Postscript
   file and print it out. You may also convert the ps file into acrobat PDF
   file by using ps2pdf from the perl package.

   The file, dns.me, is a me version of the manual. Do "nroff -me dns.me|more" 
   to view the manual.          

   If you have difficulties or any questions about how to configure the eCluster
   software, send email to our support at: support@xgforce.com. We provide 
   free support for both evaluation and registered users.

CFS(tm) CONFIGURATION:
   To run CFS, you need to start CFSD at a server and have CSD running at each 
   nodes. Specify the fullpath of the directory that you want CFSD to 
   manage(VROOT) in CFSD's configuration file, cfsd.cfg(located in 
   /install_path/etc/cfsd.cfg).

   Make sure you use the same password(key) for both CFSD and CSD. To specify 
   the key for CSD, you will need to enter a password in /install_path/etc/key. 
   Othersise, CSD will not be able to start.

   Once the CFSD daemon started at server and CSD running at each nodes, you 
   can transfer files to the managed VROOT. All files, including file 
   permissions, directory changes, ownership changes will synced across all 
   nodes listed in "cfsd.cfg".

   Both CFSD and CLMD use CSD to communicate with the configured cluster nodes.

   To start CFSD, do: /path/esd/bin/esdx start cfsd "". The double quotation at
   end is for passing switches and is needed. To stop CFSD, do: 
   /path/esd/bin/esdx stop cfsd "".

   Notes: For SUN users running version less than Solaris 8 , you need to 
   install zlib into your /usr/lib directory. To download zlib, go to:
   http://sunfreeware.com.

NIS II CONFIGURATION:
   NIS II is a completely re-implimentation of the NIS. It's more secure than the
   old NIS and is a lot simpler than SUN's NIS+. It is designed to be used for
   SYSV styled UNIXes such as: SUN Solaris and LINUX, which used simple
   passwd and shadow files to contain user and group information. It could be
   use for FreeBSD but with great caution since FreeBSD uses passwd, shadow, and
   the pwd.db/spwd.db files. For FreeBSD, only one domain group is configurable
   since all file changes has to be located at /etc, not other arbitrary 
   directories.

   NIS II can be used to sync /etc/passwd, /etc/shadow, /etc/group,
   /etc/hosts, /etc/nsswitch.conf, /etc/master.passwd, /etc/spwd.db, and
   /etc/pwd.db files. All other files listed in the map will be pushed to
   the NIS II nodes at the same directory listed in the map file. For example,
   if you listed "/usr/home/support/.profile", and you don't have the
   home/support directory created in your NIS II nodes yet, the CSD running
   on NIS II client will create the missing part of the directory and push 
   the .profile file into that directory. You can also use this mechinism to
   sync sendmail.cf configuration files and aliases files, etc. If you have
   a UNIX network of 50 machines or more, this will greatly make it easy for
   ADMs to update their passwords and manange user accounts.

   NIS II uses nisd to read in the files listed in the map file and monitor 
   those files. Should file changes, it will push map files to its clients. 
   To run NIS II, you need to have nisd(running at NIS Master Server), 
   csd(running at each NIS nodes), and nisd.cf(nisd's configuration file). 
   For SUN users running version less than Solaris 8 , you need to install 
   zlib into your /usr/lib directory. To download zlib, go to: 
   http://sunfreeware.com.

   The main configuration file is nisd.cf. The format of this file is explained
   in the /installpath/etc/nisd.cf file.

   The www.map and ftp.map in the sample file /installpath/etc/nisd.cf are 
   called nis.map. nisd uses the nis.map file to locate where the files are 
   to be pushed to NIS II clients. All NIS II clients should run csd daemon 
   in order to make the passwd, shadow, group, nsswitch.conf, hosts updates 
   working.

   nis.map files are arbitrary file names of your choice. For its file format, 
   you may refer to the sample file www.map located in the /installpath/etc 
   directory.

   The nis.map file can be located at any file systems, but the files listed in
   nis.map file must be located at the root file system where the /etc directory
   is located. Otherwise, NIS updates won't work.

   For the full path listed in the nis.map file, the csd on the clients will 
   create the same full path directory used to store tmp files if it's not
   there.

   Once you have editted the nisd.cf file to list your NIS clients and created
   the nis.map file which lists the set of files to be pushed to NIS clients,
   you will then type: esdx nisd "", to start nisd. On each clients, type:
   esdx csd "" to start csd daemon. Make sure you have the same password for 
   both nisd and csd daemons. Also make sure they are using the same port to
   talk to each other. To test, edit the file listed in your nis.map, for
   example, the passwd and the shadow file, the changes will be propagated to 
   other nodes listed in that NIS II group. 

   If you want to chmod/chown and then push these changes to NIS II clients,
   you can simply do chmod/chown on the file and then touch the file. This will
   force a push to the files to all NIS II nodes listed in that NIS II group.

   To run nisd, do "/path/esdx start nisd "". esdx is a general shell program
   used to start/stop eCluster(tm) software daemons.


SUPPORT:
Email to support@xgforce.com. Free configuration and first level training 
is provided to both evaluation and registered users.

PORT:
Please refer to README.TXT for available port information.
