September 2019 - openstack-discuss

[heat] Resource handling in Heat stacks
by Andreas Florath 08 Sep '19

08 Sep '19

Hello! Can please anybody tell me, if all resources which are created within a Heat stack belong to the stack in the way that all the resources are freed / deleted when the stack is deleted? IMHO all resources which are created during the initial creation or update of a stack, even if they are ephemeral or only internal created, must be deleted when the stack is deleted by OpenStack Heat itself. Correct? My question might see obvious, but I did not find an explicit hint in the documentation stating this. The reason for my question: I have a Heat template which uses two images to create a server (using block_device_mapping_v2). Every time I run an 'openstack stack create' and 'openstack stack delete' cycle one ephemeral volume is left over / gets not deleted. For me this sounds like a problem in OpenStack (Heat). (It looks that this is at least similar to https://review.opendev.org/#/c/341008/ which never made it into master.) Kind regards Andre

3 4

[all][elections][ptl] Combined Project Team Lead and Technical Committee Election Conclusion and Results
by Jeremy Stanley 08 Sep '19

08 Sep '19

Thank you to all candidates who put their name forward for Project Team Lead (PTL) and Technical Committee (TC) in this election. A healthy, open process breeds trust in our decision making capability thank you to all those who make this process possible. Now for the results of the PTL election process, please join me in extending congratulations to the following PTLs: * Adjutant : Adrian Turjak * Barbican : Douglas Mendizábal * Blazar : Pierre Riteau * Cinder : Brian Rosmaita * Cloudkitty : Luka Peschke * Congress : Eric Kao * Documentation : Alexandra Settle * Ec2 Api : Andrey Pavlov * Freezer : geng chc * Glance : Abhishek Kekane * Heat : Rico Lin * Horizon : Akihiro Motoki * Infrastructure : Clark Boylan * Ironic : Julia Kreger * Karbor : Pengju Jiao * Keystone : Colleen Murphy * Kolla : Mark Goddard * Kuryr : Michał Dulko * Loci : Pete Birley * Magnum : Feilong Wang * Manila : Goutham Pacha Ravi * Masakari : Sampath Priyankara * Mistral : Renat Akhmerov * Monasca : Witek Bedyk * Murano : Rong Zhu * Neutron : Sławek Kapłoński * Nova : Eric Fried * Octavia : Adam Harwell * OpenStack Charms : Frode Nordahl * Openstack Chef : Jens Harbott * OpenStack Helm : Pete Birley * OpenStackAnsible : Mohammed Naser * OpenStackClient : Dean Troyer * Oslo : Ben Nemec * Packaging Rpm : Javier Peña * Puppet OpenStack : Shengping Zhong * Qinling : Lingxian Kong * Quality Assurance : Ghanshyam Mann * Rally : Andrey Kurilin * Release Management : Sean McGinnis * Requirements : Matthew Thode * Sahara : Jeremy Freudberg * Searchlight : Trinh Nguyen * Senlin : XueFeng Liu * Solum : Rong Zhu * Storlets : Kota Tsuyuzaki * Swift : Tim Burke * Tacker : dharmendra kushwaha * Telemetry : Rong Zhu * Tricircle : chi zhang * Tripleo : Wes Hayutin * Trove : Lingxian Kong * Vitrage : Eyal Bar-Ilan * Watcher : canwei li * Zaqar : wang hao * Zun : Feng Shengqin Also please join me in congratulating the 6 newly elected members of the TC: Ghanshyam Mann (gmann) Jean-Philippe Evrard (evrardjp) Jay Bryant (jungleboyj) Kevin Carter (cloudnull) Kendall Nelson (diablo_rojo) Nate Johnston (njohnston) Full results: because there were only as many TC candidates as open seats, no poll was held and all candidates were acclaimed Elections: Election process details and results are also available here: https://governance.openstack.org/election/ -- Jeremy Stanley, on behalf of the OpenStack Technical Election Officials

15 24

[tc][neutron] Supported Linux distributions and their kernel
by Adrian Chiris 08 Sep '19

08 Sep '19

Greetings, I was wondering what is the guideline in regards to which kernels are supported by OpenStack in the various Linux distributions. Looking at [1], Taking for example latest CentOS major (7): Every "minor" version is released with a different kernel version, the oldest being released in 2014 (CentOS 7.0, kernel 3.10.0-123) and the newest released in 2018 (CentOS 7.6, kernel 3.10.0-957) While I understand that OpenStack projects are expected to support all CentOS 7.x releases. Does the same applies for the kernels they originally came out with? The reason I'm asking, is because I was working on doing some cleanup in neutron [2] for a workaround introduced because of an old kernel bug, It is unclear to me if it is safe to introduce this change. [1] https://governance.openstack.org/tc/reference/project-testing-interface.htm… [2] https://review.opendev.org/#/c/677095/ Thanks, Adrian.

