Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[AndyLiebman]

> neilb@cse.unsw.edu.au writes:
>
> Hmmm... I might work with 1.6 if you --build that raid0 (rather than
> --create),  or if you are using a 64bit processor like alpha or ia64. 
> But mdadm 1.6 definitely have problems in i386 processors when you try
> to --create a raid0 with devices larger than 2 terabytes.

I have been using "mdadm -Cv /dev/md0 -n2 -l0 -c[256 or 512] /dev/sd{a,b}" 
without incident. But I'll switch over to 1.8. 

Was the previous problem only one of creating the arrays? Or could there be 
further problems down the line with my arrays that were created with 1.6? 

By the way, do you know anybody with enough understanding of the Linux kernel 
to be able to rationally and logically tune the following items, which all 
relate to one another and MUST have some optimum settings for low-latency/high 
throughput over a network. 

* chunk size/stripe size of Hardware and Software arrays
* blockdev --setra (for each Hardware array)
* blockdev --setra (for each Linux Software array on top of Hardware)
* tcp items in /proc/sys/net/core and /proc/sys/net/ipv4 like rmem_max, 
wmem_max and the like

Regards, 
Andy Liebman

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Neil Brown]

On Wednesday November 3, AndyLiebman@aol.com wrote:
> > neilb@cse.unsw.edu.au writes:
> >
> > Hmmm... I might work with 1.6 if you --build that raid0 (rather than
> > --create),  or if you are using a 64bit processor like alpha or ia64. 
> > But mdadm 1.6 definitely have problems in i386 processors when you try
> > to --create a raid0 with devices larger than 2 terabytes.
> 
> I have been using "mdadm -Cv /dev/md0 -n2 -l0 -c[256 or 512] /dev/sd{a,b}" 
> without incident. But I'll switch over to 1.8. 

Very odd.  Is this on a 2.4 or a 2.6 linux kernel??

> 
> Was the previous problem only one of creating the arrays? Or could there be 
> further problems down the line with my arrays that were created with
> 1.6? 

If your arrays were successfully created and appear to be working,
they should be fine and should continue to work.

NeilBrown


> 
> By the way, do you know anybody with enough understanding of the Linux kernel 
> to be able to rationally and logically tune the following items, which all 
> relate to one another and MUST have some optimum settings for low-latency/high 
> throughput over a network. 
> 
> * chunk size/stripe size of Hardware and Software arrays
> * blockdev --setra (for each Hardware array)
> * blockdev --setra (for each Linux Software array on top of Hardware)
> * tcp items in /proc/sys/net/core and /proc/sys/net/ipv4 like rmem_max, 
> wmem_max and the like
> 
> Regards, 
> Andy Liebman

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[AndyLiebman]

>neilb@cse.unsw.edu.au writes:
>Release 1.8.0 adds:
>   -   --pid-file option to declare a file to record the pid in for 
>   --monitor --daemonise
>    -   Support for new "faulty" personalitiy (see md.4) - not submited to
>        kernel.org yet.
>    -   support for raid0 and linear over devices larger than 2 Terabytes.
>    -   assorted bug fixes.

Hello Neil, 

Is this to say that earlier versions of mdadm did NOT support RAID-0 on 
devices larger than 2 TB? 

I have been using mdadm 1.6 to create a 4.8 TB RAID-0 array (md0) on top of 
two 3ware 2.4 TB Hardware RAID-5 arrays (/dev/sda and /dev/sdb) -- and so far I 
don't see any problems. The RAID-0 is created without trouble. The resulting 
size is reported as 4.8 TB. The performance is awesome. And, presumably, the 
data protection is sound.

Am I missing something? 

Andy Liebman

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Neil Brown]

On Tuesday November 2, AndyLiebman@aol.com wrote:
> >neilb@cse.unsw.edu.au writes:
> >Release 1.8.0 adds:
> >   -   --pid-file option to declare a file to record the pid in for 
> >   --monitor --daemonise
> >    -   Support for new "faulty" personalitiy (see md.4) - not submited to
> >        kernel.org yet.
> >    -   support for raid0 and linear over devices larger than 2 Terabytes.
> >    -   assorted bug fixes.
> 
> Hello Neil, 
> 
> Is this to say that earlier versions of mdadm did NOT support RAID-0 on 
> devices larger than 2 TB? 
> 
> I have been using mdadm 1.6 to create a 4.8 TB RAID-0 array (md0) on top of 
> two 3ware 2.4 TB Hardware RAID-5 arrays (/dev/sda and /dev/sdb) -- and so far I 
> don't see any problems. The RAID-0 is created without trouble. The resulting 
> size is reported as 4.8 TB. The performance is awesome. And, presumably, the 
> data protection is sound.
> 
> Am I missing something? 

