headers
Rob Ross | 1 Feb 2007 18:04
Favicon

Re: Error messages on client and server

it would be interesting to see an strace of this compile, see what is 
happening.

thanks,

rob

Trach-Minh Tran wrote:
> On 01/30/2007 06:48 PM, Murali Vilayannur wrote:
>> Hi,
>> On a ^C, pvfs2-client cleans up all existing flows for that particular
>> operation
>> and it also sends those warning messages to gossip/stderr.
>> They are really only warning messages in this case since it was
>> explicitly initiated by a ^C from user.
>> Perhaps, we should change some of the gossip_err() to a gossip_debug(),
>> sam?
>>
>> All that said,
>> It is a little worrying that mpif90 was taking so much time to compile
>> a hello world program..
>> Does it ever finish? Perhaps some I/O servers are done causing the stalls?
> 
> Hello,
> 
> Here is the timing of the command ... hello.f90 is a simple program of 29 lines!
> pleiades2:ttran$ time mpif90 hello.f90
> real    1m17.831s
> user    0m0.223s
> sys     0m1.097s
(Continue reading)

Permalink | Reply | 0 comments |
headers
Suneet Chandok | 2 Feb 2007 18:02
Picon

kzalloc undefined

Hi

I am using SUSE10.0 with kernel 2.6.12.6 and I have problem compiling pvfs2.6.2. When I am doing 'make kmod' it gives an error that kzalloc undefined. I think kzalloc is not supported on 2.6.12 kernel. Is there a workaround to get it worked some how. The reason I am using 2.6.12 kernel is thats the maximum supported by lustre-1.4.8.

Please let me know if any additional information is required.

--
Suneet Chandok
Research Assistant
Parallel Software Technologies Laboratory
Department of Computer Science
University of Houston

Cell: 704-248-0718
Email: schandok <at> cs.uh.edu

_______________________________________________
Pvfs2-users mailing list
Pvfs2-users <at> beowulf-underground.org
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 11 comments |
headers
Sam Lang | 2 Feb 2007 18:31
Favicon

Re: kzalloc undefined


My message just got rejected because the patch is too big.  Suneet  
you can download it here:

http://www-unix.mcs.anl.gov/~slang/kzalloc.patch

-sam

Begin forwarded message:

> From: Sam Lang <slang <at> mcs.anl.gov>
> Date: February 2, 2007 11:20:15 AM CST
> To: "Suneet Chandok" <suneet.chandok <at> gmail.com>
> Cc: pvfs2-users <at> beowulf-underground.org
> Subject: Re: [Pvfs2-users] kzalloc undefined
>
>
> Hi Suneet,
>
> This is bug in the 2.6.2 release with older linux kernels that  
> don't have kzalloc defined.  Its been fixed in CVS, and I've  
> attached a patch that can be applied to your 2.6.2 directory.
>
> Thanks,
>
> -sam
>

On Feb 2, 2007, at 11:02 AM, Suneet Chandok wrote:

> Hi
>
> I am using SUSE10.0 with kernel 2.6.12.6 and I have problem  
> compiling pvfs2.6.2. When I am doing 'make kmod' it gives an error  
> that kzalloc undefined. I think kzalloc is not supported on 2.6.12  
> kernel. Is there a workaround to get it worked some how. The reason  
> I am using 2.6.12 kernel is thats the maximum supported by  
> lustre-1.4.8.
>
> Please let me know if any additional information is required.
>
> -- 
> Suneet Chandok
> Research Assistant
> Parallel Software Technologies Laboratory
> Department of Computer Science
> University of Houston
>
> Cell: 704-248-0718
> Email: schandok <at> cs.uh.edu
> _______________________________________________
> Pvfs2-users mailing list
> Pvfs2-users <at> beowulf-underground.org
> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 0 comments |
headers
Steve | 2 Feb 2007 22:28
Picon

metadata

Hi,

Is there any rough rule of thumb guide for the size of the metadata for the
size of the store ?