5 6

Need help trigger aodh alarm
by Anmar Salih 07 Sep '19

07 Sep '19

Hey all, I need help trigger aodh alarm to execute a simple function. I am following the instructions here <https://docs.openstack.org/qinling/rocky/user/aodh.html> but it does't work. Here are my system configurations: 1- Operating system (Ubuntu16 server) -> running on virtual machine 2- Devstack local.conf <https://paste.api-zulu.com/pebexowulo.makefile> file. 3- Devstack Stein release. Note: I tried to install Devstack on Ubuntu16 desktop and Ubuntu18 desktop but no luck. This link <https://imgur.com/a/hThei05> is the error output screen I received during the installation on Ubuntu18 desktop. Thank you in advance. Best Regards.

3 8

[infra][neutron] Requesting help to remove feature branches
by Miguel Lavalle 07 Sep '19

07 Sep '19

Dear Infra Team, We have decided to remove from the Neutron repo the following feature branches: feature/graphql feature/lbaasv2 feature/pecan feature/qos We don't need to preserve any state from these branches. In the case of the first one, no code was merged. The work in the other three branches is already merged into master. Best regards Miguel

3 5

[Octavia]-Seeking some high points on using Octavia
by Singh, Prabhjit 07 Sep '19

07 Sep '19

Hi Michael, I have been trying to get Octavia LbaaS up and running and get performance tested. It has taken me some time to get quite a few things working. While I continue to invest time in using Octavia and stay excited on some of the upcoming features. I have been asked the following questions by my leadership to which I do not have any direct answers. 1. What is the adoption of Octavia, are major organizations looking to adopt and invest in it. Can you provide some numbers 2. Roadmap wise is the Open community committed to investing in Octavia and why 3. Per your suggestion I tried to look up who are the primary companies using Octavia and haven't found a clear indication, any insight would be great. 4. Would features from haproxy 2.0 be included in Octavia 5. There are some open solutions from haproxy, Envoy, consul. How would Octavia compare. 6. Lastly, do you have enough encouragement to keep the project going, I guess I am looking for some motivation for continuing to choose to use Octavia when there are several turnkey solutions ( though offered at a price ). Currently I have been working with Redhat to answer the following questions, these are not for the community, hopefully Redhat will be able to pursue with your team. 1. How to offload logs to an external log/metrics collector 2. How to turn off logs during performance testing, I honestly do not want to do this because the performance tester is really generating live traffic which mimics a real time scenario. 3. How to set cron for rotating logs, I would think that this should be automatic. Would I need to do this everytime? 4. Do you have any way to increase performance of the amphora, my take is haproxy can handle several thousands of concurrent connections but in our case seems like we hit a threshold at 3500 - 4500 connections and then it starts to either send resets or the connections stay open for a long time. Thanks & Regards Prabhjit -----Original Message----- From: Singh, Prabhjit Sent: Tuesday, July 23, 2019 9:45 AM To: Michael Johnson <johnsomor(a)gmail.com> Cc: openstack-discuss(a)lists.openstack.org Subject: RE: [Octavia]-Seeking performance numbers on Octavia Thanks so much for the valuable insights Michael! Appreciate it and keep up the good work, as I ramp up with more dev know how hopefully I would start making contributions and can maybe convince my team to start as well. Thanks & Regards Prabhjit Singh -----Original Message----- From: Michael Johnson <johnsomor(a)gmail.com> Sent: Monday, July 22, 2019 5:48 PM To: Singh, Prabhjit <Prabhjit.Singh22(a)T-Mobile.com> Cc: openstack-discuss(a)lists.openstack.org Subject: Re: [Octavia]-Seeking performance numbers on Octavia [External] Hi Prabhjit, Comments in-line below. Michael On Sun, Jul 21, 2019 at 5:24 PM Singh, Prabhjit <Prabhjit.Singh22(a)t-mobile.com> wrote: > > Hi Michael, > > Thanks for taking the time out to send me your inputs and valuable suggestions. I do remember meeting you at the Denver Summit and hearing to a couple of your sessions. > If you wouldn't mind, I do have a few more questions and your answers would help me understand that should I continue to invest in having Octavia as one of our available LBs. > > 1. Based on your response and the amount of time you are investing in > supporting Octavia, what are some of the use cases, like for e.g. if load balancing web traffic how many transactions/connections minimum can be expected. I do understand you mentioned that it's hard to performance test Octavia but some real time situations from your testing and how customers have adopted Octavia would help me level set some expectations. This is really cloud and application specific. I would recommend you fire up an Octavia install and use your preferred tool to measure it. Some good tools are tsung, weighttp, and iperf3. > 2. We are thinking of Octavia as one of the offerings, that offers a self-serve type model. Do you know of any customers who have been able to use Octavia as one of their primary load balancers and any encouraging feedback you have gotten on Octavia. There are examples of organizations using Octavia available if you google Octavia. > 3. You suggested increasing the Ram size, I could go about making a whole new Flavor. Yes, to increase the allocated RAM for a load balancer, you would create an additional nova flavor with the specifications you would like. You can then either set this as the default nova flavor for amphora (amp_flavor_id is the setting) or you can create an Octavia flavor that specifies the nova compute flavor to use (See https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.open… for more information on Octavia flavors). > 4. I also noticed on the haproxy.conf