Open Stack

Excerpts from Octave J. Orgeron's message of 2017-02-02 12:16:15 -0700:
> Hi Doug,
> 
> Comments below..
> 
> Thanks,
> Octave
> 
> On 2/2/2017 11:27 AM, Doug Hellmann wrote:
> > Excerpts from Octave J. Orgeron's message of 2017-02-02 09:40:23 -0700:
> >> Hi Doug,
> >>
> >> One could try to detect the default engine. However, in MySQL Cluster,
> >> you can support multiple storage engines. Only NDB is fully clustered
> >> and replicated, so if you accidentally set a table to be InnoDB it won't
> >> be replicated . So it makes more sense for the operator to be explicit
> >> on which engine they want to use.
> > I think this change is probably a bigger scale item than I understood
> > it to be when you originally contacted me off-list for advice about
> > how to get started. I hope I haven't steered you too far wrong, but
> > at least the conversation is started.
> >
> > As someone (Mike?) pointed out on the review, the option by itself
> > doesn't do much of anything, now. Before we add it, I think we'll
> > want to see some more detail about how it's going used. It may be
> > easier to have that broader conversation here on email than on the
> > patch currently up for review.
> 
> Understood, it's a complicated topic since it involves gritty details in 
> SQL Alchemy and Alembic that are masked from end-users and operators 
> alike. Figuring out how to make this work did take some time on my part.
> 
> >
> > It sounds like part of the plan is to use the configuration setting
> > to control how the migration scripts create tables. How will that
> > work? Does each migration need custom logic, or can we build helpers
> > into oslo.db somehow? Or will the option be passed to the database
> > to change its behavior transparently?
> 
> These are good questions. For each service, when the db sync or db 
> manage operation is done it will call into SQL Alchemy or Alembic 
> depending on the methods used by the given service. For example, most 
> use SQL Alchemy, but there are services like Ironic and Neutron that use 
> Alembic. It is within these scripts under the <service>/db/* hierarchy 
> that the logic exist today to configure the database schema for any 
> given service. Both approaches will look at the schema version in the 
> database to determine where to start the create, upgrade, heal, etc. 
> operations. What my patches do is that in the scripts where a table 
> needs to be modified, there will be custom IF/THEN logic to check the 
> cfg.CONF.database.mysql_storage_engine setting to make the required 
> modifications. There are also use cases where the api.py or model(s).py 
> under the <service>/db/ hierarchy needs to look at this setting as well 
> for API and CLI operations where mysql_engine is auto-inserted into DB 
> operations. In those use cases, I replace the hard coded "InnoDB" with 
> the mysql_storage_engine variable.

So all existing scripts that create or modify tables will need to
be updated? That's going to be a lot of work. It will also be a lot
of work to ensure that new alter scripts are implemented using the
required logic, and that testing happens in the gates for all
projects supporting this feature to ensure there are no regressions
or behavioral changes in the applications as a result of the changes
in table definitions.

I'll let the folks more familiar with databases in general and MySQL
in particular respond to some of the technical details, but I think
I should give you fair warning that you're taking on a very big
project, especially for someone new to the community.

> It would be interesting if we could develop some helpers to automate 
> this, but it would probably have to be at the SQL Alchemy or Alembic 
> levels. Unfortunately, throughout all of the OpenStack services today we 
> are hard coding things like mysql_engine, using InnoDB specific features 
> (savepoints, nested operations, etc.), and not following the strict SQL 
> orders for modifying table elements (foreign keys, constraints, and 
> indexes). That actually makes it difficult to support other MySQL 
> dialects or other databases out of the box. SQL Alchemy can be used to 
> fix some of these things if the SQL statements are all generic and we 
> follow strict SQL rules. But to change that would be a monumental 
> effort. That is why I took this approach of just adding custom logic. 
> There is a president for this already for Postgres and DB2 support in 
> some of the OpenStack services using custom logic to deal with similar 
> differences.
> 
> As to why we should place the configuration setting into oslo.db? Here 
> are a couple of logical reasons:

Oh, I'm not questioning putting the option in oslo.db. I think that's
clearly the right place to put it.

> 
>   * The configuration block for database settings for each service comes
>     from the oslo.db namespace today under cfg.CONF.database.*. Placing
>     it here makes the location consistent across all of the services.
>   * Within the SQL Alchemy and Alembic scripts, this is one of the few
>     common namespaces that are available without bringing in a larger
>     number of modules across the services today.
>   * Many of the SQL Alchemy and Alembic scripts only import the minimal
>     set of python modules. If we imported others, we would also have to
>     initialize those name spaces which means a lot more code :(
>   * Reduces the amount of overhead required to make these changes.
> 
> >
> > Keep in mind that we do not encourage code outside of libraries to
> > rely on configuration settings defined within libraries, because
> > that limits our ability to change the names and locations of the
> > configuration variables.  If migration scripts need to access the
> > configuration setting we will need to add some sort of public API
> > to oslo.db to query the value. The function can simply return the
> > configured value.
> 
> Configuration parameters within any given service will make use of a 
> large namespace that pulls in things from oslo and the .conf files for a 
> given service. So even when an API, CLI, or DB related call is made, 
> these namespaces are key for things to work. In the case of the SQL 
> Alchemy and Alembic scripts, they also make use of this namespace with 
> oslo, oslo.db, etc. to figure out how to connect to the database and 
> other database settings. I don't think we need a public API for these 
> kinds of calls as the community already makes use of the libraries to 
> build the namespace. My oslo.db setting and patches for each service 
> just make use of the cfg.CONF.database namespace to determine the 
> correct behavior to execute.

