Re: [DISCUSS] Classloader change proposals

[Dave Marion <dm...@gmail.com>]

I have tested the new classloader as the `java.system.class.loader`, and it
works allowing me to load classes from HDFS. The reloading feature works
too, except that the ClassLoader.loadClass() logic short circuits the
loading of the new classes with the same name by calling findLoadedClass
(which returns a cached Class object). I don't believe that we will be able
to retain the reloading feature that we currently have with
`general.dynamic.classpaths` (lib/ext) in this new classloader
implementation. I have not tested it yet, but I believe that the table
context class loader reloading mechanism will work.

I know that the table context classloader feature is being used. Does
anyone know of an instance where general.dynamic.classpath is still in use
and losing the reloading feature would be unacceptable?

On Thu, Sep 24, 2020 at 10:43 AM Christopher <ct...@apache.org> wrote:

> The INFRA ticket is now closed, and the default branch has been updated.
> I also prepopulated the repo with some issue templates and GitHub
> Actions build configurations, as well as LICENSE/NOTICE and a pom.xml
> file to start it off.
>
> On Wed, Sep 23, 2020 at 10:31 AM Christopher <ct...@apache.org> wrote:
> >
> > I created the repo, but waiting on
> > https://issues.apache.org/jira/browse/INFRA-20884 to fix the GitBox
> > default branch so that the GitHub issues can be enabled.
> >
> > It should be okay to add the initial code to it, though.
> >
> > On Tue, Sep 22, 2020 at 1:40 PM Dave Marion <dm...@gmail.com> wrote:
> > >
> > > Sounds good.
> > >
> > > On Tue, Sep 22, 2020, 1:37 PM Christopher <ct...@apache.org> wrote:
> > >
> > > > Based on the conversation and direction of this, I think it probably
> > > > makes the most sense to just create a new repository under Accumulo
> > > > for it to live.
> > > >
> > > > How about 'accumulo-classloaders' for the repo name? If this is okay,
> > > > I can create it later today or tomorrow.
> > > > This is just a repo name. This is just a place to collaborate on
> > > > classloader code without constraining the scope of the repo's
> contents
> > > > too much. The actual package names and artifactId can be different,
> if
> > > > we want.
> > > >
> > > > On Tue, Sep 22, 2020 at 9:44 AM Dave Marion <dm...@gmail.com>
> wrote:
> > > > >
> > > > > [1] contains the initial set of changes to the Accumulo code base
> that
> > > > > defines the new context class loader configuration and deprecates
> the
> > > > > existing VFS ClassLoader objects. [2] contains the new
> > > > > ReloadingVFSClassLoader that can be used as the system classloader
> and a
> > > > > ClassLoaderFactory implementation for configuring contexts for
> tables and
> > > > > scans. Both build successfully and I plan on doing some local
> testing
> > > > next.
> > > > > Feedback on the design and the code is welcome.
> > > > >
> > > > > [1] https://github.com/dlmarion/accumulo/pull/2
> > > > > [2] https://github.com/dlmarion/vfs-reloading-classloader
> > > > >
> > > > > On Fri, Sep 18, 2020 at 12:19 PM Dave Marion <dm...@gmail.com>
> > > > wrote:
> > > > >
> > > > > >
> > > > > >  I tend to agree with Marc on this. If we need to push out a fix
> for
> > > > the
> > > > > > new classloader, then we can do it as needed and not have to
> rely on
> > > > > > another group of people to come to consensus on a release. Of
> course,
> > > > we
> > > > > > could maintain a fork of it in that case, but then what's the
> point? It
> > > > > > appears that VFS does have some recent activity [1], but mostly
> by one
> > > > > > person. I'm thinking that we should create a subproject for it
> and
> > > > notify
> > > > > > the commons-vfs project of its existence. If they want to copy
> it and
> > > > > > include it in their project, then they can do that.