Steve
Permalink | Reply | 2 comments |
headers
Sam Lang | 2 Feb 2007 23:15
Favicon

Re: metadata


On Feb 2, 2007, at 3:28 PM, Steve wrote:

> Hi,
>
> Is there any rough rule of thumb guide for the size of the metadata  
> for the
> size of the store ?

Hi Steve,

The size of the metadata as a percent of the size of the storage  
space depends on the sizes of the files.  I'm doing a back of the  
envelope calculation, but I think that in general, each file takes up  
somewhere around 650 bytes to store its metadata, while directories  
are a little less (maybe 500 bytes).  Remember though that the  
metadata size changes for different configurations (more servers,  
etc.).  There's also overhead related to the storage space itself,  
which is about (100K * numservers).  If you know how many files/ 
directories you have in your volume, you should be able to get a  
rough estimate of your metadata size using those numbers above.

-sam

>
> Steve
>
> _______________________________________________
> Pvfs2-users mailing list
> Pvfs2-users <at> beowulf-underground.org
> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
>
Permalink | Reply | 1 comment |
headers
Sam Lang | 2 Feb 2007 23:28
Favicon

Re: metadata


On Feb 2, 2007, at 4:15 PM, Sam Lang wrote:

>
> On Feb 2, 2007, at 3:28 PM, Steve wrote:
>
>> Hi,
>>
>> Is there any rough rule of thumb guide for the size of the  
>> metadata for the
>> size of the store ?
>
> Hi Steve,
>
> The size of the metadata as a percent of the size of the storage  
> space depends on the sizes of the files.  I'm doing a back of the  
> envelope calculation, but I think that in general, each file takes  
> up somewhere around 650 bytes to store its metadata, while  
> directories are a little less (maybe 500 bytes).  Remember though  
> that the metadata size changes for different configurations (more  
> servers, etc.).  There's also overhead related to the storage space  
> itself, which is about (100K * numservers).  If you know how many  
> files/directories you have in your volume, you should be able to  
> get a rough estimate of your metadata size using those numbers above.

I should clarify though, that the filesystem size (capacity and  
available) reported by df for PVFS volumes is just the sum of the  
sizes on each of the IO servers (capacity is actually the min of all  
servers * numservers).  This means that what df reports in terms of  
usage can be inaccurate if the PVFS storage on the IO nodes is shared  
with other files.

-sam

> -sam
>
>
>>
>> Steve
>>
>> _______________________________________________
>> Pvfs2-users mailing list
>> Pvfs2-users <at> beowulf-underground.org
>> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
>>
>
Permalink | Reply | 0 comments |
headers
Luiz Mendes | 3 Feb 2007 21:43
Picon

PVFS2 slow situation

Hi all,

I have installed PVFS2 in 2 PCS with Gigabit Network between them. I am doing write and read tests between NFS and PVFS2.

I developed a program in MPI that uses 2 processes to write to separate discs in PVFS2, and after this, the program measure time taken to write file until time to read this one recently created.

And i developed another program in MPI where 2 processes create a vector itself, and in this case, process 1 pass it vector to process 0, and process 0 gather itself vector and another from process 1. After gather this one, it write to a file. And later i read with only one process the whole file that have created.


I think that the program with PVFS2 should be faster than this second program that uses NFS, due to the fact of mpi application that uses PVFS2, each process write to a file, without communication.

But the result is the oposite of that.

Is this behavior correct? In what situations PVFS2 would be better than NFS?

Thanks
Luiz Mendes


_______________________________________________
Pvfs2-users mailing list
Pvfs2-users <at> beowulf-underground.org
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 0 comments |
headers
Sam Lang | 2 Feb 2007 18:20
Favicon

Re: kzalloc undefined


Hi Suneet,

This is bug in the 2.6.2 release with older linux kernels that don't  
have kzalloc defined.  Its been fixed in CVS, and I've attached a  
patch that can be applied to your 2.6.2 directory.

