RH133读书笔记(7)-Lab 7 Advanced Filesystem Mangement

Lab 7 Advanced Filesystem Mangement

Goal: Develop skills and knowlege related to Software RAID, LVM, quota and backup.

Estimated Duration: 120 minutes

Sequence 1: Implementing Quotas

Deliverable: A user diskhog that cannot use more than 1024k of space in /home.

Instructions:

1. Create a user diskhog.

# useradd diskhog

2. Activate user quotas for /home.

a. In /etc/fstab change the mount options for /home to include usrquota. If you do not have a separate /home partition use /.

LABEL=/home /home ext3 defaults,usrquota 1 2

b. Activate the new mount option.

# mount -o remount /home

c. Bring the system into single user mode, to ensure correct quota calculation:

# init 1

d. Run quotacheck -cu /home to create the quota file.

e. Leave single user mode

# init 5

f. Enable Quota enforcing.

# quotaon /home

3. Set the soft block quota of user diskhog to 512 1k blocks and the hard limit to 1024 1k blocks.

# setquota -u diskhog 512 1024 0 0 /home

4. Test the restrictions.

To test these restrictions, run the following commands:

su - diskhog
quota

dd if=/dev/zero of=bigfile bs=1k count=400
quota

(Should work fine.)

dd if=/dev/zero of=bigfile bs=1k count=800
quota

(Should issue a warning.)

dd if=/dev/zero of=bigfile bs=1k count=1600
quota

(Should fail to write the whole file.)

Sequence 2: Software RAID

Deliverable: A system with two RAID1 devices with two partitions each.

Instructions:

1. This lab assumes that the computers are only equipped with one hard drive. Setting up RAID on a single disk obviously does not provide any kind of fault protection.

Create four additional partitions with at least 100MB each. The partition type should be set to “Linux raid auto”.

a. Use fdisk to create four logical partitions

b. Set the Partition Type (T) to fd

c. Save and exit fdisk

2. Make sure that the kernel uses the new partition table.

# partprobe

3. Use mdadm to create two RAID 1 devices, with two partitions each. These devices should be /dev/md0 and /dev/md1. Create a filesystem on the first device.

Note: Adjust the partition numbers to match your system.

# mdadm -C /dev/md0 -a yes -l 1 -n 2 /dev/sda{6,7}
# mdadm -C /dev/md1 -a yes -l 1 -n 2 /dev/sda{8,9}

# mkfs.ext3 /dev/md0

4. Mount the filesystem under /data and put some files on it.

# mkdir /data

# mount /dev/md0 /data

Note: Do NOT put it in fstab yet.

# cp -a /lib /data

5. Get a printout of the current RAID configuration.

# mdadm --detail /dev/md0

6. It is now time to simulate a disk failure. Use mdadm --fail to mark one of your partitions as faulty. Use mdadm --remove to remove this partition from the RAID array. Check /proc/mdstat and /var/log/messages to see how the system reacts to this error.

Note: The device names in this example might differ from your setup

# mdadm --fail /dev/md0 /dev/sda7

# cat /proc/mdstat

# mdadm --remove /dev/md0 /dev/sda7

Re-add the partition to the RAID device with mdadm --add.

7. If this had been a real hard disk error, you would have to replace the broken disk and repartition it. Since our partition was never really damaged we can skip this part and re-add the partition to the RAID device with mdadm --add. After you re-add the partition, note the content of /proc/mdstat.

# mdadm -a /dev/md0 /dev/sda7

# cat /proc/mdstat

Sequence 3: Creating Logical Volumes with LVM

Deliverable: A volume group that uses one of the RAID1 devices as a physical volume and has one logical volume.

Instructions:

1. Unmount the filesystem from the last exercise and convert your first RAID device (/dev/md0) into a physical volume.

# umount /dev/md0
# pvcreate /dev/md0

2. Create a volume group named volgroup with the default extent size of 4MiB, which uses the physical volume that was just created.

