Overview of Scaling Up Partitions and File System

This topic uses the CentOS 7.6 64-bit operating system as an example to assign partitions to the disk after scaling up using the fdisk and parted partitioning tools.

Scenario

Upon completion of scaling up, you must plan the expanded capacity into the existing partition or allocate a new partition to the EVS to contain the expanded capacity.

This topic uses the CentOS 64-bit operating system as an example to assign partitions to the disk after scaling up using the fdisk and parted partitioning tools. The operation may differ depending on the operating system. This topic is for reference only.

To assign a partition to a disk after scaling up, you can select an appropriate scaling up method according to your business needs and disk condition, as follows.

EVS	Scenario	Scaling Up Method
System Disk	Planning the expanded capacity to the existing MBR partition	Expanding the existing MBR partition
	Planning the expanded capacity to a new MBR partition	Adding an MBR partition
Data Disk	Planning the expanded capacity to the existing MBR partition or GPT partition	Expanding the existing partition (MBR partition or GPT partition)
	Planning the expanded capacity to a new MBR partition or GPT partition	Adding an MBR partition or GPT partition

Restriction and Limitations

Please exercise caution with scaling up. Data loss or error may occur due to misoperation. We recommend that you back up data before scaling up. Volume Backup Service (VBS) is recommended.
Check the partition form of the EVS disk before scaling up after putting into use, as follows:

Partitioning Form	Cautions
MBR partition	A maximum of 2 TB (2048 GB). The capacity part beyond 2 TB is unavailable.
GPT partition	The capacity can be up to 18 EB (19327352832 GB). An EVS supports a maximum capacity of 32 TB (32768 GB) data disks. That is, you can scale up data disks to 32 TB.
An MBR partition must be expanded to greater than 2 TB.	The disk partitioning form must be switched from MBR to GPT. During this process, services will be interrupted and the original data on the disk will be cleared after the disk partitioning form is changed. Therefore, back up the data before capacity expansion.

Before You Begin

You have logged in to the elastic cloud server. For more information, see Overview of Linux ECS Login.
The system disks and data disks have been attached to the cloud server and initialised.
The EVS has been scaled up in the console, and attached to the elastic cloud server. For more information, see Overview of Scaling Up EVS.

Inspection Before Operation

View the partition form and select a partition tool

Before partitioning, check the partition form of the current disk. If MBR is used, choose fdisk or parted tool. If GPT is used, use the parted tool.

Run the fdisk -l command to view the partition form of the current disk. The following information is returned:

"Disk label type" indicates the partition form of the current disk, "dos" indicates that the partition form is MBR, and gpt indicates that the partition form is GPT.

Checking the file system of the disk to be scaled up

Before scaling up, check whether the file system of the disk to be scaled up can be attached properly.

Run the df -TH command to check whether the disk is attached. The following information is returned:

You can note that the file system of /dev/vdb1 is ext4 and the disk has been attached to the /mnt/sdc directory.
Run the ll /mnt/sdc command to view the files on the disk in the mount directory. If you can view files on the disk, the disk to be scaled up is normal.

Scaling Up Partitions and File System for the System Disk

Expanding the Existing MBR Partition

In this example, CentOS 7.6 64-bit is used as an example. The existing system disk /dev/vda has the /dev/vda1 partition with a capacity of 40 GB and you have expanded 10 GB for the partition in the console. The following steps will plan the newly added 10 GB into the existing /dev/vda1 partition.

Run the growpart command to check whether the cloud server is installed with the growpart tool. If the information shown in the figure is returned, the tool is installed.

Note:

If the information shown in the figure is not returned, run the yum install cloud-utils-growpart command to manually install the growpart tool.
Run the fdisk -l command to view the total capacity of system disk /dev/vda on the cloud server. The information as shown in the following figure is returned:
Run the df -TH command to view the existing capacity of system disk /dev/vda1. The information

shown in the figure is returned:

According to the returned information of two commands, the capacity of system disk /dev/vda is 50 GB, but the capacity of partition /dev/vda1 is only 40 GB. Therefore, you must scale up partition /dev/vda1.
Run the growpart /dev/vda 1 command to scale up the partition. Specify system disk /dev/vda and number the partition to be scaled up as 1. The information shown in the figure is returned:

Description
If the error "NOCHANGE: partition 1 is size xxxxxxx. it cannot be grown" is displayed, the disk of the cloud server is full (100% usage) and cannot be scaled up. In this case, back up data and delete unnecessary files or programs to release the disk space. For more information on how to clean up disk space, see Addressing the Problem with Insufficient Space of Linux-based Cloud Server in Best Practices.
Run the xfs_growfs /dev/vda1 command to scale up the file system of the existing partition /dev/vda1. The information as shown in the following figure is returned:

Description
If the file system of the disk partition is in xfs format, use the xfs_growfs tool to scale up the disk partition.
If the file system of the disk partition is in ext format, use the resize2fs tool to scale up the disk partition.
Run the df -TH command to view the capacity of partition /dev/vda1 of the system disk after scaling up. The information as shown in the following figure is returned:

As shown in the figure, the existing partition has been scaled up to 50 GB.