Thanks,

-sam
Attachment (kzalloc.patch): application/octet-stream, 643 KiB

On Feb 2, 2007, at 11:02 AM, Suneet Chandok wrote:

> Hi
>
> I am using SUSE10.0 with kernel 2.6.12.6 and I have problem  
> compiling pvfs2.6.2. When I am doing 'make kmod' it gives an error  
> that kzalloc undefined. I think kzalloc is not supported on 2.6.12  
> kernel. Is there a workaround to get it worked some how. The reason  
> I am using 2.6.12 kernel is thats the maximum supported by  
> lustre-1.4.8.
>
> Please let me know if any additional information is required.
>
> -- 
> Suneet Chandok
> Research Assistant
> Parallel Software Technologies Laboratory
> Department of Computer Science
> University of Houston
>
> Cell: 704-248-0718
> Email: schandok <at> cs.uh.edu
> _______________________________________________
> Pvfs2-users mailing list
> Pvfs2-users <at> beowulf-underground.org
> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users


_______________________________________________
Pvfs2-users mailing list
Pvfs2-users <at> beowulf-underground.org
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 9 comments |
headers
Sam Lang | 2 Feb 2007 18:26
Favicon

Fwd: kzalloc undefined


Ugh.  My message just got rejected because the patch is too big.   
Suneet you can download it here:

http://www-unix.mcs.anl.gov/~slang/kzalloc.patch

-sam

Begin forwarded message:

> From: Sam Lang <slang <at> mcs.anl.gov>
> Date: February 2, 2007 11:20:15 AM CST
> To: "Suneet Chandok" <suneet.chandok <at> gmail.com>
> Cc: pvfs2-users <at> beowulf-underground.org
> Subject: Re: [Pvfs2-users] kzalloc undefined
>
>
> Hi Suneet,
>
> This is bug in the 2.6.2 release with older linux kernels that  
> don't have kzalloc defined.  Its been fixed in CVS, and I've  
> attached a patch that can be applied to your 2.6.2 directory.
>
> Thanks,
>
> -sam
>
Attachment (kzalloc.patch): application/octet-stream, 643 KiB
>
>
> On Feb 2, 2007, at 11:02 AM, Suneet Chandok wrote:
>
>> Hi
>>
>> I am using SUSE10.0 with kernel 2.6.12.6 and I have problem  
>> compiling pvfs2.6.2. When I am doing 'make kmod' it gives an error  
>> that kzalloc undefined. I think kzalloc is not supported on 2.6.12  
>> kernel. Is there a workaround to get it worked some how. The  
>> reason I am using 2.6.12 kernel is thats the maximum supported by  
>> lustre-1.4.8.
>>
>> Please let me know if any additional information is required.
>>
>> -- 
>> Suneet Chandok
>> Research Assistant
>> Parallel Software Technologies Laboratory
>> Department of Computer Science
>> University of Houston
>>
>> Cell: 704-248-0718
>> Email: schandok <at> cs.uh.edu
>> _______________________________________________
>> Pvfs2-users mailing list
>> Pvfs2-users <at> beowulf-underground.org
>> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
>


_______________________________________________
Pvfs2-users mailing list
Pvfs2-users <at> beowulf-underground.org
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 8 comments |
headers
Luiz Mendes | 6 Feb 2007 14:19
Picon

How to make a process write to separate local disc

Hi all,

I would like to know if the hint start_iodevice works, because i want to make a benchmark where each process in different PCs writes in own local PVFS disc.

i have been trying this using hints striping_unit, start_iodevice and striping factor hints.

But is there another way to guarantee that a process writes to only one disc?

Thanks
Luiz Mendes

_______________________________________________
Pvfs2-users mailing list
Pvfs2-users <at> beowulf-underground.org
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
Permalink | Reply | 1 comment |

Gmane