# vgcreate volgroup /dev/md0

3. Create a logical volume named data , that uses all available extents. Create a filesystem on it and make sure that this filesystem is mounted to /data during boot.

Note: LVM caches device information. Due to this cache new volume groups are ignored during boot. Delete /etc/lvm/.cache, so that LVM rescans the system for volume groups.

a. Use vgdisplay to find out how many extents are available.

# vgdisplay volgroup

b. Create the logical volume with all available extents.

Replace the word "free_extents" below with the number of free extents you found when running vgdisplay not the word "free_extents"

# lvcreate -n data -l free_extents volgroup

c. # mkfs.ext3 /dev/volgroup/data

d. Edit /etc/fstab so that /data will be mounted automatically.

/dev/volgroup/data /data ext3 defaults 1 2

e. Run mount -a to mount and check for errors.

f. Delete the LVM devices cache to ensure that the new volume group is detected during boot.

# rm /etc/lvm/.cache

g. Reboot to verify that the filesystem is automatically mounted.

Sequence 4: Extending a filesystem

Deliverable: An LVM that uses both RAID1 devices as physical volumes and has one increased logical volume.

Instructions:

1. Add the second RAID device to your volume group.

# pvcreate /dev/md1

# vgextend volgroup /dev/md1

2. Increase the data logical volume and filesystem by 40MiB.

# lvextend -L +40M /dev/volgroup/data

# resize2fs -p /dev/volgroup/data

# df -h

Challenge Sequence 5: Reducing a filesystem

Deliverable: The existing logical volume is reduced by 100MiB.

Instructions:

1. To reduce the volume the filesystem has to be reduced first. Reduce the filesystem and the logical volume by 100MiB.

a. First you need to determine the current filesystem size. This can be done with df -h

b. The filesystem has to be umounted before reducing

# umount /data

c. Make sure that the filesystem is in a consistent state before reducing.

# fsck -f /dev/volgroup/data

d. Now reduce the filesystem by 100MiB. This example assumes that the original size was 140MiB.

# resize2fs /dev/volgroup/data 40M

e. It is now possible to reduce the logical volume. Be careful with this command. Wrong parameters can render your filesystem unusable.

# lvreduce /dev/volgroup/data -L 40M

f. Verify the new size, by mounting the filesystem and running df -h

Challenge Sequence 6: Backing a Logical Volume with Snapshots

Deliverable: Take a backup from an active Logical Volume by using an LVM snapshot and dump. Destroy the data on the Logical Volume and recover from the backup.

System Setup: Confirm that the Logical Volume /dev/volgroup/data from an earlier lab is mounted on /data and configured in /etc/fstab.

Instructions:

1. Copy the contents of the /var/log directory into /data. View the contents of /data.

# cp -r /var/log/* /data/

# ls /data/

2. Create a read-only Snapshot Volume of /dev/volgroup/data called data-backup, allocate 16MB of space to the Snapshot.

# lvcreate -L 16M -p r -s -n data-backup /dev/volgroup/data

3. Mount /dev/volgroup/data-backup read-only as /mnt/data-backup and view it's contents.

# mkdir /mnt/data-backup

# mount -o ro /dev/volgroup/data-backup /mnt/data-backup

4. Use dump to backup /mnt/data-backup to a file named /tmp/databackup.dump. Confirm that the backup file was created.

# dump -0u -f /tmp/data-backup.dump /mnt/data-backup

# ls -la /tmp/data-backup.dump

5. Unmount /mnt/data-backup and remove the snapshot volume.

# umount /mnt/data-backup

# lvremove /dev/volgroup/data-backup

6. Unmount /data create a new filesystem on /dev/volgroup/data. Remount /data and confirm that it only contains the lost+found directory.

# umount /data

# mkfs.ext3 /dev/volgroup/data

# mount /data

# ls /data

7. Use dump to recover the lost contents of /data from the backup. Check the contents of the directory.

# cd /data

# restore -rf /tmp/data-backup.dump

# ls