Hmmm... I might work with 1.6 if you --build that raid0 (rather than
--create),  or if you are using a 64bit processor like alpha or ia64. 
But mdadm 1.6 definitely have problems in i386 processors when you try
to --create a raid0 with devices larger than 2 terabytes.

NeilBrown

ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Neil Brown]

I am pleased to announce the availability of 
   mdadm version 1.8.0
It is available at
   http://www.cse.unsw.edu.au/~neilb/source/mdadm/
and
   http://www.{countrycode}.kernel.org/pub/linux/utils/raid/mdadm/

as a source tar-ball and (at the first site) as an SRPM, and as an RPM for i386.

mdadm is a tool for creating, managing and monitoring
device arrays using the "md" driver in Linux, also
known as Software RAID arrays.

Release 1.8.0 adds:
    -   --pid-file option to declare a file to record the pid in for 
	--monitor --daemonise
    -   Support for new "faulty" personalitiy (see md.4) - not submited to
        kernel.org yet.
    -   support for raid0 and linear over devices larger than 2 Terabytes.
    -   assorted bug fixes.

It is hoped that the next full release of mdadm will be 2.0.0
and it will have substantially re-written handling for superblocks and
array creation.  In particular, it will be able to work with the new
superblock format (version 1) supported by 2.6.
1.8.1 may be released soon which contains much of this functionality.

Development of mdadm is sponsored by CSE@UNSW: 
  The School of Computer Science and Engineering
at
  The University of New South Wales

NeilBrown  01 November 2004

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Stephan van Hienen]

On Tue, 2 Nov 2004, Neil Brown wrote:

> as a source tar-ball and (at the first site) as an SRPM,
> and as an RPM for i386.

Neil,

today i tried to rebuild the src.rpm from 1.8 (which is the latest stable 
release?)

but i got this error :

]# rpmbuild --rebuild /tmp/mdadm-1.8.1-1.src.rpm
Installing /tmp/mdadm-1.8.1-1.src.rpm
warning: user neilb does not exist - using root
warning: group neilb does not exist - using root
warning: user neilb does not exist - using root
warning: group neilb does not exist - using root
error: line 54: second %install

i fixed the mdadm.spec file, and it compiled ok
(removed the % on line 54)

maybe fix this in later releases ?

(i was compiling this on tao (based on rhes 3))

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Neil Brown]

On Sunday November 28, raid@a2000.nu wrote:
> On Tue, 2 Nov 2004, Neil Brown wrote:
> 
> > as a source tar-ball and (at the first site) as an SRPM,
> > and as an RPM for i386.
> 
> Neil,
> 
> today i tried to rebuild the src.rpm from 1.8 (which is the latest stable 
> release?)

Yes, 1.8.0 is the one to use.
> 
> but i got this error :
> 
> ]# rpmbuild --rebuild /tmp/mdadm-1.8.1-1.src.rpm
> Installing /tmp/mdadm-1.8.1-1.src.rpm
> warning: user neilb does not exist - using root
> warning: group neilb does not exist - using root
> warning: user neilb does not exist - using root
> warning: group neilb does not exist - using root
> error: line 54: second %install
> 
> i fixed the mdadm.spec file, and it compiled ok
> (removed the % on line 54)
I assume that is
   - change %install to use "make install"
changed to
   - change install to use "make install"

It's just that emacs counts it as line 53

> 
> maybe fix this in later releases ?
> 
> (i was compiling this on tao (based on rhes 3))

Thanks.
NeilBrown

Re: ANNOUNCE: mdadm 1.8.0 - A tool for managing Soft RAID under Linux

[Stephan van Hienen]

On Tue, 30 Nov 2004, Neil Brown wrote:

> I assume that is
>   - change %install to use "make install"
> changed to
>   - change install to use "make install"
>
> It's just that emacs counts it as line 53

correct line 53