I think in all cases, if you want to manage a cluster of kernels,
using the deploy folder is not necesseraly the best idea.  Or just use
rsync to copy over the changes then ...   But having a full back up
requires much more than that, it would need the preference services,
the config admin, and all other configuration files in the /etc
folder.
Btw, what's your use case for changing the startup.properties file ?

On Thu, Apr 2, 2009 at 16:45, Charles Moulliard <[email protected]> wrote:
> Using features list as the installation contract is a good idea but in this
> case, the deploy folder must be forbidden otherwise we will have a gap
> between the content of the list + what is installed using deploy folder
>
> In resume : startup.properties + features list + folder deploy = list of
> bundles to be backuped
>
>
>
> On Thu, Apr 2, 2009 at 4:27 PM, Guillaume Nodet <[email protected]> wrote:
>
>> On Thu, Apr 2, 2009 at 16:23, Jamie G. <[email protected]> wrote:
>> > 1) Backup/recover config files
>> >
>> > Interesting, would be a useful feature.
>> >
>> > 2) Log rotation (already discussed)
>> >
>> > 3) Image of bundles installed..
>> >
>> > This sounds very useful to me, kind of like taking a system snapshot
>> > that an administrator could roll back to if need be (great when dozens
>> > of bundles are involved). Would this be another web admin interface
>> > delivered feature or something we would do via the kernel shell
>> > interface? I would assume the shell would be the best place for this
>> > type of administrative function.
>>
>> I see a web console as offering the same features than the shell, so
>> imho, everything available from the web should be available from the
>> lower level shell also.
>>
>> What about building the rollback into the features service ? So that
>> if the installation fails, it would leave the runtime in the same
>> state that it was before installing the bundles.   If everything is
>> define by a feature, the state would only be the list of installed
>> feature which may be way easier to manage.
>>
>> >
>> > On Thu, Apr 2, 2009 at 11:05 AM, Charles Moulliard <[email protected]>
>> wrote:
>> >> Concerning the backup strategy, I see two points :
>> >>
>> >> 1) Backup/recover config files
>> >>
>> >> This can be achieved through web admin interface. A zip file will be
>> >> generated containing the config files and named according to the current
>> >> system date. In case of trouble (after parameter modification or
>> >> reinstallation), the user can reinstall files
>> >>
>> >> 2) Log rotation (already discussed) : can be defined in the existing cfg
>> >> file
>> >>
>> >> 3) Keep image of the bundles installed (useful to roll back install of
>> >> project = sum of bundles)
>> >> With this feature, the deployer can be roll back an installation in case
>> of
>> >> problem
>> >>
>> >> Charles
>> >>
>> >>
>> >> On Thu, Apr 2, 2009 at 3:16 PM, Jamie G. <[email protected]>
>> wrote:
>> >>
>> >>> I'd like to know more about the Backup strategy too. Are we looking
>> >>> for some best practices documentation or something more comprehensive
>> >>> (such as disaster recovery scenario)?
>> >>>
>> >>> We already have some fail over support via the kernel as Guillaume has
>> >>> noted.
>> >>>
>> >>> Cheers,
>> >>> Jamie
>> >>>
>> >>> >> - Backup strategy,
>> >>> >
>> >>> > Please explain ? Is this related to
>> >>> >
>> >>>
>> http://servicemix.apache.org/kernel/67-configuring-failover-deployments-available-in-110.html
>> >>> > ?
>> >>>
>> >>
>> >
>>
>>
>>
>> --
>> Cheers,
>> Guillaume Nodet
>> ------------------------
>> Blog: http://gnodet.blogspot.com/
>> ------------------------
>> Open Source SOA
>> http://fusesource.com
>>
>



-- 
Cheers,
Guillaume Nodet
------------------------
Blog: http://gnodet.blogspot.com/
------------------------
Open Source SOA
http://fusesource.com

Reply via email to