RADE

Global Architecture

The SCM product supports a client/server architecture. Two types of hosts are involved in the execution of a SCM command:

• The client machine: any command is executed locally on the user's host and connects to the SCM server host(s) to retrieve information
• The server machine: at least one machine is dedicated for managing SCM repositories and where SCM daemons are running.

When a command is running, it is connected with several daemons running on one or several distant hosts (SCM servers).

Fig 1: Single Server Architecture

The figure 1 shows the location of processes involved in the execution of one SCM command running on a client host: Connections are established with different daemons depending on the information to be registered or retrieved The "File Contents Server" process manages its own repository. The "Monitor" process manages data in SCM database. Data for configuration management are stored in database. The same database can be shared with the "Monitor" process.

The global architecture can be more complex depending on the size of software projects and the number of people involved in. It is possible to have several machines on which several File Servers and SCM databases processes have been installed. Whatever the number of databases or File Servers is, the way commands work is the same, the complexity is often just seen by the SCM administrator (the end user is not aware of it).

Fig 2: Multiple Servers Architecture

The figure 2 shows that several File Servers and databases can be accessed by a single command. This can occur when the command is executed from a multi-tree workspace whose trees are managed in different databases and whose associated source files are registered in different File Servers. There is only one "Monitor" process running on one server because its purpose is to control operations on all databases.

[Top]

Databases

There are two kinds of databases managed by SCM:

• SCM databases
• File Server databases.

A SCM database is used for storing any data that are not source files. It is created by the Oracle or DB2 administrator and is known by SCM commands using its SID (SID = System Identifier = a unique name for an Oracle database instance). One or more databases can be used depending on the amount of data and the number of users to manage.

Multiple databases can be used for storing either:

• Monitor process' data
  There is only one database of this type, known as the monitor database
• Site information (1 database )
  There is only one database of this type, known as the central database
• Configuration management specific data
  There can be several databases of this type, known as departmental databases

The File Server database is a repository managed by a FileServer and whose purpose is to store any user file. This kind of database is initialized by the SCM administrator using the File Server program.

Depending on the number of servers available and their characteristics, it is possible to install on a same host one SCM database and one File Server but having several File Servers or several SCM databases on a same host is not interesting.

[Top]

Trees

(Workspace) Trees are the first "objects" to be created when starting with SCM. A tree is created by specifying:

• A name
• A database name
  It will contain Configuration Management data related to objects managed in workspaces belonging to this workspace tree
• A host name and a port number identifying a File Server
  It will manage all source files for workspaces belonging to this workspace tree

Any combination of File Servers and databases is possible. The following figures show how several trees can be managed. One important thing to know is that multiple databases or File Servers should be installed on different hosts, providing better performance.

Fig 3: Single File Server / Single Database	The figure 3 shows three trees managed in a single database and whose associated sources are gathered in a single File Server repository. This is an usual way of working with SCM for either creating trees for different small projects (files and users) creating trees for managing different versions of the same projects
Fig 4: Multiple File Servers / Single Database	This architecture is interesting when managing large projects and for which a strong SCM database has been set up on a specific hardware (disk space, CPU, RAM, etc). As the amount of source files can be very big, it could be interesting to use different File Servers for managing source files associated with different workspace trees. Of course, those File Servers should run on different hosts (for performance reasons). It is also possible to install a new File Server (on a new host) for managing source files for new trees if the host(s) where the other File Server(s) run is (are) heavily loaded or if there is not enough disk space left.
Fig 5: Single File Server / Multiple Databases	This architecture is allowed but does not provide any particular feature.
Fig 6: Multiple File Servers / Multiple Databases	The figure 6 shows two databases containing some trees and for which multiple File Servers have been set up. This architecture allows to manage big projects by dispatching different projects in different databases on different servers. For instance, each division may have its own SCM server with one SCM database and one File Server. This architecture does not forbid the use of multi-tree workspaces for working in all databases at the same time (for instance for gathering developments at entreprise level).