> > > > > >
> > > > > > [1]
> https://gitbox.apache.org/repos/asf?p=commons-vfs.git;a=shortlog
> > > > > >
> > > > > > On Wed, Sep 16, 2020 at 4:48 PM Christopher <ctubbsii@apache.org
> >
> > > > wrote:
> > > > > >
> > > > > >> Only Marc joined, and we didn't talk about anything that isn't
> > > > already in
> > > > > >> this proposal, except he did mention how difficult it might be
> to try
> > > > to
> > > > > >> maintain the class loader in commons-vfs2, rather than as our
> own
> > > > small
> > > > > >> subproject, which is relatively easy.
> > > > > >>
> > > > > >> On Wed, Sep 16, 2020, 16:11 Dave Marion <dm...@gmail.com>
> wrote:
> > > > > >>
> > > > > >> > Did anyone join the call? Any notes?
> > > > > >> >
> > > > > >> > On Wed, Sep 16, 2020 at 12:44 PM Christopher <
> ctubbsii@apache.org>
> > > > > >> wrote:
> > > > > >> >
> > > > > >> > > I just want to remind everybody that I'm available in Slack
> now to
> > > > > >> > > discuss this in the ongoing video call I created in the
> #accumulo
> > > > > >> > > room, if you want to join.
> > > > > >> > >
> > > > > >> > > On Mon, Sep 14, 2020 at 10:41 PM Christopher <
> ctubbsii@apache.org
> > > > >
> > > > > >> > wrote:
> > > > > >> > > >
> > > > > >> > > > Also, if anybody is interested in a live video
> conversation to
> > > > > >> discuss
> > > > > >> > > > this interactively, I intend to be on Slack on Wednesday
> > > > afternoon
> > > > > >> > > > (EDT) starting around noon.
> > > > > >> > > >
> > > > > >> > > > On Mon, Sep 14, 2020 at 5:30 PM Christopher <
> > > > ctubbsii@apache.org>
> > > > > >> > wrote:
> > > > > >> > > > >
> > > > > >> > > > > Hi Accumulo Devs,
> > > > > >> > > > >
> > > > > >> > > > > Lately, Dave Marion (Apache ID: dlmarion) has been
> working on
> > > > > >> > > > > prototyping some new class loader concepts for Accumulo
> that
> > > > he
> > > > > >> and I
> > > > > >> > > > > have discussed, and I wanted to pitch the idea here for
> > > > > >> consideration
> > > > > >> > > > > for the project.
> > > > > >> > > > >
> > > > > >> > > > > # Background:
> > > > > >> > > > >
> > > > > >> > > > > Accumulo currently has two classloaders that are
> instantiated
> > > > at
> > > > > >> > > > > startup, and which can be used to bootstrap Accumulo
> > > > dependencies
> > > > > >> (at
> > > > > >> > > > > least, those not needed for the classloader code
> itself). This
> > > > > >> allows
> > > > > >> > > > > us to use the `general.classpaths`[1] and
> > > > > >> > > > > `general.dynamic.classpaths`[2] properties, as well as
> the
> > > > > >> > per-context
> > > > > >> > > > > classloaders (`general.vfs.*`[3] and
> > > > `table.classpath.context`[4])
> > > > > >> > for
> > > > > >> > > > > things like iterator class isolation. Since 2.0.0, we
> have
> > > > > >> deprecated
> > > > > >> > > > > `general.classpaths` and `general.dynamic.classpaths`,
> the
> > > > former
> > > > > >> > > > > supplanted by the better use of the `CLASSPATH`
> environment
> > > > > >> variable
> > > > > >> > > > > (along with much improved scripts in 2.0.0), and the
> latter
> > > > being
> > > > > >> > > > > replaceable by a user-provided class loader using the
> built-in
> > > > > >> Java
> > > > > >> > > > > property, `java.system.class.loader`[5], at their
> discretion.
