Shulou(Shulou.com)06/02 Report--

Fault description: data disaster based on UNIX system and logical failure

Fault performance:

1. Storage structure error

2. Data deletion

3. File system formatting

4. Data loss due to other reasons

II. Solutions

◆ detection

1. Detect whether there is a hardware fault, such as a hardware fault, and switch to hardware processing.

2. Check whether the fault performance is the same as the user description in a read-only way

◆ recovery

1. Backup: make a full mirror of the fault storage in a read-only manner (see appendix)

2. Perform data analysis and recovery operations in the backup.

3. Usually, the recovered data will be temporarily stored on another bank.

◆ acceptance

Verify the recovered data and confirm its correctness. If confirmed, pay fee-> hand over original media and recovered data-> issue invoice (receipt) and report.

If the data recovery result can not be recognized, return the original media without service charge, and the report can be issued free of charge.

Third, the possibility of data recovery

★ has an error with the storage structure:

No matter the error occurs in the RAID structure part, or the volume group structure, partition structure, slice structure (different UNIX has different storage management methods). If the failure occurs in the way that one or more file systems are organized on a file system basis, and the file system itself is not damaged

After the problem, there are no other destructive operations, and the reliability of data recovery will be very high.

★ for file deletion:

If no subsequent data is written after the file is deleted:

"after the AIX JFS/JFS2 file is deleted, it can usually be restored at 100%."

"after the SGI XFS file is deleted, it can usually be restored at 100%."

Vxfs file deletion, if the number of files is small, there is an algorithm that can be fully recovered successfully, otherwise it will be handled as node loss (see later description).

When other files such as SCO HTFS and UFS are deleted, nodes are usually lost.

For a UNIX file system, if the node is lost, it means that the file's attributes (size, date stamp, permissions, association with name, and so on) cannot be obtained, and in some cases, the index cannot be found. Data recovery in this case mainly depends on the internal characteristics of the file to be restored. Usually, if it is a database file such as ORACLE, it can be recovered, but compressed data, multimedia data, and so on may not be easy to recover.

If there is a write after the file is deleted, the area covered by the written data cannot be restored.

★ is formatted for file systems

If no data is written after formatting:

1. AIX JFS and JFS2, most of the files before formatting can be recovered.

2. SGI XFS, most of the files before formatting can be recovered.

3. After Vxfs formatting, we need to analyze the overlap between the new structure and the pre-formatted structure, and the reliability of recovery is between AIX JFS2 formatting and UFS formatting.

4. After formatting other file systems (such as UFS), nodes are usually lost, which is the same as deletion. Refer to the recovery reliability of nodes lost after file deletion.

If data is written after formatting, the area covered by the written data cannot be restored.

★ for other reasons for data loss

In fact, all the loss of data recovery can be attributed to whether the node is lost, the index is lost, and the data itself is lost. Deletion and formatting can usually be understood as node and index loss. If the data itself is lost, there is no sense of data recovery. If a file node, index, and data itself can be found, it can be restored at 100%.

IV. Time

Disks below 1TB (not the data capacity to be recovered) can usually be completed within 2 working days, and recovery cycles above 1TB usually increase with the increase of disk capacity.

5. Tips

In response to software failures, ★ should minimize storage operations after data loss. Sometimes, even if you turn on the computer and do nothing, it may further aggravate the disaster. If conditions permit, it is best to make a full backup of the storage after the data is corrupted.

For hardware failures, ★ should power up as little as possible after the equipment does not work properly to avoid further damage to the equipment.

6. How to avoid

Make a good backup plan and avoid single storage backup as much as possible. if the data is very important, you can consider remote backup.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.