[Top]

Accessing Administrative Commands

Commands are available in search path after executing the tck_profile command with an appropriate TCK. Two TCKs are generated when installing a CD-ROM [1]: one for all users and one for the SCM administrators only. This second TCK gives access to administrative commands like:

• adl_lock_base, adl_unlock_base
• adl_ls_contents_server, adl_mk_contents_server
• adl_mk_tree, adl_ren_tree, adl_rm_tree, adl_set_tree
• adl_mk_root_ws
• adl_monitor, adl_mo_admin
• adl_file_server, adl_fs_admin
• adl_upgrade_db

For instance, after having downloaded a CAA V5 Tools CD-ROM under the path "C:\Program Files\Dassault Systemes\T04D28", execute the tck_profile command and list the available TCKs:

C:\> "C:\Program Files\Dassault Systemes\T04D28\intel_a\code\command\tck_init.bat"
C:\>tck_list
SCM_adm_V5R5_T04D28 SCM_adm_V5R5_T04D28
V5R5_T04D28 Mkmk_SCM_V5R5_T04D28 SCM_V5R5_T04D28
C:\>

The "V5R5_T04D28" TCK corresponds to the tools brought with the CD-ROM and the "SCM_adm_V5R5_T04D28" TCK gives access to SCM administrative commands. Here is an example of the traces produced when calling the tck_profile command.

C:\>tck_profile SCM_adm_V5R5_T04D28
------------------------------------------------------------------------
SCM Variables:
----------------------------
ADL_LEVEL=5
ADL_DIFF_LOCAL=N
ADL_DIFF_LOCAL_PATH=C:\DownloadADLV5
ADL_PATH=C:\Program Files\Dassault Systemes\T04D28\intel_a
------------------------------------------------------------------------
User log cleaner started
User log cleaner ended
C:\>

SCM commands are now available in path.

[Top]

Creating Databases

Before creating databases, people must have checked

• how many organizations are going to use the Workspace Manager
• how many people will be involved in development projects
• what is the target hardware for installing SCM and File Server databases

Thinking about these topics should lead to choose a global architecture from the ones described in Trees section.

[Top]

How Many Databases Are Needed?

We advise to start with a single SCM database and a single File Server database. It is not needed to set up several databases at the beginning except if SCM is intended to be used by hundreds of developers. Depending on the number of SCM users the SCM database is created with different characteristics (initial size of tablespaces, size of extents, etc). Three types of database can be set up: small, medium and large:

• small database: for managing a small organization (up to 20 developers) and 2 or 3 workspaces per user. The number of files should be under 5000 files.
• medium database: for managing between 50 and 100 developers and up to 20000 files.
• large database : for bigger organizations or for managing lot of files.

Is it possible to add new databases for supporting new teams?

Yes it is possible. The method would be to let the first created database as the monitor database and let the new databases be the departemental databases, that is to say by creating new trees in the new databases.

[Top]

SCM Databases

The SCM initialization is performed once an empty database (or instance) has been created. Please refer to the Oracle or DB2 documentation for this task.

To complete the initialization, you must connect to the new database as administrator (DBA) and perform the following tasks:

1. Create the SCM account and possibly the associated profile
   This account will be used both in step 2 and when installing a new SCM version that needs to upgrade existing databases
2. Create the SCM tables.
   There are three kind of tables to be created depending on the futur use of the database being initialized:
   1. Tables for the "Monitor" process
   2. Tables for site information
   3. Tables for configuration management data (trees, workspaces, components, etc)

The three kinds of tables can be created in a single database as well as in different databases but

• Tables for the "Monitor" process should be created in only one database
• Tables for site information should be created in only one database (often the same as the one used by the Monitor).
  This database is known as the central database (cdb).
• Tables for configuration management data are created in each database whose purpose is to store workspace trees.
  Such databases are known as departmental databases (dept)

The creation tasks are performed by SQL programs that are generated by the adl_admin_site command. Here is the command lines to execute:

E:\> adl_admin_site -init_db -db_user xxx -db_type monitor -db_size medium -target_dir c:\temp
E:\> adl_admin_site -init_db -db_user xxx -db_type cdb -db_size medium -target_dir c:\temp
E:\> adl_admin_site -init_db -db_user xxx -db_type dept -db_size medium -target_dir c:\temp

As shown above the adl_admin_site command must be executed three times to get the SQL programs for the three kinds of tables. Those programs are generated under the directory specified by the -target_dir option.

