Skip to contentSkip to conversation

LVM multiple Snpshots implementation query

linux-lvm andiket a 2 participants 4 messages May 10, 2010 – May 11, 2010

lore
Share & tools
lore

andiket a

author
#0 lore raw

Hi,

If there are multiple snapshots then write to origin is an expensive operation.
My understanding after looking at the code of orig_wirte() is that if a write comes on a chunk on orig device then cow is initiated for all the snapshots. This is an expensive operation for sure.

Does anyone know if there was any reason behind implementing it like this?

Thanks,
Andiket.

#1 lore raw

Did you mean you want to use something like shared-snapshot? COW once? I know shared-snapshot implemented by Mikulas and others:
http://people.redhat.com/mpatocka/patches/kernel/new-snapshots/r19/

在2010-05-10,"andiket a" <[email protected]> 写道: -----原始邮件----- 发件人:"andiket a" <[email protected]>
发送时间:2010年5月10日 星期一
收件人:[email protected]
主题:[linux-lvm] LVM multiple Snpshots implementation query

Hi,

If there are multiple snapshots then write to origin is an expensive operation.
My understanding after looking at the code of orig_wirte() is that if a write comes on a chunk on orig device then cow is initiated for all the snapshots. This is an expensive operation for sure.

Does anyone know if there was any reason behind implementing it like this?

Thanks,
Andiket.

author
#2 lore raw

Thanks a lot.
I wasnt of the patch.

Yes, I am looking for a similar thing.
Right now after snapshot the cow is done for all the snapshots if there is a write on origin.

This is an obvious thing. Wondering it was missed in original lvm/dm snapshot code.

Any idea?

--- On Tue, 11/5/10, Busby.Cheung <[email protected]> wrote:

From: Busby.Cheung <[email protected]>
Subject: Re: [linux-lvm] LVM multiple Snpshots implementation query
To: "LVM general discussion and development" <[email protected]>
Date: Tuesday, 11 May, 2010, 7:44 AM

Did you mean you want to use something like shared-snapshot? COW once? I know shared-snapshot implemented by Mikulas and others:
http://people.redhat.com/mpatocka/patches/kernel/new-snapshots/r19/


在2010-05-10,"andiket a" <[email protected]> 写道: -----原始邮件-----
发件人:"andiket a" <[email protected]> 发送时间:2010年5月10日 星期一
收件人:[email protected]
主题:[linux-lvm] LVM multiple Snpshots implementation query

Hi,

If there are multiple snapshots then write to origin is an expensive operation.
My understanding after looking at the code of orig_wirte() is that if a write comes on a chunk on orig device then cow is initiated for all the snapshots. This is an expensive operation for sure.

Does anyone know if there was any reason behind implementing it like this?

Thanks,
Andiket.

网易为中小企业免费提供企业邮箱(自主域名)

-----Inline Attachment Follows-----

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/

#3 lore raw

The shared-snapshots (http://people.redhat.com/mpatocka/patches/kernel/new-snapshots/r19/) I mentioned has the the LVM2 patch :
http://people.redhat.com/msnitzer/patches/multisnap/lvm2/

在2010-05-11,"andiket a" <[email protected]> 写道: -----原始邮件----- 发件人:"andiket a" <[email protected]>
发送时间:2010年5月11日 星期二
收件人:"LVM general discussion and development" <[email protected]> 主题:Re: [linux-lvm] LVM multiple Snpshots implementation query

Thanks a lot.
I wasnt of the patch.

Yes, I am looking for a similar thing.
Right now after snapshot the cow is done for all the snapshots if there is a write on origin.

This is an obvious thing. Wondering it was missed in original lvm/dm snapshot code.

Any idea?

--- OnTue, 11/5/10, Busby.Cheung<[email protected]>wrote:

From: Busby.Cheung <[email protected]>
Subject: Re: [linux-lvm] LVM multiple Snpshots implementation query
To: "LVM general discussion and development" <[email protected]>
Date: Tuesday, 11 May, 2010, 7:44 AM

Did you mean you want to use something like shared-snapshot? COW once? I know shared-snapshot implemented by Mikulas and others:
http://people.redhat.com/mpatocka/patches/kernel/new-snapshots/r19/

在2010-05-10,"andiket a" <[email protected]> 写道: -----原始邮件----- 发件人:"andiket a" <[email protected]>
发送时间:2010年5月10日 星期一
收件人:[email protected]
主题:[linux-lvm] LVM multiple Snpshots implementation query

Hi,

If there are multiple snapshots then write to origin is an expensive operation.
My understanding after looking at the code of orig_wirte() is that if a write comes on a chunk on orig device then cow is initiated for all the snapshots. This is an expensive operation for sure.

Does anyone know if there was any reason behind implementing it like this?

Thanks,
Andiket.

网易为中小企业免费提供企业邮箱(自主域名)

-----Inline Attachment Follows-----

_______________________________________________
linux-lvm mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO athttp://tldp.org/HOWTO/LVM-HOWTO/