cncf
Health of OpenEBS
Chatter from TOC folks looking at Incubation prospects of OpenEBS:
- https://github.com/cncf/toc/pull/506
- https://cloud-native.slack.com/archives/C0MP69YF4/p1656112677562789
Looking at the devstats graphs:
Market news:
- https://www.datacore.com/news-post/datacore-acquires-mayadata/
Looks like when the main backers of OpenEBS project namely MayaData got acquired, the project has not been the same unfortunately. There has been a steep drop off in both technical activity and community aspects as well.
Questions:
- Are the maintainers still active openebs issue #3548? ( cc @AmitKumarDas @GlennBullingham @gila @ksatchit @kmova @mynktl @muratkars @pawanpraka1 @richardelling @vishnuitta )
- Is there still a desire / interest in companies other than Mayadata/Datacore to continue work on this project?
- Is there a community still interested? path forward for incubation and beyond?
thanks, Dims
Here at Civo we're using the mayastor driver pretty extensively. We've been heavily involved in the testing and proving the scalability of the driver and are looking forward the 2.0 release that's predicted for early September.
Talking to the team at DataCore, they are committed to the project at the moment and keeping it OpenSource and available for the community to use. The OpenSource side of OpenEBS did play a part in us choosing to use the software.
@DMajrekar can you please tag the folks in DataCore here on this issue? (so we can hear from them?) thanks!
@DMajrekar - we're listening! ... and will be replying shortly. I'd rather give you a coherent picture than a set of piecemeal responses.
Hi @dims – yes, OpenEBS is still very much alive and kicking!
DataCore is fully committed to the ongoing development of OpenEBS and is applying its know-how and expertise to enrich the project with a full range of data services as well as investing in growing the community.
Since the acquisition of MayaData, OpenEBS has reached a milestone 3.1 release, including the production ready Mayastor 1.0 storage engine which delivers NVMe grade performance from a disaggregated storage infrastructure.
We’ve also seen new community members getting involved. Civo (@DMajrekar) have been very supportive and active contributors through their production use of Mayastor and by making staging systems and diagnostic data available for product improvements. We continue to receive contributions from individual contributors, including a subject matter expert from a key technology player in this space who is interested in contributing designs and implementations to expand the feature set.
Earlier this week we published a change to the scheduling and cadence of community meetings to encourage wider participation and make them more accessible to those who have expressed an interest in being involved. We’re actively planning a UK-based community meetup and are open to holding them in other regions. We continue to raise awareness of the project through talks at SNIA, Data on Kubernetes and with a proposal for KubeCon NA 2022. We recognise that there is a lot more potential to build the community and will be exploring CNCF resources to assist with this.
With the acquisition of MayaData and the earlier successful incubation of LitmusChaos, some of the key maintainers have moved on to other areas of interest. We have identified several individuals whose consistent contributions and commitment make them deserving candidates as maintainers. We started updating the list a while ago, but partly due to inexperience with the process, this has stalled; we will prioritise getting it done. (@RichiH , #3548)
We are very interested in moving forward with incubation for OpenEBS, not least because of the recognition of project maturity that it represents. The last feedback we received from the Storage TAG was about the difficulty of knowing how to classify a project that contains multiple independent storage drivers. There is an obvious benefit for adopters in having a single project brand to tie things together, but it’s unclear how adoption and contribution should best be measured. As we’ve reflected on this advice, we’ve concluded that we should focus resources around the Mayastor engine which delivers the most strategic value, rather than on the other engines which are now mostly feature complete. We would welcome further guidance on how best to proceed with incubation from a CNCF perspective.
Following the acquisition, DataCore’s primary focus was to get the Mayastor engine to production quality as rapidly as possible. Achieving this has required a significant engineering investment and much of the work took place in local repos, whilst planning future involvement. Combined with the loss of some key maintainers, this resulted in a significant drop in apparent involvement as measured by devstats. The changes have now been merged into the public repos and development will continue with the open-source first approach that has always characterised OpenEBS. The devstats ‘Contributions chart’ reflects this, with a steady rise over the last three months.
Are we getting everything right? No, it’s quite clear we are not, but we are trying to! We’ve lost the active involvement, but not the support, of several key contributors. We’ve also taken longer to respond to some things than we should. However, as a primary backer of the project, DataCore is committed to investing in the OpenEBS community and in learning what it means to be good open-source citizens in this context. A measure of DataCore’s commitment to this as a long-term strategic initiative is that it envisages OpenEBS forming the nucleus of its future storage architectures.
Please bear with us as we navigate the changes and let us know where we can improve!
Thanks a lot for the details, @nconnolly1.
Are you able to guesstimate a timeline of the planned and started actions gaining traction? It's not urgent, we just want to anticipate a timeline for the graduation process.
Thanks a lot for the details, @nconnolly1.
Are you able to guesstimate a timeline of the planned and started actions gaining traction? It's not urgent, we just want to anticipate a timeline for the graduation process.
Hi @RichiH - I spoke to @GlennBullingham today and he is actively working on cleaning up the maintainers list, including contacting people to check whether they still wish to be involved. He's also cleaning up the permissions on the repos and will move on to the other areas afterwards.
I'll let @GlennBullingham comment on timescales.
@RichiH @nconnolly1
Hi @RichiH. As Nick has already said, I and some of the other maintainers/contributors are actively working to make updates and improvements to project administration in a number of areas, as well to put processes and documentation in place to ensure that this will be just the beginning of a sustainable initiative of better community engagement.
-
Maintainers list I'm reviewing this with current maintainers at this time. Some personnel changes related to the acquisition of MayaData by DataCore, plus some natural drift over the course of the project, means that some changes can be expected. Frankly, since this is the first time we've conducted such a review, we're learning on the job and will be reviewing the appropriate CNCF guidelines for this. To that last point, the timescale for this action will be the earliest opportunity that such practices, (I'm thinking of any mandated recipient response waiting times etc.) provide for.
-
OpenEBS Roadmap We've been preparing a draft update to this, which includes surfacing current work in progress on Mayastor by DataCore-based contributors, and expected release cadences. I expect the results of this review to be available to the community via GitHub by the end of this week (Friday 9th September).
-
Contribution Guidelines and Practices As a priority, we’re looking to compile a simple, consistent and well-documented set of practices for the same across the entire project. My expectation being to have something approved by the current maintainers and in-place within the next 30 days.
-
Contributions to Mayastor by the community are both welcome and essential. In order to make this as easy as possible and a rewarding experience for those contributors, we will be providing the community with the body of existing product and project documentation, which for Mayastor has tended historically to reside only with the maintainers. I expect this work to begin within the next 1-2 weeks, and to be complete within 30-45 days.
Should I have misunderstood your reference to timescales, or should there be other timescales of interest which I’ve not addressed here then please let me know.
Is there anything else that I can clarify or help with in the meantime?
Regards, Glenn
Thanks for the level of detail; looks good and we can mentally pencil in mid/end of October.
Sent by mobile; please excuse my brevity..