E:\> dir C:\TEMP\*.sql
 Directory of c:\temp

06/10/00  16:07                 25 978 CreateTablesCdb.sql
06/10/00  16:07                 66 736 CreateTablesDept.sql
06/10/00  16:03                  3 874 CreateTablesMonitor.sql
06/10/00  16:07                    262 DropTablesCdb.sql
06/10/00  16:07                  1 446 DropTablesDept.sql
06/10/00  16:03                    151 DropTablesMonitor.sql
06/10/00  16:07                    695 EstimateStatisticsCdb.sql
06/10/00  16:07                  3 390 EstimateStatisticsDept.sql
06/10/00  16:03                    387 EstimateStatisticsMonitor.sql
06/10/00  16:07                  3 209 InitInstanceCdb.sql
06/10/00  16:07                  3 209 InitInstanceDept.sql
06/10/00  16:03                  3 209 InitInstanceMonitor.sql
              12 File(s)        112 546 bytes

Step 1: the InitInstanceXXX.sql program is the first one to be executed and must be executed only one time on a given database, i.e. if only one database is used for managing monitor, site and configuration management tables, only one of InitInstanceCdb.sql, InitInstanceDept.sql and InitInstanceMonitor.sql programs will be used.

Step 2: each of the CreateTablesXXX.sql programs must be executed on the database that will manage XXX information. If a single database is used, the three programs must be executed on it.
Notice: several tables are duplicated between databases and then the SQL programs can generate errors when a single database is used.

Step 3: the EstimateStatisticsXXX.sql programs must be executed regularly (weekly for instance) to keep performances to be the best.

Note: the DropTablesXXX.sql programs are provided just for cleaning up a database before restarting its creation. They must not be used in a normal database initialization process.

[Top]

SCM Databases Connection

By default, the connection to SCM databases is done with "adl" user and "adl" password. To use another user and/or password, execute the adl_admin_site command:

E:\> adl_admin_site -init_db_connect -db_user user -db_password password

This command stores the user and the encoded password in the site profile. So, this command must be executed after the installation on each machine.

[Top]

File Server Data

Prerequisites: "Monitor" and site information tables must have been created and the Monitor started.

File Server data are composed of two root directories into which the File Server will format its own repository files. Here are the steps to follow to create a new file Server database.

Step 1: declare the new File Server in the central database by the adl_mk_contents_server command

• Specify the name of the host where the File Server will be launched
• Specify the port number on which it will be listening
• Give a brief description of its purpose (for information only)

E:\> adl_mk_contents_server -host tartampion -port 23451 -desc "file server for CAA dev project"
Creating the contents server with the host name "tartampion" and the port number 23451 Command successful.

Step 2: log on the host where the File Server will be launched

• Use the SCM administrator identity
• Create two directories on one (or two) disks local to this host (reliability is not ensured when using network file systems)

E:\> mkdir E:\FileServer\FSFiles
E:\> mkdir E:\FileServer\FSData

• Format these directories by starting the File Server using the -format option
  1. Take care that the -format option must be used only when initializing a new File Server since any data under the two directories will be lost
  2. Specify the first directory by using the -files option
  3. Specify the second directory by using the -data option
  4. Specify the port number by using the -port option (as registered at step 1)

E:\> adl_file_server -files E:\FileServer\FSFiles -data E:\FileServer\FSData -format -port 23451
Formatting storage area. The format procedure is successfully terminated. E:\> 

Step 3: when started with the -format option, the File Server stops after few seconds. The corresponding repository is then ready to store source files. Read the "How to start it" section to know how to start the File Server.

