LVM 1.0.8 $Date: 2003/11/17 15:58:54 $ Logical Volume Manager frequently asked questions: -------------------------------------------------- Q0: Where can I find more information about the Logical Volume Manager for Linux? A0: Please see and/or join our mailing list at by following the subscription instructions at . You can find pointers to LVM howtos on the Web there as well. Q1: Why can't I create my volume group "mygroup"? A1: You can't create a volume group with the same name as an existing one. You can't create more than 99 volume groups at the moment. You are only able to use inititalized physical volumes (see pvcreate(8)). Q2: Why can't I create a physical volume with pvcreate(8)? A2: Maybe pvcreate(8) complains that the physical volume has already been initialized. So... think about "pvcreate -f[f] ...". But DON'T DO THAT, if the physical volume belongs to another volume group! Remember to set the partition system id to 0x8e with fdisk(8) before trying pvcreate(8) on it. Q3: Why am I not able to extend a logical volume? A3: Your volume group is full or you've already reached the maximum logical volume size in that volume group. Logical volume size is limited by the size of the physical extents times their maximum amount, which only can be set at volume group creation time. The default physical extent size is 4MB which limits ;-) logical volumes to a maximum of 256 Gigabyte (see vgcreate(8), vgdisplay(8)). If your volume group isn't full or you didn't reach the current logical volume size limit, your logical volume may have striped or contiguous allocation policy. Have a look at the physical volumes with vgdisplay or pvdisplay(8) to figure out, if there are not enough free (contiguous) physical extents. Q4: Why can't I move my logical volume(s) away from a physical volume with pvmove(8). A4: Look at the free space on all destination disks you want to use (or which are implicitly used) AND at the attributes of the logical volumes to be moved. Remember: you can't move a contiguous logical volume when there isn't enough free contiguous space on any destination disk. In this case you can think about changing from contiguous allocation policy to next free and do the attribute change with lvchange(8). You can't move a striped logical volume either, if there isn't enough space for the complete stripe on any destination physical volume. You can't move to physical volumes which are NOT allocatable. Think about changing this with pvchange(8). Q5: My striped logical volume works horrible slowly, what's going on? A5: If you put it on two or more physical volumes based on partitions on one disk, you will not gain any performance. Remember: you are allowed to use two or more partitions of one disk as physical volumes (this only makes sense for next free allocated logical volumes on those physical volumes). If you have attached two IDE disks to one adapter, you can't get parallel i/o on these two disks. Q6: Why can I not rename my volume group / logical volume? A6: You have to deactivate them before you are allowed to rename them (see lvrename(8), vgrename(8)). Q7: The LVM kernel patch in the LVM distribution is not compatible with my Linux version. Where's help? A7: If this relly is the case after following the instructions contained in PATCHES/README please send an email to . Q8: A LVM command was just working when my system crashed... Help please!? A8: Bring your system back online and look at the volume group backup files in /etc/lvmconf. There's at least one called /etc/lvmconf/VolumeGroupName.conf and possible more in the backup history called /etc/lvmconf/VolumeGroupName.conf.*.old. You can use these backup files to bring the configuration back to the one before the crash (see vgcfgrestore(8)). Q9: Why are my logical volumes limited to 256 GB in size? A9: This is NOT an absolute limit but it depends on the physical extent size you configured at volume group creation time. Please use option -s of the vgcreate command to give a larger physical extent size. For example with a physical extent size of 524288 KB (512 MB) you are able to map a logical volume of 32 Terabyte. Remember that actual 32 bit kernels are limited to 2 Terabytes (with 512 byte block size). Q10: Why can't I split my volume group my_vg? A10: The physical volumes you want to split into another volume group may NOT have logical extents of logical volumes belonging to the original volume group you started with. Please use pvmove to seperate the logical volumes. Q11: Why can't I merge my two volume groups my_vg1 and my_vg2? A11: A merged volume group can't go beyond the physical or logical volume limits of the destination volume group. This means for eg. that you can't merge my_vg1 with 20 logical volumes and my_vg2 with 30 logical volumes getting my_vg1, if my_vg1 has a 31 logical volume limit. You are only able to merge (currently) volume groups with equal physical extent sizes. Q12: How can I move parts of my logical volume with very intensive I/O to a different physical volume? A12: Please look at pvmove(8) and use the logical extent syntax to do the job. Q13: I have a compile problem and/or a runtime problem with the LVM. Where to ask for help? A13: Please send me a bug mail request to or to the linux-lvm mailing list. Q14: Where can I ask for a missing feature or for a better way to implement something in the LVM? A14: Please send me an enhancement mail request to or to the linux-lvm mailing list. Q15: Why is LVM not in the stock kernel? A15: LVM has been in the stock kernel since 2.3.47 Q16: Where can I send a patch for the LVM? A16: Please remember to make a unified diff to the original LVM distribution and mail the diff with your comments to Q17: Why am I not able to create my 211th logical volume? vgdisplay tells me about a limit of 256 logical volumes. A17: The total amount of 256 logical volumes is shared among all volume groups. You have to delete logical volumes in different volume groups to be able to create the new one. Q18: Can I have my root filesystem in a logical volume? A18: Yes you can. There's basic support since LVM 0.7 to create an initial RAM disk containing the necessary executables, device specials etc. to switch to a logical volume containing e root filesystem. See script lvmcreate_initrd(8). Nevertheless you have to setup the neccesary logical volumes with filesystems, /etc/lilo.conf and /etc/fstab by hand. Q19: Why does LVM not support mirroring? A19: Mirroring or other levels of RAID on a per Logical Volume base decreases reliability because you have to take care of which Logical Volumes are mirrored and which ones are not. I recommend to use dedicated hardware RAID subsystems or Multiple Devices to have the redundancy below the LVM. In this case you just don't care about Logical Volume data redundancy and you don't run into the dangerous situation that your data is not redundant by accident. If you do have these kind of devices in place you can setup RAID 10/40/50 if you want to because the LVM supports RAID0. Combining for eg. several hardware RAID5 subsystems in a volume group to set up LVM-RAID0 gives you reliability, performance and flexibility at the same time.