> > > > > >> > > > >
> > > > > >> > > > > # The Problem:
> > > > > >> > > > >
> > > > > >> > > > > The main problem with the current code is: complexity.
> > > > Accumulo is
> > > > > >> > > > > already complex enough without needing to be in the
> business
> > > > of
> > > > > >> > > > > developing and supporting complex custom class loading
> > > > features,
> > > > > >> > > > > especially when users have viable alternatives that can
> be
> > > > better
> > > > > >> > > > > supported by independent, dedicated projects.
> Furthermore,
> > > > these
> > > > > >> > > > > custom class loaders also have a dependency on
> commons-vfs2,
> > > > which
> > > > > >> > has
> > > > > >> > > > > been the source of numerous problems and bugs that we
> have
> > > > needed
> > > > > >> to
> > > > > >> > > > > deal with, and which affect Accumulo, even though they
> are not
> > > > > >> > > > > necessarily bugs in Accumulo itself. This also brings
> in a
> > > > lot of
> > > > > >> > > > > optional dependencies that aren't needed by users who
> don't
> > > > rely
> > > > > >> on
> > > > > >> > > > > these features.
> > > > > >> > > > >
> > > > > >> > > > > # The Requirements:
> > > > > >> > > > >
> > > > > >> > > > > In spite of these problems, I believe we still want to
> enable
> > > > the
> > > > > >> use
> > > > > >> > > > > cases that our classloaders are currently enabling.
> > > > > >> > > > >
> > > > > >> > > > > Specifically,
> > > > > >> > > > > 1) the ability to have separate contexts for iterator
> class
> > > > > >> isolation
> > > > > >> > > > > (A/B testing of iterators, updating iterators in a live
> > > > system,
> > > > > >> > etc.),
> > > > > >> > > > > and
> > > > > >> > > > > 2) the ability for users to bootstrap their class path
> from
> > > > some
> > > > > >> > other
> > > > > >> > > > > distributed storage than local disk.
> > > > > >> > > > >
> > > > > >> > > > > # The Proposal:
> > > > > >> > > > >
> > > > > >> > > > > 1. Create a new reloading vfs class loader, with similar
> > > > > >> > functionality
> > > > > >> > > > > as our current two-classloaders that do the reloading
> and
> > > > provide
> > > > > >> vfs
> > > > > >> > > > > features, that can be easily used as a system class
> loader,
> > > > if the
> > > > > >> > > > > user chooses to, and deprecate (for removal in 3.0) the
> > > > built-in
> > > > > >> > > > > implementations. This class loader could not only be
> used with
> > > > > >> > > > > Accumulo, but it could also be used by any other
> project that
> > > > > >> chooses
> > > > > >> > > > > to use it, because it will not have much, if any,
> dependencies
> > > > > >> beyond
> > > > > >> > > > > commons-vfs2, and will certainly not depend on Accumulo.
> > > > Creating
> > > > > >> > this
> > > > > >> > > > > separate class loader provides us a path forward to
> simplify
> > > > > >> Accumulo
> > > > > >> > > > > by removing these features from Accumulo directly (the
> > > > properties
> > > > > >> are
> > > > > >> > > > > already deprecated), and enabling it to be maintained
> > > > > >> independently.
> > > > > >> > > > > 2. Create a new class loader factory property in
> Accumulo,
> > > > with
> > > > > >> > > > > corresponding SPI interface, for users to provide their
> own
> > > > > >> > > > > implementation of a class loader factory, that can map a
> > > > per-table
> > > > > >> > > > > "context" to a ClassLoader of the implementation's
> choosing.