[Top]

The SCM Daemons

There are two types of daemons running on SCM servers: the Monitor and one or several File Servers.

The Monitor

Purpose

The Monitor is the first program to be run since it controls all operations performed in all databases (SCM, File Server).

Its purpose is

• to log every commands run on the site
• to manage concurrent access on objects like trees, workspaces, images
• to trigger recovery mechanism after a command crash

[Top]

How to Start It

Log on the host where the Monitor will be launched and change to the SCM administrator identity.

Set up the current environment using the TCK for administration and execute the adl_monitor command.

Note that the '&' character can be used on Unix platform for running the process as a background process. It is not possible to run it the same way on Windows platform. SCM provides the adl_admin_site command for registering a service for that purpose.

[Top]

How to Stop It

From any host and being the SCM administrator, set up the current environment using the TCK for administration and execute the adl_mo_admin command with the -stop option. This option causes the Monitor to wait for the current active commands to finish and then stop.

E:\>adl_mo_admin  -stop
# ADLMo - 0100: Monitor will stop at the end of the currently processed commands

The -abort option can be used for stopping the Monitor by aborting any running command.

E:\>adl_mo_admin  -abort
The monitor is aborting

Please, consult the "Monitoring Commands" section for learning more about the adl_mo_admin command.

[Top]

Using Windows Services

On Windows platform, the adl_admin_site command provides the -create_service for registering a new service to launch a SCM daemon (Monitor, File Server or Multi-site Transfer Manager) on the host where the service is declared.

To declare a service for the monitor, log on the SCM server having system privileges, use the TCK for administration to have access to the adl_admin_site command and execute the following command line:

E:\>adl_admin_site -create_service -service_type monitor -start_at_boot

The -start_at_boot option is used for specifying to the service system manager that the given service must be started at boot time.

Starting The Monitor Service

Fig 7: Control Panel/Services

The adl_admin_site command just declares a new service but does not start it. To start the monitor, use the Windows service tool. This tool is found by opening the "Start/Settings" menu and selecting the "Control Panel" item. Then run the "Services" programs and find the service named "SCM Monitor V5R5_T04D28" where V5R5_T04D28 is the TCK corresponding to the monitor installation.

Fig 8: Starting the Monitor Service

To start the Monitor service, select the corresponding line and click on the "Start" button: a popup window appears while the service is starting the service status changes to "Started" when the service has been started The way the service is started (manual or automatic) can be changed whatever the service was registered using the adl_admin_site command. The startup type can be also changed to "Disabled" if you want to prevent the system (or anyone) to start the Monitor for a while.

Only one service can be created on a given host for a given daemon and a given installation.

[Top]

Stopping the Monitor Service

Fig 9: Stopping the Monitor Service

Open the "Services" program like when starting the Monitor service, select the line corresponding to the monitor and click on the "Stop" button: a popup window appears while the service is stopping the "Started" status disappears

If a File Server service has been registered on the same host and has been started, the "Services" program will display a message indicating the File Server will be stopped too.

[Top]

The File Server

Purpose

The File Server is a process that receives requests from SCM commands for storing or extracting files.There is no direct user commands for requesting it, all requests are performed by workspace commands.

One File Server can store files for one or several workspace trees wherever these trees are managed (in one or several SCM databases, on a single or different servers).

[Top]

How to Start It

Log on the host where the File Server will be launched and change to the SCM administrator identity.

Set up the current environment using the TCK for administration and execute the adl_file_server command:

1. Specify the File Server directories by using the -files and -data options
2. Specify the port number by using the -port option

E:\>adl_file_server -port 23451 -files E:\FileServer\FSFiles -data E:\FileServer\FSData
Recovery start.
Processing REDO log.
REDO log processed.
Recovery stop.
Server start.

