[kolla] [swift] Seeking reviews for reinstating Swift support for 2025.2
Hello Kolla and OpenStack community, A few months ago, Swift support was removed from the Kolla project due to several challenges, including the need for role modernization, the lack of dedicated maintainership, and persistent CI failures. Following that, I volunteered to take on this effort. Over the last few months, I have been working to address these issues and am happy to report significant progress: - *Modernized Roles:* The Swift roles have been fully updated to align with the latest Kolla and Kolla-Ansible development standards. - *Reliable CI:* The associated Zuul jobs are now consistently green, with all tests successfully validated. - *Maintainership:* I have stepped up to maintain the Swift service within Kolla, ensuring its long-term health and stability. A considerable amount of effort has been invested to ensure this implementation is robust and maintainable for the community. With the *2025.2 release deadline just four weeks away*, the final and most critical step is to get these patches reviewed and merged. Your feedback and reviews are crucial to landing this work in time for the release. Reinstating Swift support would be a valuable addition for many operators. Any time you can spare to look over the patches would be greatly appreciated. You can find the main patch series here: https://review.opendev.org/q/topic:%22dev-swift-modernization%22 Thank you for your consideration and for all your contributions to the OpenStack community. Best regards, Amir Hossein (IRC: amir58118)
Hi Amir, As we have pointed out many times - the venue to talk about this is the weekly Kolla meeting. Kolla is a cycle trailing project, so it does not have the same deadline as other projects. Swift service has been removed just one cycle ago, because there were no maintainers - we can see your effort, but we don’t have a guarantee you’re going to disappear just after that gets merged - and there are no other contributors on the Swift patches than you, and you have not been long enough in the community for us to make that leap of faith. On the weekly meeting where there was a discussion about reinstating Swift - there was agreement that we might review it and ultimately merge it - but it is low priority. Majority of the Kolla project believes Swift should be an external service - just like Ceph is - but given your persistence we have sort of agreed for you to try to make that contribution. Currently - the Kolla code is nearly 1:1 to what it was before, I already commented on that - it needs to be adapted. Do you have data how many operators will use Swift in Kolla? Since we haven’t had any contributors for that functionality for years - I have reasonable doubt for that. Best regards, Michal
On 7 Sep 2025, at 13:12, Amir Hossein Ahmadi <sahahmadi96@gmail.com> wrote:
Hello Kolla and OpenStack community,
A few months ago, Swift support was removed from the Kolla project due to several challenges, including the need for role modernization, the lack of dedicated maintainership, and persistent CI failures.
Following that, I volunteered to take on this effort. Over the last few months, I have been working to address these issues and am happy to report significant progress:
Modernized Roles: The Swift roles have been fully updated to align with the latest Kolla and Kolla-Ansible development standards.
Reliable CI: The associated Zuul jobs are now consistently green, with all tests successfully validated.
Maintainership: I have stepped up to maintain the Swift service within Kolla, ensuring its long-term health and stability.
A considerable amount of effort has been invested to ensure this implementation is robust and maintainable for the community.
With the 2025.2 release deadline just four weeks away, the final and most critical step is to get these patches reviewed and merged. Your feedback and reviews are crucial to landing this work in time for the release.
Reinstating Swift support would be a valuable addition for many operators. Any time you can spare to look over the patches would be greatly appreciated.
You can find the main patch series here: https://review.opendev.org/q/topic:%22dev-swift-modernization%22
Thank you for your consideration and for all your contributions to the OpenStack community.