To enable automatic disk partition mounting upon startup, see Enabling Automatic Disk Mounting Upon Startup.

Adding an MBR partition

In this example, CentOS 7.6 64-bit is used as an example. The existing system disk /dev/vda has the /dev/vda1 partition with a capacity of 50 GB and you have expanded 10 GB for the disk /dev/vda in the console. This example shows how to use the fdisk tool to allocate a new MBR partition to the newly expanded 10 GB capacity of the system disk and mount it to the /opt directory.

Run the fdisk -l command to check the partition information of the disk. The following information is returned:

The current capacity of system disk /dev/vda is 60 GB.
Run the fdisk /dev/vda command to enter the fdisk partitioning tool. The information shown in the figure is returned:
Type "n" and press Enter to create a new partition. The following information is returned:

Note:
p means a primary partition and e means an extended partition.
When the MBR partition is used, a maximum of four primary partitions or three primary partitions and one extended partition can be created. In addition, an extended partition cannot be used directly. Instead, it must be divided into several logical partitions.
Type "p" and press Enter to create a primary partition. The information as shown in the following figure is returned:
Type the default partition No. as "2" and press Enter . The information as shown in the following figure is returned:
The cylinder number of the new partition, including the start cylinder number and the end cylinder number. Use the default values here and press the Enter key to confirm. Users can also customize the number value here. Note that the start cylinder value must be lower than the end cylinder value. The information as shown in the following figure is returned:
Type "p" and press Enter to view the newly created partition. The newly created partition is /dev/vda2 and the information shown in the figure is returned:
Type "w" and press Enter to write the partition result into the partition table. The following information is returned:

The information returned as shown above indicates that the partition is created.

Note:

If you make an error in the partition operation, type "q" to exit the fdisk partitioning tool. The previous partition results are not retained and subsequent operations are not affected.

Run the partprobe command to synchronize the changes to the new partition table to the operating system.
Run the mkfs -t ext4 /dev/vda2 command to set the format of the file system of the new partition /dev/vda2 to ext4. The information shown in the figure is returned:

Description
For an EVS with a large capacity, the mkfs command may take a long time to execute. To shorten the command execution time, you can add the following parameters to the mkfs command:
1. For ext2, ext3, and ext4 file systems, add the -E nodiscard parameter
2. For the xfs file system, add the -K parameter
The formatting process takes a while. Observe the running status of the system. If "done" is displayed in the returned information, the formatting process is complete, as shown in the figure.
Run the mount /dev/vda2 /opt command to mount the newly created partition /dev/vda2 to the directory where space is to be expanded. In this example, the newly created partition is mounted to /opt.

Note:

We recommend that you mount the newly created partition to an empty directory or run the mkdir command to create a directory. If the newly created partition is mounted to a non-empty directory, the existing subdirectories and files in the directory will be hidden. If a mount directory is specified and the directory is not empty, you can temporarily move the subdirectories and files in the directory to another directory, and move the subdirectories and files back again after the newly created partition is mounted.
Run the df -TH command to check the mount result. The information shown in the figure is returned:

To enable automatic disk partition mounting upon startup, see Enabling Automatic Disk Mounting Upon Startup.

Scaling Up Partitions and File System for the Data Disk

This topic describes how to scale up disk partitions and file system for data disks in a Linux environment.

Expanding the Existing Partition (MBR partition and GPT partition)

In this example, the CentOS 7.6 64-bit operating system is used as an example. The existing data disk /dev/vda has the /dev/vdb1 partition with a capacity of 10 GB and you have expanded the data disk to 20 GB in the console. This example shows how to plan the added 10 GB capacity into /dev/vdb1 in the existing MBR partition or GPT partition.

Run the growpart command to check whether the cloud server is installed with the growpart tool. If the information shown in the figure is returned, the tool is installed.

