thr3ads.net - Lustre devel - [Lustre-devel] Control file striping [Nov 2012]

If this information is useful, please help other people find it:
Share via:

Kalyana C

2012-Nov-03 17:41 UTC

[Lustre-devel] Control file striping

Hello,

I am trying to figure out a way to:

set striping across a set of OSTs of my choice.

lfs setstripe command and API call does not allow for this currently. The
user can only set the starting OST through the index parameter. I
appreciate why there is no such feature currently; you do not want to let
the users have control on how OSTs are used since it can result in OST
usage imbalance.

However, I do have a valid usage case for such a feature and I am trying to
figure out the best way to do it.

One way to achieve the same result is through the use of OST pools.
However, the userland tools do not work from a client node, it has to be
run on the MGS. There are also other restrictions as described here:
http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools

I am also digging through the Lustre source and header files to see if
there are other (undocumented) ways to do this.

So, my questions are as follows:

1) Is there currently a way for a user to control the striping of a file /
directory on a selected set of OSTs? If yes, is it possible to do it
through an API call or a command line utility?

2) If answer to (1) is No, Is there a way to do this with admin rights from
the MDS or client nodes?

3) If pools are the only way to achieve this, is there an API call to
create, manage and destroy pools from a user application? Is there any
impact on performance of MDS components if pools are dynamically created
and deleted on demand? Can the API be called from a client node (root or
user) or does it have to be on the MGS?

Thank you.

Regards,
Kalyana Chadalavada
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-devel/attachments/20121103/1884bff0/attachment.html

Atchley, Scott

2012-Nov-05 17:47 UTC

head link

[Lustre-devel] Control file striping

On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com> wrote:
> Hello,
> 
> I am trying to figure out a way to:
> 
> set striping across a set of OSTs of my choice. 
> 
> lfs setstripe command and API call does not allow for this currently. The
user can only set the starting OST through the index parameter. I appreciate why
there is no such feature currently; you do not want to let the users have
control on how OSTs are used since it can result in OST usage imbalance.
> 
> However, I do have a valid usage case for such a feature and I am trying to
figure out the best way to do it.
I have a valid use for it as well. It would be nice to have this feature back.

Scott
> 
> One way to achieve the same result is through the use of OST pools.
However, the userland tools do not work from a client node, it has to be run on
the MGS. There are also other restrictions as described here:
http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
> 
> I am also digging through the Lustre source and header files to see if
there are other (undocumented) ways to do this.
> 
> So, my questions are as follows:
> 
> 1) Is there currently a way for a user to control the striping of a file /
directory on a selected set of OSTs? If yes, is it possible to do it through an
API call or a command line utility?
> 
> 2) If answer to (1) is No, Is there a way to do this with admin rights from
the MDS or client nodes?
> 
> 3) If pools are the only way to achieve this, is there an API call to
create, manage and destroy pools from a user application? Is there any impact on
performance of MDS components if pools are dynamically created and deleted on
demand? Can the API be called from a client node (root or user) or does it have
to be on the MGS?
> 
> Thank you.
> 
> Regards,
> Kalyana Chadalavada 
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel

Ben Evans

2012-Nov-05 17:57 UTC

head link

[Lustre-devel] Control file striping

Have you considered using a combination of OST Pools and striping fit
your needs?

-----Original Message-----
From: lustre-devel-bounces at lists.lustre.org
[mailto:lustre-devel-bounces at lists.lustre.org] On Behalf Of Atchley,
Scott
Sent: Monday, November 05, 2012 12:47 PM
To: Kalyana C
Cc: lustre-devel at lists.lustre.org
Subject: Re: [Lustre-devel] Control file striping

On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com> wrote:
> Hello,
> 
> I am trying to figure out a way to:
> 
> set striping across a set of OSTs of my choice. 
> 
> lfs setstripe command and API call does not allow for this currently.The user can only set the starting OST through the index parameter. I
appreciate why there is no such feature currently; you do not want to
let the users have control on how OSTs are used since it can result in
OST usage imbalance. > 
> However, I do have a valid usage case for such a feature and I amtrying to figure out the best way to do it. 

I have a valid use for it as well. It would be nice to have this feature
back.