> > > > > >> > > > >
> > > > > >> > > > > The result of doing these two things will allow us to
> more
> > > > > >> flexibly
> > > > > >> > > > > support user class loading needs, without being directly
> > > > > >> responsible
> > > > > >> > > > > for class loading implementations inside Accumulo's core
> > > > code. All
> > > > > >> > the
> > > > > >> > > > > same functionality that is available today will
> continue to
> > > > exist,
> > > > > >> > but
> > > > > >> > > > > will be configured differently. The resulting code in
> Accumulo
> > > > > >> will
> > > > > >> > be
> > > > > >> > > > > dramatically simpler, as we would no longer have any
> complex
> > > > class
> > > > > >> > > > > loading implementations in our code base, and we would
> no
> > > > longer
> > > > > >> have
> > > > > >> > > > > any direct dependency on commons-vfs2, which has been
> > > > problematic.
> > > > > >> > > > > Independent implementations may use commons-vfs2, or
> something
> > > > > >> else,
> > > > > >> > > > > but will be more easily testable and maintainable as
> > > > independent
> > > > > >> > > > > projects that are pluggable in Accumulo.
> > > > > >> > > > >
> > > > > >> > > > > Dave has already been working on prototyping these
> proposed
> > > > > >> changes,
> > > > > >> > > > > and it is looking very feasible.
> > > > > >> > > > >
> > > > > >> > > > > We are now ready to:
> > > > > >> > > > > 1. get feedback on the overall proposal, and
> > > > > >> > > > > 2. decide on where to maintain the separate class
> loader.
> > > > > >> > > > >
> > > > > >> > > > > For where to maintain, the options seem to be: A) try to
> > > > donate to
> > > > > >> > > > > commons-vfs2 OR B) maintain as a new repository,
> > > > > >> > > > > accumulo-vfs-classloader.
> > > > > >> > > > >
> > > > > >> > > > > Note that we have not yet proposed the idea of a
> user-facing,
> > > > > >> > > > > configurable, reloading vfs classloader to the
> commons-vfs2
> > > > > >> > > > > developers. We wanted to get our own community's
> feedback on
> > > > this
> > > > > >> > > > > first.
> > > > > >> > > > >
> > > > > >> > > > > Please discuss.
> > > > > >> > > > >
> > > > > >> > > > > Thanks,
> > > > > >> > > > > Christopher (in collaboration with Dave)
> > > > > >> > > > >
> > > > > >> > > > > [1]:
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > >
> https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_classpaths
> > > > > >> > > > > [2]:
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > >
> https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_dynamic_classpaths
> > > > > >> > > > > [3]:
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > >
> https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_vfs_context_classpath_prefix
> > > > > >> > > > > [4]:
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > >
> https://accumulo.apache.org/docs/2.x/configuration/server-properties#table_classpath_context
> > > > > >> > > > > [5]:
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > >
> https://docs.oracle.com/javase/8/docs/api/java/lang/ClassLoader.html#getSystemClassLoader%E2%80%93
> > > > > >> > >
> > > > > >> >
> > > > > >>
> > > > > >
> > > >
>

Re: [DISCUSS] Classloader change proposals

[Christopher <ct...@apache.org>]

I cannot think of a good use case where somebody would want dynamic
class reloading from jars in the startup context. That seems risky and
prone to bugs. Other than a few user-pluggable components, such as
balancer, volume chooser, etc., most of the startup context contains
stuff that isn't safe to reload (ZK jars, Hadoop jars, and all their
dependencies, for example). The biggest use case for reloading seems
to be with the per-table contexts, and that doesn't require the system
class loader to support reloading.

We *could* make the startup context pluggable with its own
classloader, rather than rely on the built-in Java system class loader
mechanism, and that startup context classloader could be one that
supports reloading. However, this would also have its own problems,
like where would it get its config from when we haven't yet
bootstrapped enough of the application to set all the Accumulo code
for config handling, logging, etc. yet.

Given the available options, I'm inclined to just drop the reloading
behavior from the separate class loader intended for use as the system
class loader (but still leave reloading as a feature of the per-table
context classloading). If somebody really needs the reloading behavior
in the startup context, they can implement it later, as either a PR,
or as a completely separate launcher/wrapper or system class loader,
or people can run Accumulo in an application server. All of these
options seem far outside the scope of "Accumulo" work, though. I think
it'd be enough if we just made the basic non-reloading system class
loader available as an option (along with the reloading per-table
class loading).

