[ops] database archiving tool

Thomas Goirand zigo at debian.org
Wed May 22 11:24:19 UTC 2019


On 5/22/19 10:34 AM, Pierre-Samuel LE STANG wrote:
> Thomas Goirand <zigo at debian.org> wrote on mer. [2019-mai-22 09:09:25 +0200]:
>> Hi Pierre,
>>
>> That's really the kind of project that I would prefer not to have to
>> exist. By this I mean, it'd be a lot nicer if this could be taken care
>> of project by project, with something like what Nova does (ie:
>> nova-manage db archive_deleted_rows).
>>
>> In such configuration, that's typically something that could be added as
>> a cron job, automatically configured by packages.
>>
>> Now, a question for other OPS reading this thread: how long should be
>> the retention? In Debian, we use to have the unsaid policy that we don't
>> want too much retention, to project privacy. Though in operation, we may
>> need at least a few weeks of history, so we can do support. If I was to
>> configure a cron job for nova, for example, what parameter should I set
>> to --before (when #556751 is merged)? My instinct would be:
>>
>> nova-manage db archive_deleted_rows \
>> 	--before $(date -d "-1 month" +%Y-%m-%d)
>>
>> Your thoughts everyone?
> 
> Hi Thomas,
> 
> Thanks for your feedback, I really appreciate it. The tool is designed to be
> customized and to fit your needs. It means that you can have one configuration
> per project or one configuration for all projects.
> 
> So you might imagine having a configuration for glance which exclude images
> table and one configuration for nova with a higher or lower retention.
> 
> --
> PS

This looks super nice then!

Will you provide a standard configuration for every OpenStack project?
It'd be nice if your package had a conf.d folder where one could drop
the config for every project. That way, every OpenStack project package
could drop a configuration there.

Cheers,

Thomas Goirand (zigo)



More information about the openstack-discuss mailing list