Scott
> 
> One way to achieve the same result is through the use of OST pools.However, the userland tools do not work from a client node, it has to be
run on the MGS. There are also other restrictions as described here:
http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools> 
> I am also digging through the Lustre source and header files to see if
there are other (undocumented) ways to do this. > 
> So, my questions are as follows:
> 
> 1) Is there currently a way for a user to control the striping of afile / directory on a selected set of OSTs? If yes, is it possible to do
it through an API call or a command line utility?> 
> 2) If answer to (1) is No, Is there a way to do this with admin rights
from the MDS or client nodes? > 
> 3) If pools are the only way to achieve this, is there an API call tocreate, manage and destroy pools from a user application? Is there any
impact on performance of MDS components if pools are dynamically created
and deleted on demand? Can the API be called from a client node (root or
user) or does it have to be on the MGS?> 
> Thank you.
> 
> Regards,
> Kalyana Chadalavada 
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel
_______________________________________________
Lustre-devel mailing list
Lustre-devel at lists.lustre.org
http://lists.lustre.org/mailman/listinfo/lustre-devel

Kalyana C

2012-Nov-05 18:01 UTC

head link

[Lustre-devel] Control file striping

> Have you considered using a combination of OST Pools and stripingfit your needs?

This is the only way possible that I currently know of.

But is it possible to manage OST pools via an API call? Can it be done from
a client under a normal user application? Or does it have the same
restrictions as described at
http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools

Regards,
Kalyan

On Mon, Nov 5, 2012 at 11:57 AM, Ben Evans <Ben.Evans at terascala.com>
wrote:
> Have you considered using a combination of OST Pools and striping fit
> your needs?
>
> -----Original Message-----
> From: lustre-devel-bounces at lists.lustre.org
> [mailto:lustre-devel-bounces at lists.lustre.org] On Behalf Of Atchley,
> Scott
> Sent: Monday, November 05, 2012 12:47 PM
> To: Kalyana C
> Cc: lustre-devel at lists.lustre.org
> Subject: Re: [Lustre-devel] Control file striping
>
> On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com> wrote:
>
> > Hello,
> >
> > I am trying to figure out a way to:
> >
> > set striping across a set of OSTs of my choice.
> >
> > lfs setstripe command and API call does not allow for this currently.
> The user can only set the starting OST through the index parameter. I
> appreciate why there is no such feature currently; you do not want to
> let the users have control on how OSTs are used since it can result in
> OST usage imbalance.
> >
> > However, I do have a valid usage case for such a feature and I am
> trying to figure out the best way to do it.
>
> I have a valid use for it as well. It would be nice to have this feature
> back.
>
> Scott
>
> >
> > One way to achieve the same result is through the use of OST pools.
> However, the userland tools do not work from a client node, it has to be
> run on the MGS. There are also other restrictions as described here:
> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
> >
> > I am also digging through the Lustre source and header files to see if
> there are other (undocumented) ways to do this.
> >
> > So, my questions are as follows:
> >
> > 1) Is there currently a way for a user to control the striping of a
> file / directory on a selected set of OSTs? If yes, is it possible to do
> it through an API call or a command line utility?
> >
> > 2) If answer to (1) is No, Is there a way to do this with admin rights
> from the MDS or client nodes?
> >
> > 3) If pools are the only way to achieve this, is there an API call to
> create, manage and destroy pools from a user application? Is there any
> impact on performance of MDS components if pools are dynamically created
> and deleted on demand? Can the API be called from a client node (root or
> user) or does it have to be on the MGS?
> >
> > Thank you.
> >
> > Regards,
> > Kalyana Chadalavada
> >
> > _______________________________________________
> > Lustre-devel mailing list
> > Lustre-devel at lists.lustre.org
> > http://lists.lustre.org/mailman/listinfo/lustre-devel
>
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel
>-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-devel/attachments/20121105/3f279f1b/attachment.html

White, Cliff

2012-Nov-05 18:16 UTC

head link

[Lustre-devel] Control file striping

Creating/destroying pools must be done as per the manual, from MDS.
Do you need to frequently create/destroy pools, or would you have a fixed setup?

The normal method is to map various pools (and other striping policies) to
directories, then user space applications
can access specific pools by pointing at the requisite directory, thus requiring
no special API.
Once pools are created, the directory creation/striping is all done on a client
via lfs setstripe, the liblustreapi can
also be used by clients to set striping via C calls.