The process is then ready to server requests.

Note that the '&' character can be used on Unix platform for running the process as a background process. It is not possible to run it the same way on Windows platform. SCM provides the adl_admin_site command for registering a service for that purpose.

[Top]

How to Stop It

The adl_fs_admin command is used for monitoring File Servers. It provides the -stop and -abort options for stopping a File Server.

From any host and being the SCM administrator, set up the current environment using the TCK for administration and execute the adl_fs_admin command with the -stop option. This option causes the File Server to wait for the current active commands to finish and then stop.

E:\>adl_fs_admin  -host algue -port 23451 -stop

The -abort option can be used for stopping the File Server by aborting any active transaction.

E:\>adl_fs_admin  -host algue -port 23451 -abort

Note: even if the adl_fs_admin command finishes, the File Server may take few seconds to stop because some sub-processes must stop first before the main process can stop itself. The adl_fs_admin command can be used for testing whether a File Server is running or not:

E:\>adl_fs_admin -host algue -port 2340 -ping
While opening a connection with "algue:2340" (host:port_offset).
# ADLSOCK - CONNECT: The connection was rejected.
Check network and server machine with "ping" or "ftp" command.

[Top]

Using Windows Services

On Windows platform, the adl_admin_site command provides the -create_service for registering a new service to launch a SCM daemon (Monitor, File Server or Multi-site Transfer Manager) on the host where the service is declared.

To declare a service for the File Server, log on the corresponding SCM server having system privileges, use the TCK for administration to have access to the adl_admin_site command and execute the following command line:

E:\>adl_admin_site -create_service -service_type fcs -port 23451 -fs_files E:\FileServer\FSFiles -fs_data E:\FileServer\FSData -start_at_boot

The -start_at_boot option is used for specifying to the service system manager that the given service must be started at boot time.

The parameters of the -port, -fs_files and -fs_data options must be those specified when initializing the File Server repository (see File Server Database).

If the File Server service is registered after having registered the Monitor Service, a dependency is set up between the two services that will cause the Monitor to be automatically started (if it is not already started) when the File Server is being started.

[Top]

Starting the FCS Service

Fig 10: Control Panel/Services

The adl_admin_site command just declares a new service but does not start it. To start the File Server, use the Windows service tool. This tool is found by opening the "Start/Settings" menu and selecting the "Control Panel" item. Then run the "Services" programs and find the service named "SCM File Contents Server V5R5_T04D28" where V5R5_T04D28 is the TCK corresponding to the File Server installation.

Fig 11: Starting the FCS Service

To start the FCS service, select the corresponding line and click on the "Start" button: a popup window appears while the service is starting the service status changes to "Started" when the service has been started The way the service is started (manual or automatic) can be changed whatever the service was registered using the adl_admin_site command. The startup type can be also changed to "Disabled" if you want to prevent the system (or anyone) to start the File Server for a while.

[Top]

Stopping the FCS Service

Fig 12: Stopping the FCS Service

Open the "Services" program like when starting the FCS service, select the line corresponding to the File Server and click on the "Stop" button: a popup window appears while the service is stopping the "Started" status disappears

[Top]

The Intersite Transfers

It is possible to work on the same project, even if the different developers are located on different sites. For this purpose, see the documentation related to the [2] intersite transfers.

The Multi-site Transfer Manager

Purpose

The Multi-site Transfer Manager controls data replication between multiple SCM installations. Having two installations on a W.A.N., launching the Transfer Manager on one site allows to send or receive data from the other site.

[Top]

How to Start It

Log on the host where the Multi-site Transfer Manager will be launched and change to the identity needed for starting it. Note that it is possible to use s specific identity for starting the manager regarding the remark here after.

The Transfer Manager will execute commands in the workspaces dedicated to replicating data and this will be done under the same login as the one used for starting it.

