questions
Hi
Is there a limit for the maximum size of a LVM diskgroup? (internally in LVM-code, or because of other limits in the kernel) The 2TB (1 TB) maximum devicesize in the linux-kernel doesn't apply, does it?
How does LVM handle that one of it's devices changes size? E.g. if one of the physical disks is a RAIDset, and more disks are added to it? Is there any way to tell LVM to use more or less of the disk?
Thanks.
Signature · 2 lines
Ragnar Kj�rstad BigStorage
Quoted 6 lines · Ragnar Kj�rstad
Hi
Is there a limit for the maximum size of a LVM diskgroup? (internally in LVM-code, or because of other limits in the kernel) The 2TB (1 TB) maximum devicesize in the linux-kernel doesn't apply, does it?
256 (which is limited to 128 SCSI disks and 20? IDE discs) * maximum PV size which is limited to 2 TB --> 296 TB in case you'll find 2 TB IDE disks ;-)
Quoted 4 lines
How does LVM handle that one of it's devices changes size? E.g. if one of the physical disks is a RAIDset, and more disks are added to it? Is there any way to tell LVM to use more or less of the disk?
LVM can't handle size changes directly today.
You need to go with
a. additional partitions on the size increased device b. additional LUNs exposing the added capacity (typical with some
HW RAID device models)
and create additional PVs on them.
Basically we could implement an enhancement to pvchange(8) in order to support growing of PVs.
Quoted 12 lines
Thanks.
Signature · 6 lines
Ragnar Kj�rstad BigStorage _______________________________________________ linux-lvm mailing list [email protected] http://lists.sistina.com/mailman/listinfo/linux-lvm
Signature · 15 lines
Regards,
Heinz -- The LVM Guy --
*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Heinz Mauelshagen Sistina Software Inc.
Senior Consultant/Developer Am Sonnenhang 11
56242 Marienrachdorf
Germany
[email protected] +49 2626 141200
FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-Quoted 4 lines · Ragnar Kj_rstad
Is there a limit for the maximum size of a LVM diskgroup? (internally in LVM-code, or because of other limits in the kernel) The 2TB (1 TB) maximum devicesize in the linux-kernel doesn't apply, does it?
Yes, even LVM has the 2TB limit at this point. This is because LVM presents itself to the rest of the kernel as a block device (just like any other), so the kernel 2TB limit for block devices still holds.
Heinz talked about allowing LVM to have PAGE_SIZE (4k) blocks (unlike the rest of the kernel, which uses 512 byte blocks) so it may be possible to go up to 16TB with LVM without a huge amount of kernel redesign.
Cheers, Andreas
Signature · 3 lines
Andreas Dilger \ "If a man ate a pound of pasta and a pound of antipasto,
\ would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/ -- DogbertQuoted 9 lines · Andreas Dilger
Quoted 4 lines · Ragnar Kj_rstad
Is there a limit for the maximum size of a LVM diskgroup? (internally in LVM-code, or because of other limits in the kernel) The 2TB (1 TB) maximum devicesize in the linux-kernel doesn't apply, does it?
Yes, even LVM has the 2TB limit at this point. This is because LVM presents itself to the rest of the kernel as a block device (just like any other), so the kernel 2TB limit for block devices still holds.
For the volume, but not for the diskgroup, right?
Quoted 3 lines
Heinz talked about allowing LVM to have
PAGE_SIZE(4k) blocks (unlike the rest of the kernel, which uses 512 byte blocks) so it may be possible to go up to 16TB with LVM without a huge amount of kernel redesign.
This sounds really interesting.
What exactly would be needed to do that?
I believe there have been issues with lvm block sizes and XFS (maybe other filesystems too). Would changing it to 4k create new issues? Would it be possible to increese it even futher? if the filesystem used more than 4k blocksize?
Thanks.
Signature · 2 lines
Ragnar Kj�rstad BigStorage
Quoted 6 lines · Ragnar Kj_rstad
Quoted 3 lines · Andreas Dilger
Yes, even LVM has the 2TB limit at this point. This is because LVM presents itself to the rest of the kernel as a block device (just like any other), so the kernel 2TB limit for block devices still holds.
For the volume, but not for the diskgroup, right?
Yes, the 2TB limit is for a single LV. You can have up to 255 LVs in total I think, so 512TB total usable space with LVM, assuming your configuration works in this way. Probably the 255 LV limit will be a problem before the 512TB limit is...
Quoted 5 lines
Quoted 3 lines
Heinz talked about allowing LVM to have
PAGE_SIZE(4k) blocks (unlike the rest of the kernel, which uses 512 byte blocks) so it may be possible to go up to 16TB with LVM without a huge amount of kernel redesign.This sounds really interesting. What exactly would be needed to do that?
Change all of the code paths in the kernel that assume block size is 512 bytes...
Quoted 2 lines
I believe there have been issues with lvm block sizes and XFS (maybe other filesystems too). Would changing it to 4k create new issues?
Yes, probably.
Quoted 2 lines
Would it be possible to increese it even futher? if the filesystem used more than 4k blocksize?
Probably not. There will likely be lots of problems for block size > PAGE_SIZE. Of course on ia64 and alpha PAGE_SIZE = 8kB, so you can go to 32TB I think, assuming the rest of the issues are fixed.
Cheers, Andreas
Signature · 3 lines
Andreas Dilger \ "If a man ate a pound of pasta and a pound of antipasto,
\ would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/ -- DogbertQuoted 2 lines · Andreas Dilger
Change all of the code paths in the kernel that assume block size is 512 bytes...
IIRC some S/390 devices already use 4K block size.
Christoph
Signature · 1 line
Of course it doesn't work. We've performed a software upgrade.
Quoted 5 lines · Christoph Hellwig
Quoted 2 lines · Andreas Dilger
Change all of the code paths in the kernel that assume block size is 512 bytes...
IIRC some S/390 devices already use 4K block size.
Yes. They have so called DASD drives which are 4K hard sectored.
Anyway. if you change the BLOCK_SIZE definition in lvm.h from 1 to 4k you can build LVs larger than 2TB (or even 4TB which is the default with the 1k block size used by LVM today).
You can't go beyond PAGE_SIZE though because a whole pile of code in the kernel doesn't support that.
That means that you could even go to 8k block size on machines supporting that page size.
Remember: you loose comapatibility with 4k page sized systems in this case.
Quoted 9 lines
Christoph
Signature · 5 lines
Of course it doesn't work. We've performed a software upgrade. _______________________________________________ linux-lvm mailing list [email protected] http://lists.sistina.com/mailman/listinfo/linux-lvm
Signature · 15 lines
Regards,
Heinz -- The LVM Guy --
*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Heinz Mauelshagen Sistina Software Inc.
Senior Consultant/Developer Am Sonnenhang 11
56242 Marienrachdorf
Germany
[email protected] +49 2626 141200
FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-