Beginner Configuration Management

Source: http://space.itpub.net /? Uid-12231076-action-viewspace-itemid-193

Recently I am working on a company project. In this project, I am not responsibleTestIn addition, we also collect cm (Configuration Management) and metric data. The test also belongs to the Quality Assurance Department. I am a tester and also responsible for configuration management. The company will make reasonable use of resources, I just don't pay for it. Now that the project is in the outline design stage and the requirement baseline has just been imported into the database, let's talk about how the configuration administrator feels during the demand development period.

People who have done configuration management know that this job is hard and easy to say. For new users in this field, it is hard to imagine that there is no supporting document template for cmme. Although we have some document templates, they are not perfect and there are no successful cases for reference. It is a painful experience.

I have never done configuration management before, and common configuration management tools such as VSS and CVS won't be used. Now I want to learn how to sell it. for document management, we chose VSS. According to the meaning of the project team, the Library was created first, these include the project baseline database, personal development database, project controlled database, and process controlled database. The building process of the database will not be mentioned. I believe everyone will.

The following describes how to add users and assign permissions. Users are allocated along the way according to the system role permission table of the configuration library. The permission is roughly: baseline database: only the configuration administrator, that is, I have all permissions, and others are read-only;

Personal Development Library: PM and CM have all permissions. Others have all operations except deletion on their folders. developers can operate on each other's items in their folders, apart from being deleted, ppqa and testers can only read files from others.

Project controlled database: PM and CM have all permissions, developers and ppqa read-only permissions, testers have all permissions to delete the controlled database in the test part, and read-only permissions to other files.

Process controlled database: PM has all permissions. CM has some permissions for some folders, all permissions for other folders, and only read-only permissions, developers, testers, and ppqa have the permission to delete some files and only read-only permission on other files.

Looking at the role permission table, I wrote it out. Oh, isn't it messy? Maybe my description is not clear enough. Ignore it first.

Next, there is no management purpose, because no Configuration Management Plan is written, but what others tell us to do is to write Configuration Management weekly reports and configuration status reports every week, another thing is to back up data. After just one month, I attended several training and project group meetings to know that configuration management was not as simple as I thought, we need to monitor a lot of things, not simply making statistics and backups.

Now I have found some problems: the document submission is messy. Sometimes, when collecting statistics, I find that the document has been deleted. I only know how many documents have been deleted, but I don't know which ones have been deleted, in addition, anyone in the controlled database can submit the changes and check out the changes. If n people change a document, it is obvious that the controlled database has no significance. I reflected this issue at the group meeting. After discussion, the permissions were re-divided.

The new authorization is as follows:

• Controlled Database: the permissions remain unchanged. CM has all permissions and others are read-only.
• Personal Development Library: Personal operations on personal folders, read-only to others. PM and CM can operate the folders of all users.
• Project product controlled database: CM has all permissions. Others are read-only.
• Process product controlled database: CM has all permissions. Others are read-only.

This is clear, and the controlled database also plays a role in control. I don't know if it is reasonable, but it is at least easier than the previous management, all the documents to be submitted to the controlled database are put by myself, which makes it much easier to collect statistics. The reports are submitted by baseline or change, so the backups are also regular, improving ......

Now I have to do more and more clearly, fill in the configuration management plan, which was originally written at the beginning of the project requirements, and now quickly complete back, the configuration item status table: the document naming format, process domain, storage location, and submission time are provided for reference in the project development stage. In the past, backup was self-determined and there were no backup records or statistical records. Now you need to fill in the daily backup application form.

After the work is clear, everything is easy. In the past, due to permission confusion, the submission of documents was messy. For the sake of security, we had to back up data every day, adding a lot of work, in addition, the backup server of the company has not been arranged yet. To put it on my own machine, the project has produced a lot of documents and the data volume is growing, the daily backup data volume is too large (I used a full backup, and it seems that Incremental backup is not provided in VSS. Now, as both the baseline database and the controlled database are put in the same directory, you can back up the database once a week without fear of data loss, let them manage it on their own. I only need to follow my Configuration Management Plan and ask for data from them. Isn't it easy.

When it comes to backup, I started to copy files completely to the local device. The advantage is that it is simple and easy to operate. During recovery, I copied the backup back to the server to overwrite the original data, however, it has the disadvantage of large data volumes. Currently, the. Ssa backup file is generated, which saves a lot of space. If you also generate a backup method in. Ssa format, pay attention to the following points for restoration:

1. If the original archive all the date backup is performed and the original project is not deleted.
Then, "permanently delete" the original project and restore it.

2. If the original archive this version and older are used and you are sure to reply to the original version
Then, "permanently delete" the original project and restore it. (If there are documents shared with other projects in projects, the latest version will not be overwritten)

Note:
1. The Archive this version and older method is not recommended. Unless you want to save hard disk space, delete the original version after the backup,
To avoid trouble during recovery.
2. "permanently delete" a project can be performed in VSS release E, or in VSS administrator.

In fact, CM is still responsible for version control. As for how to control the version, the personal method will be different, but the purpose is the same. Generally, the company may use two days of servers, one as the project staff development library, the other as the controlled Library and the baseline library. In this way, the configuration Administrator is required to fetch data from the project development database every time and use his workbench as an intermediary. In this case, the same directory as the database can be created on his workbench, you can directly get the Developer Library and the controlled library from the project to the local, and check the content to be updated in the future and then check in
To the controlled database. We use a controlled database. The baseline database and the project developer library are on one server, but the operations are similar. In my opinion, the controlled database and the Developer Library of the project should be separated for security reasons, so that the permission management can be simpler.

I saw a document about how to establish a relatively "secure" VSS Data Protection System and sorted it out for your reference.

1. Set a security folder andDatabaseDirectory parallelism (for example, if the database is in D:/doc/VSS/AAA, the directory is set to D:/doc/VSS/lock) and this directory is hidden, set the permissions of common team members to readable.

Write but not column directory. You cannot directly access this directory through a browser to prevent accidental deletion.

2. The directory where the VSS library is located remains unchanged, but only the srcsafe. ini file is retained. Everything else is moved to the above Security folder. (Create d:/doc/VSS/lock/AAA to hold the files and subfolders in the original AAA)

3. in the file srcsafe. in INI, point the path and file to the actual content (add the relative path, for example, the original is data_path = data, and now change .. /lock/AAA/data, and so on ).

After srcsafe. INI is modified, a backup is made to restore the damaged data.

4. Add a random file name (such as asdhfn3.7d) to the security directory to monitor the reading and writing of the file (check whether there is unauthorized access)

5. This security system does not eliminate hardware faults and damage caused by the permission of others to use administrator. Therefore, regular backup is still required.

To sum up, the completed and ongoing documents include:

1.  
   1. Configuration status report
   2. Daily backup application
   3. Configuration Management Plan
   4. Configuration Management weekly
   5. Configuration Library System
   6. Configuration item status table

With these table specifications, I believe the subsequent configuration management should be clearer.

Now, the baseline for the demand stage has been released and passed the review to enter the outline design and detailed design stages. to grow in the project, you can see the progress.