On Mon, Oct 5, 2020 at 2:44 PM Dave Marion <dm...@gmail.com> wrote:
>
> I have tested the new classloader as the `java.system.class.loader`, and it
> works allowing me to load classes from HDFS. The reloading feature works
> too, except that the ClassLoader.loadClass() logic short circuits the
> loading of the new classes with the same name by calling findLoadedClass
> (which returns a cached Class object). I don't believe that we will be able
> to retain the reloading feature that we currently have with
> `general.dynamic.classpaths` (lib/ext) in this new classloader
> implementation. I have not tested it yet, but I believe that the table
> context class loader reloading mechanism will work.
>
> I know that the table context classloader feature is being used. Does
> anyone know of an instance where general.dynamic.classpath is still in use
> and losing the reloading feature would be unacceptable?
>
> On Thu, Sep 24, 2020 at 10:43 AM Christopher <ct...@apache.org> wrote:
>
> > The INFRA ticket is now closed, and the default branch has been updated.
> > I also prepopulated the repo with some issue templates and GitHub
> > Actions build configurations, as well as LICENSE/NOTICE and a pom.xml
> > file to start it off.
> >
> > On Wed, Sep 23, 2020 at 10:31 AM Christopher <ct...@apache.org> wrote:
> > >
> > > I created the repo, but waiting on
> > > https://issues.apache.org/jira/browse/INFRA-20884 to fix the GitBox
> > > default branch so that the GitHub issues can be enabled.
> > >
> > > It should be okay to add the initial code to it, though.
> > >
> > > On Tue, Sep 22, 2020 at 1:40 PM Dave Marion <dm...@gmail.com> wrote:
> > > >
> > > > Sounds good.
> > > >
> > > > On Tue, Sep 22, 2020, 1:37 PM Christopher <ct...@apache.org> wrote:
> > > >
> > > > > Based on the conversation and direction of this, I think it probably
> > > > > makes the most sense to just create a new repository under Accumulo
> > > > > for it to live.
> > > > >
> > > > > How about 'accumulo-classloaders' for the repo name? If this is okay,
> > > > > I can create it later today or tomorrow.
> > > > > This is just a repo name. This is just a place to collaborate on
> > > > > classloader code without constraining the scope of the repo's
> > contents
> > > > > too much. The actual package names and artifactId can be different,
> > if
> > > > > we want.
> > > > >
> > > > > On Tue, Sep 22, 2020 at 9:44 AM Dave Marion <dm...@gmail.com>
> > wrote:
> > > > > >
> > > > > > [1] contains the initial set of changes to the Accumulo code base
> > that
> > > > > > defines the new context class loader configuration and deprecates
> > the
> > > > > > existing VFS ClassLoader objects. [2] contains the new
> > > > > > ReloadingVFSClassLoader that can be used as the system classloader
> > and a
> > > > > > ClassLoaderFactory implementation for configuring contexts for
> > tables and
> > > > > > scans. Both build successfully and I plan on doing some local
> > testing
> > > > > next.
> > > > > > Feedback on the design and the code is welcome.
> > > > > >
> > > > > > [1] https://github.com/dlmarion/accumulo/pull/2
> > > > > > [2] https://github.com/dlmarion/vfs-reloading-classloader
> > > > > >
> > > > > > On Fri, Sep 18, 2020 at 12:19 PM Dave Marion <dm...@gmail.com>
> > > > > wrote:
> > > > > >
> > > > > > >
> > > > > > >  I tend to agree with Marc on this. If we need to push out a fix
> > for
> > > > > the
> > > > > > > new classloader, then we can do it as needed and not have to
> > rely on
> > > > > > > another group of people to come to consensus on a release. Of
> > course,
> > > > > we
> > > > > > > could maintain a fork of it in that case, but then what's the
> > point? It
> > > > > > > appears that VFS does have some recent activity [1], but mostly
> > by one
> > > > > > > person. I'm thinking that we should create a subproject for it
> > and
> > > > > notify
> > > > > > > the commons-vfs project of its existence. If they want to copy
> > it and
> > > > > > > include it in their project, then they can do that.