Note:

If the information shown in the figure is not returned, run the yum install cloud-utils-growpart command to manually install the growpart tool.
Run the fdisk -l command to view the partition information of the disk. The information shown in the figure is returned:
Run the df -TH command to view the capacity of partition /dev/vdb1 of the data disk. The information shown in the figure is returned:

As shown in the figure, the capacity of partition /dev/vdb1 is 10 GB, but the capacity of disk /dev/vdb is 20 GB. Therefore, you must expand the capacity of partition /dev/vdb1.
Run the growpart /dev/vda 1 command, specify the partition of the data disk to be scaled up (the data disk is /dev/vdb and the partition number is 1), and run the growpart command to scale up the partition. The information shown in the figure is returned:
Run the resize2fs /dev/vdb1 command to scale up the file system of the existing partition /dev/vdb1. The information as shown in the following figure is returned:

Description
1. If the file system of the disk partition is in xfs format, use the xfs_growfs tool to scale up the disk partition.
2. If the file system of the disk partition is in ext format, use the resize2fs tool to scale up the disk partition.
Run the df -TH command to view the capacity of partition /dev/vdb1 of the data disk after scaling up. The information shown in the figure is returned:

In this case, the capacity of partition /dev/vdb1 of the data disk is expanded to 20 GB.

To enable automatic disk partition mounting upon startup, see Enabling Automatic Disk Mounting Upon Startup.

Adding an MBR partition

In this example, the CentOS 7.6 64-bit operating system is used as an example. The existing data disk /dev/vda has the /dev/vdb1 partition with a capacity of 20 GB and you have expanded the data disk to 30 GB in the console. This example shows how to allocate a new MBR partition to the newly added 10 GB capacity.

Run the fdisk -l command to check the partition information of the disk. The following information is returned:

The current capacity of system disk /dev/vdb is 30 GB.
Run the fdisk /dev/vdb command to enter the fdisk partitioning tool. The information shown in the figure is returned:
Type "n" and press Enter to create a new partition. The following information is returned:

Note:
p means a primary partition and e means an extended partition. When the MBR partition is used, a maximum of four primary partitions or three primary partitions and one extended partition can be created. In addition, an extended partition cannot be used directly. Instead, it must be divided into several logical partitions.
Type "p" and press Enter to create a primary partition. The information as shown in the following figure is returned:
Type the partition No. as "2". Use the default value and press Enter . The information as shown in the following figure is returned:
Confirm the cylinder number of the new partition, including the start cylinder number and the end cylinder number. Use the default values here and press the Enter key to confirm. Users can also customize the number value here. Note that the start cylinder value must be lower than the end cylinder value. The information as shown in the following figure is returned:
Type "p" and press Enter to view the newly created partition. The newly created partition is /dev/vdb2 and the information shown in the figure is returned:
Type "w" and press Enter to write the partition result into the partition table. The following information is returned:

The information returned as shown above indicates that the partition is created.
Note:

If you make an error in the partition operation, type "q" to exit the fdisk partitioning tool. The previous partition results are not retained and subsequent operations are not affected.
Run the partprobe command to synchronize the changes to the new partition table to the operating system.
Run the mkfs -t ext4 /dev/vdb2 command to set the format of the file system of the new partition /dev/vdb2 to ext4. The information shown in the figure is returned:

Description
For an EVS with a large capacity, the mkfs command may take a long time to execute. To shorten the command execution time, you can add the following parameters to the mkfs command:
1. For ext2, ext3, and ext4 file systems, add the -E nodiscard parameter
2. For the xfs file system, add the -K parameter
  
  The formatting process takes a while. Observe the running status of the system. If "done" is displayed in the returned information, the formatting process is complete, as shown in the figure.
Run the mkdir /mnt/sdc1 command to create the /mnt/sdc1 directory to which you want to mount the partition.
Run the mount /dev/vdb2 /mnt/sdc1 command to mount the newly created partition /dev/vdb2 to the directory where space is to be expanded. In this example, the newly created partition is mounted to /mnt/sdc1.

Note:

We recommend that you mount the newly created partition to an empty directory or run the mkdir command to create a directory. If the newly created partition is mounted to a non-empty directory, the existing subdirectories and files in the directory will be hidden. If a mount directory is specified and the directory is not empty, you can temporarily move the subdirectories and files in the directory to another directory, and move the subdirectories and files back again after the newly created partition is mounted.
Run the df -TH command to check the mount result. The information shown in the figure is returned:

To enable automatic disk partition mounting upon startup, see Enabling Automatic Disk Mounting Upon Startup.

Adding an GPT Partition

In this example, the CentOS 7.6 64-bit operating system is used as an example. The existing data disk /dev/vda has the GPT partition /dev/vdb1 with a capacity of 10 GB and you have expanded the data disk to 20 GB in the console. This example shows how to allocate a new GPT partition /dev/vdb2 for the added 10 GB capacity.