Best regards,
Amir Hossein (IRC: amir58118)
Hi Michal, Thank you for your detailed feedback and for highlighting the community’s concerns. Let me address them point by point. *Long-term maintainership:* This is not just my individual effort — it is the result of teamwork. Our team has already invested more than 100 hours of testing and development, and over 10 hours of internal review meetings to bring Swift support up to Kolla’s current standards. While my name is the one visible on the patches, this work reflects a collective effort. We are committed to maintaining Swift in Kolla beyond this cycle, continuously monitoring CI and fixing issues to ensure long-term stability. *Operator adoption:* You asked whether there is data about operators using Swift in Kolla. One of the reviewers, Roman Krček, has already pointed to the OpenStack User Survey which shows that ~40% of production deployments run Swift, with another ~15% either testing or interested [1]. This indicates that object storage remains critical for many operators. *On the importance of object storage in cloud infrastructure:* The philosophy of using objects as the right abstraction layer in cloud computing is rooted in the fundamental needs of this domain. Just as Linux abstracts hardware complexity through files, cloud platforms abstract infrastructure complexity through objects. This model provides flexibility, scalability, and ease of resource management — which is why object storage is so central to modern cloud environments. *Code adaptation and alignment: * Thank you for the valuable feedback you have already provided on both the Kolla and Kolla-Ansible patches. Your comments have been very helpful in highlighting where improvements are needed. We will make sure to address all of these points so that the Swift roles are not only restored, but also fully modernized and aligned with the current Kolla architecture. *Next steps:* We will continue this discussion in the weekly Kolla meeting and present both our technical progress and our long-term commitment there. Our goal is not just to merge these patches, but to demonstrate persistence in maintaining Swift as a first-class citizen in Kolla. Thank you for your consideration and for giving our team the opportunity to show sustained commitment. Best regards, Amir Hossein (IRC: amir58118) [1] https://openinfra.org/user-survey/ On Mon, Sep 8, 2025 at 10:41 AM Michał Nasiadka <mnasiadka@gmail.com> wrote:
Hi Amir,
As we have pointed out many times - the venue to talk about this is the weekly Kolla meeting. Kolla is a cycle trailing project, so it does not have the same deadline as other projects.
Swift service has been removed just one cycle ago, because there were no maintainers - we can see your effort, but we don’t have a guarantee you’re going to disappear just after that gets merged - and there are no other contributors on the Swift patches than you, and you have not been long enough in the community for us to make that leap of faith.
On the weekly meeting where there was a discussion about reinstating Swift - there was agreement that we might review it and ultimately merge it - but it is low priority. Majority of the Kolla project believes Swift should be an external service - just like Ceph is - but given your persistence we have sort of agreed for you to try to make that contribution.
Currently - the Kolla code is nearly 1:1 to what it was before, I already commented on that - it needs to be adapted.
Do you have data how many operators will use Swift in Kolla? Since we haven’t had any contributors for that functionality for years - I have reasonable doubt for that.
Best regards, Michal
On 7 Sep 2025, at 13:12, Amir Hossein Ahmadi <sahahmadi96@gmail.com> wrote:
Hello Kolla and OpenStack community,
A few months ago, Swift support was removed from the Kolla project due to several challenges, including the need for role modernization, the lack of dedicated maintainership, and persistent CI failures.
Following that, I volunteered to take on this effort. Over the last few months, I have been working to address these issues and am happy to report significant progress:
-
*Modernized Roles:* The Swift roles have been fully updated to align with the latest Kolla and Kolla-Ansible development standards. -
*Reliable CI:* The associated Zuul jobs are now consistently green, with all tests successfully validated. -
*Maintainership:* I have stepped up to maintain the Swift service within Kolla, ensuring its long-term health and stability.
A considerable amount of effort has been invested to ensure this implementation is robust and maintainable for the community.
With the *2025.2 release deadline just four weeks away*, the final and most critical step is to get these patches reviewed and merged. Your feedback and reviews are crucial to landing this work in time for the release.
Reinstating Swift support would be a valuable addition for many operators. Any time you can spare to look over the patches would be greatly appreciated.
You can find the main patch series here: https://review.opendev.org/q/topic:%22dev-swift-modernization%22
Thank you for your consideration and for all your contributions to the OpenStack community.
Best regards,
Amir Hossein (IRC: amir58118)
On Mon, Sep 8, 2025 at 10:41 AM Michał Nasiadka <mnasiadka@gmail.com> wrote:
Hi Amir,
As we have pointed out many times - the venue to talk about this is the weekly Kolla meeting. Kolla is a cycle trailing project, so it does not have the same deadline as other projects.