> > > > > > >
> > > > > > > [1]
> > https://gitbox.apache.org/repos/asf?p=commons-vfs.git;a=shortlog
> > > > > > >
> > > > > > > On Wed, Sep 16, 2020 at 4:48 PM Christopher <ctubbsii@apache.org
> > >
> > > > > wrote:
> > > > > > >
> > > > > > >> Only Marc joined, and we didn't talk about anything that isn't
> > > > > already in
> > > > > > >> this proposal, except he did mention how difficult it might be
> > to try
> > > > > to
> > > > > > >> maintain the class loader in commons-vfs2, rather than as our
> > own
> > > > > small
> > > > > > >> subproject, which is relatively easy.
> > > > > > >>
> > > > > > >> On Wed, Sep 16, 2020, 16:11 Dave Marion <dm...@gmail.com>
> > wrote:
> > > > > > >>
> > > > > > >> > Did anyone join the call? Any notes?
> > > > > > >> >
> > > > > > >> > On Wed, Sep 16, 2020 at 12:44 PM Christopher <
> > ctubbsii@apache.org>
> > > > > > >> wrote:
> > > > > > >> >
> > > > > > >> > > I just want to remind everybody that I'm available in Slack
> > now to
> > > > > > >> > > discuss this in the ongoing video call I created in the
> > #accumulo
> > > > > > >> > > room, if you want to join.
> > > > > > >> > >
> > > > > > >> > > On Mon, Sep 14, 2020 at 10:41 PM Christopher <
> > ctubbsii@apache.org
> > > > > >
> > > > > > >> > wrote:
> > > > > > >> > > >
> > > > > > >> > > > Also, if anybody is interested in a live video
> > conversation to
> > > > > > >> discuss
> > > > > > >> > > > this interactively, I intend to be on Slack on Wednesday
> > > > > afternoon
> > > > > > >> > > > (EDT) starting around noon.
> > > > > > >> > > >
> > > > > > >> > > > On Mon, Sep 14, 2020 at 5:30 PM Christopher <
> > > > > ctubbsii@apache.org>
> > > > > > >> > wrote:
> > > > > > >> > > > >
> > > > > > >> > > > > Hi Accumulo Devs,
> > > > > > >> > > > >
> > > > > > >> > > > > Lately, Dave Marion (Apache ID: dlmarion) has been
> > working on
> > > > > > >> > > > > prototyping some new class loader concepts for Accumulo
> > that
> > > > > he
> > > > > > >> and I
> > > > > > >> > > > > have discussed, and I wanted to pitch the idea here for
> > > > > > >> consideration
> > > > > > >> > > > > for the project.
> > > > > > >> > > > >
> > > > > > >> > > > > # Background:
> > > > > > >> > > > >
> > > > > > >> > > > > Accumulo currently has two classloaders that are
> > instantiated
> > > > > at
> > > > > > >> > > > > startup, and which can be used to bootstrap Accumulo
> > > > > dependencies
> > > > > > >> (at
> > > > > > >> > > > > least, those not needed for the classloader code
> > itself). This
> > > > > > >> allows
> > > > > > >> > > > > us to use the `general.classpaths`[1] and
> > > > > > >> > > > > `general.dynamic.classpaths`[2] properties, as well as
> > the
> > > > > > >> > per-context
> > > > > > >> > > > > classloaders (`general.vfs.*`[3] and
> > > > > `table.classpath.context`[4])
> > > > > > >> > for
> > > > > > >> > > > > things like iterator class isolation. Since 2.0.0, we
> > have
> > > > > > >> deprecated
> > > > > > >> > > > > `general.classpaths` and `general.dynamic.classpaths`,
> > the
> > > > > former
> > > > > > >> > > > > supplanted by the better use of the `CLASSPATH`
> > environment
> > > > > > >> variable
> > > > > > >> > > > > (along with much improved scripts in 2.0.0), and the
> > latter
> > > > > being
> > > > > > >> > > > > replaceable by a user-provided class loader using the
> > built-in
> > > > > > >> Java
> > > > > > >> > > > > property, `java.system.class.loader`[5], at their
> > discretion.