Run the lsblk command to view the partition information of the disk. The information shown in the following figure is returned:
Run the parted /dev/vdb command to enter the parted partitioning tool. The information shown in the following figure is returned:
Type "unit s" and press Enter to set the metering unit of the disk to cylinder.
Type "p" and press Enter to view the partition status of the current data disk. The information shown in the figure is returned:

If the Error message shown in the figure is displayed, select type "Fix" after "Fix/Ignore/Cancel?". The information of the GPT partition table is usually stored at the start of the disk, but a copy is also backed up at the end of the disk. When the disk capacity increases, the end position of the partition table is also extended backward. Therefore, you must type "Fix" as prompted to correct the backup of the partition table information to the end position of the new disk. The information shown in the following figure is returned:

If the "Warning" message is displayed after you type "Fix", continue to type "Fix". This step cannot be ignored. Only when you type "Fix", the system will automatically set the partition form of the expanded capacity to GPT. The information shown in the following figure is returned:
You must remember the end cylinder value in the specific partition information given here, that is, the value in the End column. In this example, the value is 20969471s.
Type the mkpart command in the format of "mkpart disk partition name start cylinder value end cylinder value", and press Enter .

Note:
Configure the command parameters as needed. In this example, as shown in Step 5, the end cylinder value of the existing partition of the data disk is 20969471s. Therefore, for the newly created partition /dev/vdb2, set the start cylinder value to 20969472s, the next position of the existing partition. The end cylinder value is to 100%, indicating the full capacity. The name of the disk partition is set to "data". As a result, in this example, type the command "mkpart data 20969472s 100%".
If a question-like message appears, type Yes in the Yes/No field to agree with the operation of placing the partition at the end of the disk.
Type "p" and press Enter to view the newly created partition. The information shown in the shown in returned:
Type "q" and press Enter to exit the parted partitioning tool. The information shown in the shown in returned:
Run the mkfs -t ext4 /dev/vdb2 command to set the file system of the newly created partition /dev/vdb2. The format of the file system is ext4. The information as shown in the following figure is returned:
Description

If the format of the file system is set to xfs, run the mkfs -t xfs command. The disk formatting process takes a while. If "done" is displayed in the returned information, the formatting process is complete. Do not quit when formatting is in progress. For an EVS with a large capacity, the mkfs command may take a long time to execute. To shorten the command execution time, you can add the following parameters to the mkfs command:

For ext2, ext3, and ext4 file systems, add the -E nodiscard parameter

For the xfs file system, add the -K parameter

Run the mkdir /mnt/sdc1 command to create the /mnt/sdc1 directory to which you want to mount the partition.
Run the mount /dev/vdb2 /mnt/sdc1 command to mount the newly created partition /dev/vdb2 to the newly created /mnt/sdc1 directory. The operations and returned information are shown in the following figure:

Note:

We recommend that you mount the newly created partition to an empty directory or run the mkdir command to create a directory. If the newly created partition is mounted to a non-empty directory, the existing subdirectories and files in the directory will be hidden. If a mount directory is specified and the directory is not empty, you can temporarily move the subdirectories and files in the directory to another directory, and move the subdirectories and files back again after the newly created partition is mounted.

Run the df -TH command to check the mount result. The information shown in the shown in returned:

To enable automatic disk partition mounting upon startup, see Enabling Automatic Disk Mounting Upon Startup.

Setting Automatic Disk Mounting upon Startup

If you want to automatically mount disks when the cloud server system starts, do not specify /dev/vdb1 in /etc/fstab because the sequential encoding of devices in the cloud may change when the cloud server stops or starts. For example, /dev/vdb1 may change to /dev/vdb2. We recommend that you use UUID to configure automatic disk mounting. The Universally Unique Identifier (UUID) of a disk is the universally unique string that the Linux system provides for disk partitions.

Run the blkid /dev/vdb1 command to query the UUID of the disk partition. The following information is returned:
Run the vi /etc/fstab command to open the fstab file using the VI editor. Type "i" to go to the editing mode, move the cursor to the end of the file, and press Enter to add the following content. In the UUID= field, please enter the UUID you queried in the previous step, as shown in the figure:
Press ESC, type ":wq", and press Enter . Save settings and exit the editor.
To verify the automatic mounting function, unmount the mounted partition and run the umount /dev/vdb1 command, as shown in the figure:
Run mount -a to reload all the contents of the /etc/fstab file, as shown in the figure:
Run the mount |grep /mnt/sdc command to query the mount of the file system. The operations and returned information are shown in the following figure:

If the information shown in the figure is returned, the automatic disk mounting is successfully set.

Elastic Volume Service

Scaling Up Partitions and File System (Linux)