I may not have been entirely clear. You need to add a function to
oslo.db to allow a user of oslo.db to read the configuration value
without knowing what that option name is. There are two reasons for
this policy:

1. Configuration options are supposed to be completely transparent
   to the application developer using the library, otherwise they
   would be parameters to the classes or functions in the library
   instead of deployer-facing configuration options.

   oslo.config allows us to rename configuration options transparently
   to deployers (they get a warning about the new name or location
   for the option in the config file, but the library knows both
   locations).

   The rename feature does not work when accessing options
   programmatically, because we do not consider configuration options
   to be part of the API of a library.  That means that cfg.CONF.foo.bar
   can move to cfg.CONF.blah.bletch, and your code using it by the
   old name will break.

2. Accessing configuration options depends on having them registered,
   and a user of the library that owns a configuration option may not
   know which functions in the library to call to register the options.
   As a result, they may try to use an option before it is actually
   defined. Using an access function to read the value of an option
   allows the library to ensure the option is registered before trying
   to return the value.

For those reasons, in cases where a configuration option needs to
be exposed outside of the library we require a function defined
inside the library where we can have unit tests that will break if
the configuration option is renamed or otherwise changed, and so
we can handle those changes without breaking applications consuming
the library.

In this case, the migration scripts are outside of oslo.db, so they
will need a public function added to oslo.db to access the configuration
value. The function should first ensure that the new option is
registered, and then return the configured value.

> 
> >
> > What other behaviors are likely to be changed by the new option?
> > Will application runtime behavior need to know about the storage
> > engine?
> 
> The changes will be transparent to the application runtime behavior. The 
> APIs and CLI tools call into the <service>/db/api.py as the entry point 
> for database calls. Behind this you usually have a models.py that is 
> aware of the database schema to understand the layout of things. So the 
> underlining structure is abstracted away from the run-time. These entry 
> points sometimes do require minor modifications to handle any hard coded 
> issues or intercept functions like savepoints and nested operations. 
> Again I use the cfg.CONF.database namespace to check for the appropriate 
> behavior and implement IF/THEN logic to do the right thing.
> 
> Some of my design objectives for all of these patches are:
> 
>   * Zero impact on OpenStack functionality and usability (API, CLI, user
>     experience)
>   * No loss in database structure. Consistent foreign keys, constraints,
>     indexes, etc.
>   * Minimal impact on column size and/or types to fit within NDB table
>     row limits. Many columns are over-sized today.
>   * Validate functionality of APIs, service processes, and CLI. Tempest
>     is our friend :)
>   * Zero impact for users not using MySQL Cluster (NDB).
> 
> >
> > Doug
> >
> >> Thanks,
> >> Octave
> >>
> >> On 2/2/2017 6:46 AM, Doug Hellmann wrote:
> >>> Excerpts from Octave J. Orgeron's message of 2017-02-01 20:33:38 -0700:
> >>>> Hi Folks,
> >>>>
> >>>> I'm working on adding support for MySQL Cluster to the core OpenStack
> >>>> services. This will enable the community to benefit from an
> >>>> active/active, auto-sharding, and scale-out MySQL database. My approach
> >>>> is to have a single configuration setting in each core OpenStack service
> >>>> in the oslo.db configuration section called mysql_storage_engine that
> >>>> will enable the logic in the SQL Alchemy or Alembic upgrade scripts to
> >>>> handle the differences between InnoDB and NDB storage engines
> >>>> respectively. When enabled, this logic will make the required table
> >>>> schema changes around:
> >>>>
> >>>>     * Row character length limits 65k -> 14k
> >>>>     * Proper SQL ordering of foreign key, constraints, and index operations
> >>>>     * Interception of savepoint and nested operations
> >>>>
> >>>> By default this functionality will not be enabled and will have no
> >>>> impact on the default InnoDB functionality. These changes have been
> >>>> tested on Kilo and Mitaka in previous releases of our OpenStack
> >>>> distributions with Tempest. I'm working on updating these patches for
> >>>> upstream consumption. We are also working on a 3rd party CI for
> >>>> regression testing against MySQL Cluster for the community.
> >>>>
> >>>> The first change set is for oslo.db and can be reviewed at:
> >>>>
> >>>> https://review.openstack.org/427970
> >>>>
> >>>> Thanks,
> >>>> Octave
> >>>>
> >>> Is it possible to detect the storage engine at runtime, instead of
> >>> having the operator configure it?
> >>>
> >>> Doug
> >>>
> >>> __________________________________________________________________________
> >>> OpenStack Development Mailing List (not for usage questions)
> >>> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> >>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
> > __________________________________________________________________________
> > OpenStack Development Mailing List (not for usage questions)
> > Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>