> > > > > > >> > > > >
> > > > > > >> > > > > # The Problem:
> > > > > > >> > > > >
> > > > > > >> > > > > The main problem with the current code is: complexity.
> > > > > Accumulo is
> > > > > > >> > > > > already complex enough without needing to be in the
> > business
> > > > > of
> > > > > > >> > > > > developing and supporting complex custom class loading
> > > > > features,
> > > > > > >> > > > > especially when users have viable alternatives that can
> > be
> > > > > better
> > > > > > >> > > > > supported by independent, dedicated projects.
> > Furthermore,
> > > > > these
> > > > > > >> > > > > custom class loaders also have a dependency on
> > commons-vfs2,
> > > > > which
> > > > > > >> > has
> > > > > > >> > > > > been the source of numerous problems and bugs that we
> > have
> > > > > needed
> > > > > > >> to
> > > > > > >> > > > > deal with, and which affect Accumulo, even though they
> > are not
> > > > > > >> > > > > necessarily bugs in Accumulo itself. This also brings
> > in a
> > > > > lot of
> > > > > > >> > > > > optional dependencies that aren't needed by users who
> > don't
> > > > > rely
> > > > > > >> on
> > > > > > >> > > > > these features.
> > > > > > >> > > > >
> > > > > > >> > > > > # The Requirements:
> > > > > > >> > > > >
> > > > > > >> > > > > In spite of these problems, I believe we still want to
> > enable
> > > > > the
> > > > > > >> use
> > > > > > >> > > > > cases that our classloaders are currently enabling.
> > > > > > >> > > > >
> > > > > > >> > > > > Specifically,
> > > > > > >> > > > > 1) the ability to have separate contexts for iterator
> > class
> > > > > > >> isolation
> > > > > > >> > > > > (A/B testing of iterators, updating iterators in a live
> > > > > system,
> > > > > > >> > etc.),
> > > > > > >> > > > > and
> > > > > > >> > > > > 2) the ability for users to bootstrap their class path
> > from
> > > > > some
> > > > > > >> > other
> > > > > > >> > > > > distributed storage than local disk.
> > > > > > >> > > > >
> > > > > > >> > > > > # The Proposal:
> > > > > > >> > > > >
> > > > > > >> > > > > 1. Create a new reloading vfs class loader, with similar
> > > > > > >> > functionality
> > > > > > >> > > > > as our current two-classloaders that do the reloading
> > and
> > > > > provide
> > > > > > >> vfs
> > > > > > >> > > > > features, that can be easily used as a system class
> > loader,
> > > > > if the
> > > > > > >> > > > > user chooses to, and deprecate (for removal in 3.0) the
> > > > > built-in
> > > > > > >> > > > > implementations. This class loader could not only be
> > used with
> > > > > > >> > > > > Accumulo, but it could also be used by any other
> > project that
> > > > > > >> chooses
> > > > > > >> > > > > to use it, because it will not have much, if any,
> > dependencies
> > > > > > >> beyond
> > > > > > >> > > > > commons-vfs2, and will certainly not depend on Accumulo.
> > > > > Creating
> > > > > > >> > this
> > > > > > >> > > > > separate class loader provides us a path forward to
> > simplify
> > > > > > >> Accumulo
> > > > > > >> > > > > by removing these features from Accumulo directly (the
> > > > > properties
> > > > > > >> are
> > > > > > >> > > > > already deprecated), and enabling it to be maintained
> > > > > > >> independently.