Set up the current environment using the TCK for administration and execute the adl_transfer_mngr command. The port number that is specified is the one that will be used for starting data transfers afterwards.

E:\>adl_transfer_mngr -port 23452 -cmdfile %ADL_ADMIN_PATH%\AuthorizedCommands.txt (*)

(*) Syntax for Windows

The process is then ready to serve requests.

Note that the '&' character can be used on Unix platform for running the process as a background process. It is not possible to run it the same way on Windows platform. SCM provides the adl_admin_site command for registering a service for that purpose.

On UNIX platform it is possible that the transfers fail due to system resource limits. It is advised to use the adl_start_transfer_mngr.sh command instead of adl_transfer_mngr directly in order to let it increase those limits.

$ adl_start_transfer_mngr.sh -port 23452 -cmdfile $ADL_ADMIN_PATH\AuthorizedCommands.txt (**)

(*) Syntax for Unix

The file "AuthorizedCommands.txt" that appears in the previous examples contains the list of commands that are not SCM commands but that are required for sharing components between sites. It must be specified on Unix platform but is automatically referenced on Windows platform.

[Top]

How to Stop It

From any host and being the SCM administrator, set up the current environment using the TCK for administration and execute the adl_transfer_client command with the -stop_server option. This option causes the Multi-site Transfer Manager to stop. Note that the server is stopped even there are some multi-site transfers under way.

E:\>adl_transfer_client -server <hostname> -port 23452 -stop_server

• The <hostname> is the name of the machine where the Transfer Manager is running.
• The port number is the one used for starting the daemon
• The -stop_server option specifies the daemon must stop.

[Top]

Using Windows Services

On Windows platform, the adl_admin_site command provides the -create_service for registering a new service to launch a SCM daemon (Monitor, File Server or Multi-site Transfer Manager) on the host where the service is declared.

To declare a service for the Multi-site Transfer Manager, log on the corresponding SCM server having system privileges, use the TCK for administration to have access to the adl_admin_site command and execute the following command line:

E:\>adl_admin_site -create_service -service_type transfer -port 23452 -start_at_boot

The -start_at_boot option is used for specifying to the service system manager that the given service must be started at boot time.

[Top]

Starting the SCM Transfer Service

Fig 13: Control Panel/Services

The adl_admin_site command just declares a new service but does not start it. To start the File Server, use the Windows service tool. This tool is found by opening the "Start/Settings" menu and selecting the "Control Panel" item. Then run the "Services" programs and find the service named "SCM Transfer Manager V5R8_T08" where V5R8_T08 is the TCK corresponding to the Transfer Manager installation.

Fig 14: Starting the Transfer Manager Service

To start the Transfer Manager service, select the corresponding line and click on the "Start" button: a popup window appears while the service is starting the service status changes to "Started" when the service has been started The way the service is started (manual or automatic) can be changed whatever the service was registered using the adl_admin_site command. The startup type can be also changed to "Disabled" if you want to prevent the system (or anyone) to start the Transfer Manager for a while.

[Top]

Stopping the SCM Transfer Service

Fig 15: Stopping the Transfer Manager Service

Open the "Services" program like when starting the Transfer Manager service, select the line corresponding to the Transfer Manager service and click on the "Stop" button: a popup window appears while the service is stopping the "Started" status disappears

[Top]

Monitoring Commands

All commands run under Monitor control. Therefore it is possible to list or control the running commands at any time by retrieving information from the Monitor.

The adl_mo_admin command gives access to information managed by the Monitor process.

This command is available after setting the environment with the TCK for administration. Here are options available for querying the Monitor.

(*) In specific cases only. The command can get these information from the current environment.

[Top]

In Short

Commands store information in two kinds of repositories: File Server repositories and SCM databases.

Several File Server repositories can be set up, each of them being managed by a dedicated File Server process.

All commands run under the control of a Monitor process. This process can be queried using the adl_mo_admin command.

[Top]

References

History

Copyright © 2000, Dassault Systèmes. All rights reserved.