Re: dm-cache

[Ming Zhao]

The 2.6.29 version on github may work.

Can you describe the problem you see? I am cc'ing to the whole group
in case Stephen or Douglas has an answer.

- Ming

On Sun, Sep 4, 2011 at 1:56 PM, Dulcardo Arteaga Clavijo
<dulc...@gmail.com> wrote:
> Hello,
> I get the module compiled for kernel version 2.6.39.1, but dom0 is using
> 2.6.32.x and this version can not be patched with dm-cache. I have been
> trying to patch a kernel 2.6.39 with dom0 patches to get dm-cache working. I
> having some problems I hope I can get this working today. I will let you
> know
> -Dulcardo
>
> On Sun, Sep 4, 2011 at 12:46 AM, Ming Zhao <mi...@cs.fiu.edu> wrote:
>>
>> Dulcardo,
>>
>> How's the progress with setting up dm-cache? Below is the problem
>> description from XLS Hosting. Read it carefully and ask me if any
>> questions. Come up with a plan (with Daniel) and discuss with me next
>> week.
>>
>> "The nature of our workload (with a small number of players, but not a
>> predictable set, brining most of the pain) would make it the sanest
>> approach to pool virtual servers into a common cache. With dm-cache
>> the way it is set up right now,-r every individual Logical Volume on
>> the SAN at the moment a guest is started (and tear it down when a
>> guest disappears). This will waste precious cache space on unworthy
>> players, but guarantee that guests will not deal with possibly stale
>> data if they are moved around different hardware nodes.
>>
>>       2. Create a dm-cache per LVM Volume Group. This would allow
>> most of the cache to be used by whichever guest produces the biggest
>> i/o load. We would have to take special care to flush the cache
>> whenever we see a guest machine attached to it move around the
>> cluster.
>>
>> I think what we're trying to do in approach 2 (but not necessarily
>> executed that way) is the way forward, but invalidating the entire
>> cache whenever one virtual machine moves around is sub-optimal. We'd
>> be looking for a way to work around that, for example by allowing us
>> to invalidate the cache for a specific range of blocks.
>>
>> Another (perhaps even cooler) approach would be to let dm-cache could
>> use one _shared_ cache-device for multiple cache maps. Then we could
>> execute approach 1 but get the performance of approach 2."
>>
>>
>> - Ming
>
>

[Dulcardo Arteaga Clavijo]

Hello All,

I have been trying to get dm-cache working on Dom0, this is where I am now.

kernel: 2.6.18.8

dm-cache: patch for kernel 2.6.19

Description: The patch is applied without problems, the kernel has the xen patches for dom0 and it works successfully. I can load the dm_cache module after booting in dom0.

kernel: 2.6.29

dm-cache: patch for kernel 2.6.29

Descriptiong: the dm-cache is applied successfully, dm_cache.ko is created, but after booting it fails booting for dom0 and also for domU, actually this version is not a supported kernel vesion for xen, there exist different version of this kernel for fedora and others distribution, I tried with one of these

kernel: 2.6.32

dm-cache: latest dm-cache patch

Description: this kernel version is supported by XEN and it is located in XEN repositories it works fine as a dom0 kernel, but the dm-cache can not be applied the following error appear during compilation:

drivers/md/dm-cache.c: In function ‘cache_ctr’:

drivers/md/dm-cache.c:1481: warning: passing argument 4 of ‘dm_get_device’ makes integer from pointer without a cast

drivers/md/dm-cache.c:1481: error: too few arguments to function ‘dm_get_device’

drivers/md/dm-cache.c:1488: warning: passing argument 4 of ‘dm_get_device’ makes integer from pointer without a cast

drivers/md/dm-cache.c:1488: error: too few arguments to function ‘dm_get_device’

kernel: 2.6.39.4

dm-cache: lastes dm-cache patch

description: the dm-cache patch is applied successfully,  dm_cache.ko is generated without problems. it boots good for domU, but when I tried to booted from dom0, a black screen appear at the very beginning, like a multiboot failure, and nothing happen after that. I need to check the erros here.

But the kernel 2.6.39 is a unstable version from xen, according to xen mailing list, it is missing the backend drivers. still in progress.

Please let me know your comments.

-Dulcardo

[Dulcardo Arteaga]

Hi,

This error appear when you are using different kernel version.

What is your kernel version? Linux flavor?.

-Dulcardo.

On Thu, Apr 11, 2013 at 2:12 PM, <rodri.b...@gmail.com> wrote:

I am having this problem... Any solutions?

Thanks.

--
You received this message because you are subscribed to the Google Groups "dm-cache" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dm-cache+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
-Dulcardo