Swift service has been removed just one cycle ago, because there were no maintainers - we can see your effort, but we don’t have a guarantee you’re going to disappear just after that gets merged - and there are no other contributors on the Swift patches than you, and you have not been long enough in the community for us to make that leap of faith.
On the weekly meeting where there was a discussion about reinstating Swift - there was agreement that we might review it and ultimately merge it - but it is low priority. Majority of the Kolla project believes Swift should be an external service - just like Ceph is - but given your persistence we have sort of agreed for you to try to make that contribution.
Currently - the Kolla code is nearly 1:1 to what it was before, I already commented on that - it needs to be adapted.
Do you have data how many operators will use Swift in Kolla? Since we haven’t had any contributors for that functionality for years - I have reasonable doubt for that.
Best regards, Michal
On 7 Sep 2025, at 13:12, Amir Hossein Ahmadi <sahahmadi96@gmail.com> wrote:
Hello Kolla and OpenStack community,
A few months ago, Swift support was removed from the Kolla project due to several challenges, including the need for role modernization, the lack of dedicated maintainership, and persistent CI failures.
Following that, I volunteered to take on this effort. Over the last few months, I have been working to address these issues and am happy to report significant progress:
-
*Modernized Roles:* The Swift roles have been fully updated to align with the latest Kolla and Kolla-Ansible development standards. -
*Reliable CI:* The associated Zuul jobs are now consistently green, with all tests successfully validated. -
*Maintainership:* I have stepped up to maintain the Swift service within Kolla, ensuring its long-term health and stability.
A considerable amount of effort has been invested to ensure this implementation is robust and maintainable for the community.
With the *2025.2 release deadline just four weeks away*, the final and most critical step is to get these patches reviewed and merged. Your feedback and reviews are crucial to landing this work in time for the release.
Reinstating Swift support would be a valuable addition for many operators. Any time you can spare to look over the patches would be greatly appreciated.
You can find the main patch series here: https://review.opendev.org/q/topic:%22dev-swift-modernization%22
Thank you for your consideration and for all your contributions to the OpenStack community.
Best regards,
Amir Hossein (IRC: amir58118)
Hi, I’m also interested in this initiative and appreciate the effort Amir has put in. I’ll try to support by reviewing and contributing where possible. Best, Sara
Hi All, Also interested in the Swift Support, We would switch the management of double digits PetaByte Swift clusters to Kolla ansible then. Olivier On 2025-09-08T18:01:41.000+02:00, <sarah.kh09@gmail.com> wrote:
Hi, I’m also interested in this initiative and appreciate the effort Amir has put in. I’ll try to support by reviewing and contributing where possible. Best, Sara
On 2025-09-08 11:34:17 +0330 (+0330), Amir Hossein Ahmadi wrote: [...]
*Operator adoption:*
You asked whether there is data about operators using Swift in Kolla. One of the reviewers, Roman Krček, has already pointed to the OpenStack User Survey which shows that ~40% of production deployments run Swift, with another ~15% either testing or interested [1]. This indicates that object storage remains critical for many operators. [...]
Chiming in from a general OpenStack (not necessarily Kolla) user perspective, in OpenDev we have a mix of providers with Swift and with a Swift-like RadowGW API backed by Ceph. We only use the Swift providers for publishing CI logs because Ceph-based providers complained that their configuration didn't scale to the thousands of containers and billions of objects we were constantly adding and removing. I get the impression RadosGW is a convenient way to avoid deploying resources for yet another storage service if you're already depending on Ceph for other reasons, but has its own limitations and is not always a suitable substitute for actual Swift. -- Jeremy Stanley
participants (5)
-
Amir Hossein Ahmadi
-
Jeremy Stanley
-
Michał Nasiadka
-
Olivier Chaze
-
sarah.kh09＠gmail.com