Best
cliffw


On Nov 5, 2012, at 10:01 AM, Kalyana C <ckkccf at gmail.com<mailto:ckkccf
at gmail.com>>
 wrote:
> Have you considered using a combination of OST Pools and striping fit your
needs?
This is the only way possible that I currently know of.

But is it possible to manage OST pools via an API call? Can it be done from a
client under a normal user application? Or does it have the same restrictions as
described at http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools

Regards,
Kalyan

On Mon, Nov 5, 2012 at 11:57 AM, Ben Evans <Ben.Evans at
terascala.com<mailto:Ben.Evans at terascala.com>> wrote:
Have you considered using a combination of OST Pools and striping fit
your needs?

-----Original Message-----
From: lustre-devel-bounces at lists.lustre.org<mailto:lustre-devel-bounces at
lists.lustre.org>
[mailto:lustre-devel-bounces at lists.lustre.org<mailto:lustre-devel-bounces
at lists.lustre.org>] On Behalf Of Atchley,
Scott
Sent: Monday, November 05, 2012 12:47 PM
To: Kalyana C
Cc: lustre-devel at lists.lustre.org<mailto:lustre-devel at
lists.lustre.org>
Subject: Re: [Lustre-devel] Control file striping

On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com<mailto:ckkccf
at gmail.com>> wrote:
> Hello,
>
> I am trying to figure out a way to:
>
> set striping across a set of OSTs of my choice.
>
> lfs setstripe command and API call does not allow for this currently.The user can only set the starting OST through the index parameter. I
appreciate why there is no such feature currently; you do not want to
let the users have control on how OSTs are used since it can result in
OST usage imbalance.>
> However, I do have a valid usage case for such a feature and I amtrying to figure out the best way to do it.

I have a valid use for it as well. It would be nice to have this feature
back.

Scott
>
> One way to achieve the same result is through the use of OST pools.However, the userland tools do not work from a client node, it has to be
run on the MGS. There are also other restrictions as described here:
http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools>
> I am also digging through the Lustre source and header files to see if
there are other (undocumented) ways to do this.>
> So, my questions are as follows:
>
> 1) Is there currently a way for a user to control the striping of afile / directory on a selected set of OSTs? If yes, is it possible to do
it through an API call or a command line utility?>
> 2) If answer to (1) is No, Is there a way to do this with admin rights
from the MDS or client nodes?>
> 3) If pools are the only way to achieve this, is there an API call tocreate, manage and destroy pools from a user application? Is there any
impact on performance of MDS components if pools are dynamically created
and deleted on demand? Can the API be called from a client node (root or
user) or does it have to be on the MGS?>
> Thank you.
>
> Regards,
> Kalyana Chadalavada
>
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org<mailto:Lustre-devel at
lists.lustre.org>
> http://lists.lustre.org/mailman/listinfo/lustre-devel
_______________________________________________
Lustre-devel mailing list
Lustre-devel at lists.lustre.org<mailto:Lustre-devel at lists.lustre.org>
http://lists.lustre.org/mailman/listinfo/lustre-devel

_______________________________________________
Lustre-devel mailing list
Lustre-devel at lists.lustre.org<mailto:Lustre-devel at lists.lustre.org>
http://lists.lustre.org/mailman/listinfo/lustre-devel

Atchley, Scott

2012-Nov-05 18:28 UTC

head link

[Lustre-devel] Control file striping

On Nov 5, 2012, at 1:16 PM, "White, Cliff" <cliff.white at
intel.com> wrote:
> Creating/destroying pools must be done as per the manual, from MDS.
> Do you need to frequently create/destroy pools, or would you have a fixed
setup?
In our case, we need to frequently change the OSTs. Possibly per file.

