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

Brief data failure scenario:

Users use two disks to form a raid0, one of which turns on the yellow light, and after being kicked out by the raid card, the raid crashes. The following focuses on the rescue of data at that time.

The hard disk used by the user is two pieces of SAS 300G. First, the hard disk is removed from the machine, then it is directly connected to the win environment by SAS HBA, and the hard disk is marked as offline in the disk management to ensure that it is read-only in the process of operation and protect the security of the original data.

After mirroring all the sectors at the bottom of the two hard disks, the disk order and stripe size are analyzed by the file system, and the original raid environment is built by software virtual reorganization. After further parsing the ntfs file system, we finally see the data. At this time, a new problem arises. After copying the data directly, the original system and applications need to be redeployed. And because without the support of software service providers, it is not easy to implement, so think of the complete migration of the built raid to the new raid environment, it can be the same as before the damage. Saved a lot of time.

Customers have learned a lesson from the new raid environment and decided to use three disks to build raid5, so that even if a disk is offline, raid can still be degraded and will not crash immediately, giving users a chance to replace a new hard disk.

After installing a new raid card that supports raid5 and plugging in a new hard drive, we have created a raid5 volume and now begin to study how to migrate the resulting data.

Overview of data migration:

Because the front panel of the server is managed by the raid card, inserting a new disk directly will not be directly identified under the system. You need to create a raid under the raid card before you can use it. Moreover, it is limited to the capacity of a single disk, so you cannot use this solution, so you can study other methods.

Because the front panel of the server has a DVD optical drive, and now the server optical drive and the motherboard are connected with the sata channel, so you can open the cover and use the sata port above to connect a sata hard disk, and you can move the data back in pe or linux live cd mode, and this is the fastest way. However, when we are ready to implement, we find that the sata used in this machine is not a standard size interface type, but mini sata. Because there is no ready-made adapter card in hand, this method is not good either. In fact, we can also use USB to do it when the amount of data is small, but because the usb of most servers is still too slow in USB2.0, the time is unacceptable for large amounts of data.

In the end, we absolutely use a novel way to move back to the data-the Internet.

At this time, we need to start a linux live cd first. Generally, we use linux system rescue cd to configure the ip of the server under ifconfig after the linux is started. Then we put the data on a win 2008 R2 machine and open the nfs service (off by default) in the win environment. "Service Manager-role-add role-check File Service-check Network File system Service for installation You need to restart the computer after the first installation. "

After the restart, we operate on the folder where the mirrored data is stored. Select share this folder in the right-click-NFS sharing tab, and then select allow root directory access in the permissions, and select read and write access type.

After the settings on the Win side are complete, we will take a look at the settings on the Linux side, and ifconfig will view the current network configuration.

Because we need to assign an ip to him, here we assign the network card "enp4s0" and the ip address is assigned to 10.3.12.3 with a subnet mask of 255.0.0.0, use the following command: ifconfig enp4s0 10.3.12.3 255.0.0.0 and then use ifconfig to view the ip address.

After configuring ip, check to see if the network comes with, command: ping 10.1.1.1

Then check to see if the directory of the NFS share on the 10.1.1.1 machine is accessible, command: showmount-e 10.1.1.1.

The source and destination machines are now connected, create a directory mkdir / mnt/bysjhf in linux

Once created, we mount the mirrored data to mount 10.1.1.1:/data / mnt/bysjhf-o nolock under the newly created folder under linux

After mounting, check the mount point information df-k

After confirming that it has been mounted, go to this folder and check the image files in the folder:

Root@sysresccd / mnt/bysjhf ls

And check the hard disk and partition information: fdisk-l

After confirming the source and target devices, perform the mirror operation:

Dd if=/mnt/bysjhf/data.img of=/dev/sda bs=10M

In the gigabit network environment, the speed of NFS can run to 70M/S, which is already a very ideal speed. After waiting for the completion of dd, we restart the ibm X3650 server and choose raid to boot. The expected windows startup page finally appears, the previous hard work has not been wasted, and the complete data migration is successful.

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.