Open Stack

Hello guys,

I've got a similar question about cache-manager and the presence of a 
shared filesystem for instances images.
I'm currently reading the source code in order to find out how this is 
managed but before I would be curious how you achieve this on production 
servers.

For example images not used by compute node A will probably be cleaned 
on the shared FS despite the fact that compute B use it, that's the main 
problem.

How do you handle _base guys ?

Thanks
Kind regards,

Le 01/04/2015 08:07, Gary Kotton a écrit :
> At the moment the aging is pretty rudimentary. There were patches in the
> Icehouse cycles that tried to address issues like this but they did not
> get any reviews and were eventually abandoned.
> It would be nice if there were different options for aging, maybe we
> should even try and explore something pluggable so that people could add
> drivers for various use cases.
> I would be happy to work on this in the next cycle if you would like to
> collaborate
> Thanks
> Gary
>
> On 3/31/15, 10:56 PM, "Michael Still"<mikal at stillhq.com>  wrote:
>
>> Ok, so that sounds like a totally reasonable feature request to me.
>> Can you please file a bug so its tracked and let me know the bug
>> number?
>>
>> Thanks,
>> Michael
>>
>> On Wed, Apr 1, 2015 at 1:27 AM, Chris Buccella
>> <chris.buccella at verilume.com>  wrote:
>>> The instance disk images are stored in ceph.
>>>
>>> This isn't a situation we hit often. But when several users are booting
>>> large, short-lifecyle instances (say, Windows, or Hadoop images to use
>>> with
>>> Sahara), they get the "no valid host" error. The only recourse then is
>>> for
>>> an admin to manually clean up the cache, or wait for the periodic cache
>>> cleanup.
>>>
>>>
>>> -Chris
>>>
>>> On Tue, Mar 31, 2015 at 1:01 AM, Michael Still<mikal at stillhq.com>
>>> wrote:
>>>> Your request is reasonable, but not currently handled by the code. Its
>>>> a bit concerning that your cache is so disk constrained -- is it a
>>>> separate filesystem from the instance disk images?
>>>>
>>>> Michael
>>>>
>>>> On Tue, Mar 31, 2015 at 3:26 PM, Chris Buccella
>>>> <chris.buccella at verilume.com>  wrote:
>>>>> I'm running Juno. Here's my situation:
>>>>>
>>>>> 1) There are a bunch of cached images on the compute nodes
>>>>> (/var/lib/nova/instances/_base/)
>>>>> 2) I want to keep the images in the cache as long as possible (for
>>>>> performance reasons)
>>>>> 3) I want to be able to spawn instances with an uncached image
>>>>>
>>>>> What I've observed is that if the cache is full, it won't
>>>> automatically
>>>>> be
>>>>> cleaned up when a new image tries to be cached. Instead, I get a "no
>>>>> valid
>>>>> host" error when trying to boot a new instance. I've found the
>>>> following
>>>>> relevant config parameters:
>>>>>
>>>>> image_cache_manager_interval
>>>>> remove_unused_base_images
>>>>> remove_unused_original_minimum_age_seconds
>>>>>
>>>>> These work, but don't really help me. I don't want time-based cache
>>>>> management, but on-demand cache management. Basically, if
>>>>> removed_unused_base_images is true, the oldest image should be
>>>> removed
>>>>> from
>>>>> the cache to make room for the new one.
>>>>>
>>>>> Has anyone else encountered this situation? Maybe there's something
>>>> I'm
>>>>> missing?
>>>>>
>>>>> Thanks,
>>>>>
>>>>>
>>>>> -Chris
>>>>>
>>>>> _______________________________________________
>>>>> Mailing list:
>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>>>>> Post to     :openstack at lists.openstack.org
>>>>> Unsubscribe :
>>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>>>>>
>>>>
>>>> --
>>>> Rackspace Australia
>>
>> -- 
>> Rackspace Australia
>>
>> _______________________________________________
>> Mailing list:
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>> Post to     :openstack at lists.openstack.org
>> Unsubscribe :
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
> _______________________________________________
> Mailing list:http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
> Post to     :openstack at lists.openstack.org
> Unsubscribe :http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>