Scott
> 
> The normal method is to map various pools (and other striping policies) to
directories, then user space applications
> can access specific pools by pointing at the requisite directory, thus
requiring no special API.
> Once pools are created, the directory creation/striping is all done on a
client via lfs setstripe, the liblustreapi can
> also be used by clients to set striping via C calls.
> 
> Best
> cliffw
> 
> 
> On Nov 5, 2012, at 10:01 AM, Kalyana C <ckkccf at
gmail.com<mailto:ckkccf at gmail.com>>
> wrote:
> 
>> Have you considered using a combination of OST Pools and striping fit
your needs?
> 
> This is the only way possible that I currently know of.
> 
> But is it possible to manage OST pools via an API call? Can it be done from
a client under a normal user application? Or does it have the same restrictions
as described at http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
> 
> Regards,
> Kalyan
> 
> On Mon, Nov 5, 2012 at 11:57 AM, Ben Evans <Ben.Evans at
terascala.com<mailto:Ben.Evans at terascala.com>> wrote:
> Have you considered using a combination of OST Pools and striping fit
> your needs?
> 
> -----Original Message-----
> From: lustre-devel-bounces at
lists.lustre.org<mailto:lustre-devel-bounces at lists.lustre.org>
> [mailto:lustre-devel-bounces at
lists.lustre.org<mailto:lustre-devel-bounces at lists.lustre.org>] On
Behalf Of Atchley,
> Scott
> Sent: Monday, November 05, 2012 12:47 PM
> To: Kalyana C
> Cc: lustre-devel at lists.lustre.org<mailto:lustre-devel at
lists.lustre.org>
> Subject: Re: [Lustre-devel] Control file striping
> 
> On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at
gmail.com<mailto:ckkccf at gmail.com>> wrote:
> 
>> Hello,
>> 
>> I am trying to figure out a way to:
>> 
>> set striping across a set of OSTs of my choice.
>> 
>> lfs setstripe command and API call does not allow for this currently.
> The user can only set the starting OST through the index parameter. I
> appreciate why there is no such feature currently; you do not want to
> let the users have control on how OSTs are used since it can result in
> OST usage imbalance.
>> 
>> However, I do have a valid usage case for such a feature and I am
> trying to figure out the best way to do it.
> 
> I have a valid use for it as well. It would be nice to have this feature
> back.
> 
> Scott
> 
>> 
>> One way to achieve the same result is through the use of OST pools.
> However, the userland tools do not work from a client node, it has to be
> run on the MGS. There are also other restrictions as described here:
> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
>> 
>> I am also digging through the Lustre source and header files to see if
> there are other (undocumented) ways to do this.
>> 
>> So, my questions are as follows:
>> 
>> 1) Is there currently a way for a user to control the striping of a
> file / directory on a selected set of OSTs? If yes, is it possible to do
> it through an API call or a command line utility?
>> 
>> 2) If answer to (1) is No, Is there a way to do this with admin rights
> from the MDS or client nodes?
>> 
>> 3) If pools are the only way to achieve this, is there an API call to
> create, manage and destroy pools from a user application? Is there any
> impact on performance of MDS components if pools are dynamically created
> and deleted on demand? Can the API be called from a client node (root or
> user) or does it have to be on the MGS?
>> 
>> Thank you.
>> 
>> Regards,
>> Kalyana Chadalavada
>> 
>> _______________________________________________
>> Lustre-devel mailing list
>> Lustre-devel at lists.lustre.org<mailto:Lustre-devel at
lists.lustre.org>
>> http://lists.lustre.org/mailman/listinfo/lustre-devel
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org<mailto:Lustre-devel at
lists.lustre.org>
> http://lists.lustre.org/mailman/listinfo/lustre-devel
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org<mailto:Lustre-devel at
lists.lustre.org>
> http://lists.lustre.org/mailman/listinfo/lustre-devel
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel

Kalyana C

2012-Nov-05 18:28 UTC

head link

[Lustre-devel] Control file striping

I will need to setup striping on specific OSTs on demand. Its not a fixed
setup.

Here is what I am thinking, with my current understanding:

*) user app sends the details (OST ids, name of the pool) to a listener on
MGS
*) a process on MGS executes the command and sends back an ack
*) user app sets up striping on the newly created pool
*) pool is deleted once striping is set

I was hoping to find out if there was a better (and easier) way to do this.

A simple usage case is as follows:
There are a mix of disks with varying performance in the file system.
Depending on the application, I would like to change the disks used for
that instance. Say, 10% spindle, 90% SSD, 50-50 for another etc.

The same can be applied to a mix of storage devices in the same file system
- disks with varying levels of redundancy or even varying devices.

