I'm not sure that the comparison to Kubernetes is very apt - I don't think
anyone is under the illusion that Mesos is a contender for it, that ship
has long sailed.

Also, features availability is only one of the many aspects which should be
considered when choosing among potential candidate technologies:
reliability, scalability, operational complexity, extensibility and fitness
to the actual use case are all aspects which drove us to choose Mesos over
the many alternative back when we made the decision - which we haven't
regretted since.

To give a very bad analogy, Firefox is a much better and feature complete
browser than Curl, however it doesn't mean that the latter doesn't have its
use cases, even in 2021.

As far as I'm concerned, I'd be quite happy with Mesos continuing as a
"distributed systems kernel" as described on http://mesos.apache.org/,
allowing interested individuals and organisations to build on top of it -
we're certainly happy with this paradigm at my company.

Now obviously it doesn't mean that we shouldn't consider adding new
features - for example something which was requested by some users a few
months ago and which we ended up implementing ourselves at the framework,
agent resource and custom executor level was support for NUMA topology
awareness, which could probably make sense to add to Mesos.
For more inspiration for new features, I think Nomad and Slurm might be
worth looking into as inspiration.

Also, simply continuing to address existing bug reports and MRs would I
think be a good starting point to try to revive the project, get potential
new contributors familiar with the code base, etc.
And merging some changes maintained in third party forks such as the ones
mentioned by the people from Criteo who also mentioned an interest in
keeping Mesos going forward.

People like Javi Roman seem to have some idea on potential new features -
those would be great to hear as well!

However the main barrier to any progress I can see right now - and which
was discussed in the previous thread - is that none of the current
maintainers seem to have any time to dedicate to the project, including
reviewing existing patches/MR. I still haven't seen any suggestion from the
current project members on ways to address that.
I know some people objected that electing new committers might not be
enough to revive the project - I most certainly agree it might very well
not be sufficient, however I think it is a necessary condition, unless some
of the current maintainers are willing to dedicate some time to onboarding
potential new contributors.



On Sun, 28 Feb 2021, 16:42 Jorge Machado, <jom...@me.com.invalid> wrote:

> Hi Samuel,
> To be honest, I would not invest any more Time on Mesos. The features from
> Kubernetes are just way better. :)
> > On 28. Feb 2021, at 12:54, Samuel Marks <sam...@offscale.io> wrote:
> >
> > Decouple Apache ZooKeeper, enabling Apache Mesos to run completely
> without
> > ZooKeeper. Specifically enable a choice between ZooKeeper, etcd, and
> consul.
> >
> > My organisation is somewhat interested in contributing this. We tried in
> > the past but came across some hurdles on the Mesos organisation end. Open
> > to trying again, but will need a clear pathway to getting this accepted.
> >
> > Samuel Marks
> > Charity <https://sydneyscientific.org> | consultancy <
> https://offscale.io>
> > | open-source <https://github.com/offscale> | LinkedIn
> > <https://linkedin.com/in/samuelmarks>
> >
> >
> > On Sun, Feb 28, 2021 at 7:39 PM Qian Zhang <zhq527...@gmail.com> wrote:
> >
> >> Hi Folks,
> >>
> >> To reboot this awesome project, I'd like to collect feature requests for
> >> Mesos. Please let us know your requirements for Mesos and whether you or
> >> your organization would like to contribute to the implementation of the
> >> requirements. Thanks!
> >>
> >>
> >> Regards,
> >> Qian Zhang
> >>

Reply via email to