> > > > > > >> > > > > 2. Create a new class loader factory property in
> > Accumulo,
> > > > > with
> > > > > > >> > > > > corresponding SPI interface, for users to provide their
> > own
> > > > > > >> > > > > implementation of a class loader factory, that can map a
> > > > > per-table
> > > > > > >> > > > > "context" to a ClassLoader of the implementation's
> > choosing.
> > > > > > >> > > > >
> > > > > > >> > > > > The result of doing these two things will allow us to
> > more
> > > > > > >> flexibly
> > > > > > >> > > > > support user class loading needs, without being directly
> > > > > > >> responsible
> > > > > > >> > > > > for class loading implementations inside Accumulo's core
> > > > > code. All
> > > > > > >> > the
> > > > > > >> > > > > same functionality that is available today will
> > continue to
> > > > > exist,
> > > > > > >> > but
> > > > > > >> > > > > will be configured differently. The resulting code in
> > Accumulo
> > > > > > >> will
> > > > > > >> > be
> > > > > > >> > > > > dramatically simpler, as we would no longer have any
> > complex
> > > > > class
> > > > > > >> > > > > loading implementations in our code base, and we would
> > no
> > > > > longer
> > > > > > >> have
> > > > > > >> > > > > any direct dependency on commons-vfs2, which has been
> > > > > problematic.
> > > > > > >> > > > > Independent implementations may use commons-vfs2, or
> > something
> > > > > > >> else,
> > > > > > >> > > > > but will be more easily testable and maintainable as
> > > > > independent
> > > > > > >> > > > > projects that are pluggable in Accumulo.
> > > > > > >> > > > >
> > > > > > >> > > > > Dave has already been working on prototyping these
> > proposed
> > > > > > >> changes,
> > > > > > >> > > > > and it is looking very feasible.
> > > > > > >> > > > >
> > > > > > >> > > > > We are now ready to:
> > > > > > >> > > > > 1. get feedback on the overall proposal, and
> > > > > > >> > > > > 2. decide on where to maintain the separate class
> > loader.
> > > > > > >> > > > >
> > > > > > >> > > > > For where to maintain, the options seem to be: A) try to
> > > > > donate to
> > > > > > >> > > > > commons-vfs2 OR B) maintain as a new repository,
> > > > > > >> > > > > accumulo-vfs-classloader.
> > > > > > >> > > > >
> > > > > > >> > > > > Note that we have not yet proposed the idea of a
> > user-facing,
> > > > > > >> > > > > configurable, reloading vfs classloader to the
> > commons-vfs2
> > > > > > >> > > > > developers. We wanted to get our own community's
> > feedback on
> > > > > this
> > > > > > >> > > > > first.
> > > > > > >> > > > >
> > > > > > >> > > > > Please discuss.
> > > > > > >> > > > >
> > > > > > >> > > > > Thanks,
> > > > > > >> > > > > Christopher (in collaboration with Dave)
> > > > > > >> > > > >
> > > > > > >> > > > > [1]:
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > >
> > https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_classpaths
> > > > > > >> > > > > [2]:
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > >
> > https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_dynamic_classpaths
> > > > > > >> > > > > [3]:
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > >
> > https://accumulo.apache.org/docs/2.x/configuration/server-properties#general_vfs_context_classpath_prefix
> > > > > > >> > > > > [4]:
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > >
> > https://accumulo.apache.org/docs/2.x/configuration/server-properties#table_classpath_context
> > > > > > >> > > > > [5]:
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > >
> > https://docs.oracle.com/javase/8/docs/api/java/lang/ClassLoader.html#getSystemClassLoader%E2%80%93
> > > > > > >> > >
> > > > > > >> >
> > > > > > >>
> > > > > > >
> > > > >
> >