Regards,
Kalyan

On Mon, Nov 5, 2012 at 12:16 PM, White, Cliff <cliff.white at intel.com>
wrote:
>  Creating/destroying pools must be done as per the manual, from MDS.
> Do you need to frequently create/destroy pools, or would you have a fixed
> setup?
>
>  The normal method is to map various pools (and other striping policies)
> to directories, then user space applications
> can access specific pools by pointing at the requisite directory, thus
> requiring no special API.
> Once pools are created, the directory creation/striping is all done on a
> client via lfs setstripe, the liblustreapi can
> also be used by clients to set striping via C calls.
>
>  Best
> cliffw
>
>
>  On Nov 5, 2012, at 10:01 AM, Kalyana C <ckkccf at gmail.com>
>  wrote:
>
>  > Have you considered using a combination of OST Pools and striping
> fit your needs?
>
>  This is the only way possible that I currently know of.
>
>  But is it possible to manage OST pools via an API call? Can it be done
> from a client under a normal user application? Or does it have the same
> restrictions as described at
> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
>
>  Regards,
> Kalyan
>
> On Mon, Nov 5, 2012 at 11:57 AM, Ben Evans <Ben.Evans at
terascala.com>wrote:
>
>> Have you considered using a combination of OST Pools and striping fit
>> your needs?
>>
>> -----Original Message-----
>> From: lustre-devel-bounces at lists.lustre.org
>> [mailto:lustre-devel-bounces at lists.lustre.org] On Behalf Of Atchley,
>> Scott
>> Sent: Monday, November 05, 2012 12:47 PM
>> To: Kalyana C
>> Cc: lustre-devel at lists.lustre.org
>> Subject: Re: [Lustre-devel] Control file striping
>>
>> On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com>
wrote:
>>
>> > Hello,
>> >
>> > I am trying to figure out a way to:
>> >
>> > set striping across a set of OSTs of my choice.
>> >
>> > lfs setstripe command and API call does not allow for this
currently.
>> The user can only set the starting OST through the index parameter. I
>> appreciate why there is no such feature currently; you do not want to
>> let the users have control on how OSTs are used since it can result in
>> OST usage imbalance.
>> >
>> > However, I do have a valid usage case for such a feature and I am
>> trying to figure out the best way to do it.
>>
>> I have a valid use for it as well. It would be nice to have this
feature
>> back.
>>
>> Scott
>>
>> >
>> > One way to achieve the same result is through the use of OST
pools.
>> However, the userland tools do not work from a client node, it has to
be
>> run on the MGS. There are also other restrictions as described here:
>> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
>> >
>> > I am also digging through the Lustre source and header files to
see if
>> there are other (undocumented) ways to do this.
>> >
>> > So, my questions are as follows:
>> >
>> > 1) Is there currently a way for a user to control the striping of
a
>> file / directory on a selected set of OSTs? If yes, is it possible to
do
>> it through an API call or a command line utility?
>> >
>> > 2) If answer to (1) is No, Is there a way to do this with admin
rights
>> from the MDS or client nodes?
>> >
>> > 3) If pools are the only way to achieve this, is there an API call
to
>> create, manage and destroy pools from a user application? Is there any
>> impact on performance of MDS components if pools are dynamically
created
>> and deleted on demand? Can the API be called from a client node (root
or
>> user) or does it have to be on the MGS?
>> >
>> > Thank you.
>> >
>> > Regards,
>> > Kalyana Chadalavada
>> >
>> > _______________________________________________
>> > Lustre-devel mailing list
>> > Lustre-devel at lists.lustre.org
>> > http://lists.lustre.org/mailman/listinfo/lustre-devel
>>
>> _______________________________________________
>> Lustre-devel mailing list
>> Lustre-devel at lists.lustre.org
>> http://lists.lustre.org/mailman/listinfo/lustre-devel
>>
>
>  _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel
>
>
>-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-devel/attachments/20121105/86822b0b/attachment.html

Christopher J. Morrone

2012-Nov-05 18:42 UTC

head link

[Lustre-devel] Control file striping

Pools only address one kind of problem:  constraining object allocation 
to a set of OSTs.

But what if the user wants to guarantee that their files'' objects are 
evenly distributed across the OSTs, whether in a pool or not?  Even a 
slight unevenness in object allocation (one OST getting more objects 
than the others) can drastically change the apparent aggregate 
throughput that the user sees.

Most of the applications that I am aware of essentially do:

   compute
   barrier
   write files
   barrier
   repeat

So any unevenness of object allocation means that the entire application 
sits and waits for that one OST that is more heavily loaded than the rest.

Which generally means for us that Lustre''s object allocation algorithms
that attempt to provide space-balancing have noticeably negative impacts 
on application performance.

So I agree that I would like to see the set-stripe API and wire 
protocols expanded to allow specifying all of the desired object 
indices, not just the first one.

It would probably be good to make the use of the ability to manually set 
object locations permission-based, and allow the sysadmin to decide if 
users can use that functionality.  Best would be if they could set it on 
a per-user basis, but I''d settle for a single large on/off switch.

Chris

On 11/05/2012 09:57 AM, Ben Evans wrote:> Have you considered using a combination of OST Pools and striping fit
> your needs?
>
> -----Original Message-----
> From: lustre-devel-bounces at lists.lustre.org
> [mailto:lustre-devel-bounces at lists.lustre.org] On Behalf Of Atchley,
> Scott
> Sent: Monday, November 05, 2012 12:47 PM
> To: Kalyana C
> Cc: lustre-devel at lists.lustre.org
> Subject: Re: [Lustre-devel] Control file striping
>
> On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com> wrote:
>
>> Hello,
>>
>> I am trying to figure out a way to:
>>
>> set striping across a set of OSTs of my choice.
>>
>> lfs setstripe command and API call does not allow for this currently.
> The user can only set the starting OST through the index parameter. I
> appreciate why there is no such feature currently; you do not want to
> let the users have control on how OSTs are used since it can result in
> OST usage imbalance.
>>
>> However, I do have a valid usage case for such a feature and I am
> trying to figure out the best way to do it.
>
> I have a valid use for it as well. It would be nice to have this feature
> back.
>
> Scott
>
>>
>> One way to achieve the same result is through the use of OST pools.
> However, the userland tools do not work from a client node, it has to be
> run on the MGS. There are also other restrictions as described here:
> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
>>
>> I am also digging through the Lustre source and header files to see if
> there are other (undocumented) ways to do this.
>>
>> So, my questions are as follows:
>>
>> 1) Is there currently a way for a user to control the striping of a
> file / directory on a selected set of OSTs? If yes, is it possible to do
> it through an API call or a command line utility?
>>
>> 2) If answer to (1) is No, Is there a way to do this with admin rights
> from the MDS or client nodes?
>>
>> 3) If pools are the only way to achieve this, is there an API call to
> create, manage and destroy pools from a user application? Is there any
> impact on performance of MDS components if pools are dynamically created
> and deleted on demand? Can the API be called from a client node (root or
> user) or does it have to be on the MGS?
>>
>> Thank you.
>>
>> Regards,
>> Kalyana Chadalavada
>>
>> _______________________________________________
>> Lustre-devel mailing list
>> Lustre-devel at lists.lustre.org
>> http://lists.lustre.org/mailman/listinfo/lustre-devel
>
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel
> .
>

Atchley, Scott

2012-Nov-05 19:00 UTC

head link

[Lustre-devel] Control file striping

On Nov 5, 2012, at 1:42 PM, Christopher J. Morrone <morrone2 at llnl.gov>
wrote:
> Pools only address one kind of problem:  constraining object allocation 
> to a set of OSTs.
> 
> But what if the user wants to guarantee that their files'' objects
are
> evenly distributed across the OSTs, whether in a pool or not?  Even a 
> slight unevenness in object allocation (one OST getting more objects 
> than the others) can drastically change the apparent aggregate 
> throughput that the user sees.
> 
> Most of the applications that I am aware of essentially do:
> 
>   compute
>   barrier
>   write files
>   barrier
>   repeat
> 
> So any unevenness of object allocation means that the entire application 
> sits and waits for that one OST that is more heavily loaded than the rest.
> 
> Which generally means for us that Lustre''s object allocation
algorithms
> that attempt to provide space-balancing have noticeably negative impacts 
> on application performance.
> 
> So I agree that I would like to see the set-stripe API and wire 
> protocols expanded to allow specifying all of the desired object 
> indices, not just the first one.
> 
> It would probably be good to make the use of the ability to manually set 
> object locations permission-based, and allow the sysadmin to decide if 
> users can use that functionality.  Best would be if they could set it on 
> a per-user basis, but I''d settle for a single large on/off switch.
> 
> Chris
Ditto.

Scott
> 
> On 11/05/2012 09:57 AM, Ben Evans wrote:
>> Have you considered using a combination of OST Pools and striping fit
>> your needs?
>> 
>> -----Original Message-----
>> From: lustre-devel-bounces at lists.lustre.org
>> [mailto:lustre-devel-bounces at lists.lustre.org] On Behalf Of Atchley,
>> Scott
>> Sent: Monday, November 05, 2012 12:47 PM
>> To: Kalyana C
>> Cc: lustre-devel at lists.lustre.org
>> Subject: Re: [Lustre-devel] Control file striping
>> 
>> On Nov 3, 2012, at 1:41 PM, Kalyana C <ckkccf at gmail.com>
wrote:
>> 
>>> Hello,
>>> 
>>> I am trying to figure out a way to:
>>> 
>>> set striping across a set of OSTs of my choice.
>>> 
>>> lfs setstripe command and API call does not allow for this
currently.
>> The user can only set the starting OST through the index parameter. I
>> appreciate why there is no such feature currently; you do not want to
>> let the users have control on how OSTs are used since it can result in
>> OST usage imbalance.
>>> 
>>> However, I do have a valid usage case for such a feature and I am
>> trying to figure out the best way to do it.
>> 
>> I have a valid use for it as well. It would be nice to have this
feature
>> back.
>> 
>> Scott
>> 
>>> 
>>> One way to achieve the same result is through the use of OST pools.
>> However, the userland tools do not work from a client node, it has to
be
>> run on the MGS. There are also other restrictions as described here:
>> http://wiki.lustre.org/index.php/Creating_and_Managing_OST_Pools
>>> 
>>> I am also digging through the Lustre source and header files to see
if
>> there are other (undocumented) ways to do this.
>>> 
>>> So, my questions are as follows:
>>> 
>>> 1) Is there currently a way for a user to control the striping of a
>> file / directory on a selected set of OSTs? If yes, is it possible to
do
>> it through an API call or a command line utility?
>>> 
>>> 2) If answer to (1) is No, Is there a way to do this with admin
rights
>> from the MDS or client nodes?
>>> 
>>> 3) If pools are the only way to achieve this, is there an API call
to
>> create, manage and destroy pools from a user application? Is there any
>> impact on performance of MDS components if pools are dynamically
created
>> and deleted on demand? Can the API be called from a client node (root
or
>> user) or does it have to be on the MGS?
>>> 
>>> Thank you.
>>> 
>>> Regards,
>>> Kalyana Chadalavada
>>> 
>>> _______________________________________________
>>> Lustre-devel mailing list
>>> Lustre-devel at lists.lustre.org
>>> http://lists.lustre.org/mailman/listinfo/lustre-devel
>> 
>> _______________________________________________
>> Lustre-devel mailing list
>> Lustre-devel at lists.lustre.org
>> http://lists.lustre.org/mailman/listinfo/lustre-devel
>> _______________________________________________
>> Lustre-devel mailing list
>> Lustre-devel at lists.lustre.org
>> http://lists.lustre.org/mailman/listinfo/lustre-devel
>> .
>> 
> 
> _______________________________________________
> Lustre-devel mailing list
> Lustre-devel at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-devel

Kalyana C

2012-Nov-05 19:41 UTC

head link

[Lustre-devel] Control file striping

> So I agree that I would like to see the set-stripe API and wire
> protocols expanded to allow specifying all of the desired object
> indices, not just the first one.
> It would probably be good to make the use of the ability to manually set
> object locations permission-based, and allow the sysadmin to decide if
> users can use that functionality.  Best would be if they could set it on
> a per-user basis, but I''d settle for a single large on/off switch.

This would be nice to have.


Regards,
Kalyan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-devel/attachments/20121105/58f8aed9/attachment.html

Lustre devel - Nov 2012 - Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping

[Lustre-devel] Control file striping