the maxconns is set to 2000, should I increase this, does this affect the connection per server, which you said 64000 conns per server, so if I have 10 servers can I expect somewhere close to 640000 sessions? I think you are looking at the haproxy.conf file provided by your operating system package. Octavia does not use this file, it creates it's own HAProxy configuration files as needed under /var/lib/octavia inside the amphora. The default, if the user does not specify one at listener creation, is 1,000,000. > 5. Based on some of the limitations and the dev work in progress, I think the most important feature that would make Octavia a real solid offering would be the Active-Active and Autoscaling feature. I brought this up with you in our brief conversation at the summit, and you did mention that its not a top priority at this time and you are looking for some help. I have noticed a lot of documentation has been updated on this feature, do you think with the available document and progress I could spin up a distributor and manage sessions between Amphora or it's not complete yet. Active/Active is still on our roadmap, but unfortunately the people that were working on it had to stop for personal reasons. There may be some folks picking up this work again soon. At this point the Active/Active patches up for review are non-functional and still a work in progress. > 6. We have a Triple O setup, do you think I can make the above tweaks with the Triple O setup. I think you are able to make various adjustments to Octavia with Triple O, but I do not have specifics on that. > Thanks & Regards > > Prabhjit Singh > Systems Design and Strategy - Magentabox > | O: (973) 397-4819 | M: (973) 563-4445 > > > > -----Original Message----- > From: Michael Johnson <johnsomor(a)gmail.com> > Sent: Friday, July 19, 2019 6:00 PM > To: Singh, Prabhjit <Prabhjit.Singh22(a)T-Mobile.com> > Cc: openstack-discuss(a)lists.openstack.org > Subject: Re: [Octavia]-Seeking performance numbers on Octavia > > [External] > > > Hi Prabhjit, > > As you have mentioned, it is very challenging to get accurate performance results in cloud environments. There are a large number(very large in fact) of factors that can impact the overall performance of OpenStack and Octavia. > > In our OpenDev testing environment, we only have software emulation virtual machines available (Qemu running with the TCG engine) which performs extremely poorly. This means that the testing environment does not reflect how the software is used in real world deployments. > An example of this is simply booting a VM can take up to ten minutes on Qemu with TCG when it takes about twenty seconds on a real OpenStack deployment. > With this resource limitation, we cannot effectively run performance benchmarking test jobs on the OpenDev environment. > > Because of this, we don't publish performance numbers as they will not reflect what you can achieve in your environment. > > Let me try to speak to your bullet points: > 1. The Octavia team has never (to my knowledge) claimed the Amphora driver is "carrier grade". We do consider the Amphora driver to be "operator grade", which speaks to a cloud operator's perspective versus the previous offering that did not support high availability, have appropriate maintenance tooling, upgrade paths, performance, etc. > To me, "carrier grade" has an additional level of requirements including performance, latency, scale, and availability SLAs. This is not what the Octavia Amphora driver is currently ready for. That said, third party provider drivers for Octavia may be able to provide a "carrier grade" level of load balancing for OpenStack. > 2. As for performance tuning, much of this is either automatically handled by Octavia or are dependent on the application you are load balancing and your cloud deployment. For example we have many configuration settings to tune how many retries we attempt when interacting with other services. In performing and stable clouds, these can be tuned down, in others the defaults may be appropriate. If you would like faster failover, at the expense of slightly more network traffic, you can tune the health monitoring and keepalived_vrrp settings. We do not currently have a performance tuning guide for Octavia but would support someone authoring one. > 3. We do not currently have a guide for this. I will say with the version of HAproxy currently being shipped with the distributions, going beyond the 1vCPU per amphora does not gain you much. With the release of HAProxy 2.0 this has changed and we expect to be adding support for vertically scaling the Amphora in future releases. Disk space is only necessary if you are storing the flow logs locally, which I would not recommend for a performance load balancer (See the notes in the log offloading guide: > https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.open…). > Finally, the RAM usage is a factor of the number of concurrent connections and if you are enabling TLS on the load balancer. For typical load balancing loads, the default is typically fine. However, if you have high connection counts and/or TLS offloading, you may want to experiment with increasing the available RAM. > 4. The source IP issue is a known issue > (https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstoryboar…). We have not prioritized addressing this as we have not had anyone come forward that they needed this in their deployment. If this is an issue impacting your use case, please comment on the story to that effect and provide a use case. This will help the team prioritize this work. > Also, patches are welcome! If you are interested in working on this issue, I can help you with information about how this could be added. > It should also be noted that it is a limitation of 64,000 connections per-backend server, not per load balancer. > 5. The team uses the #openstack-lbaas IRC channel on freenode and is happy to answer questions, etc. > > To date, we have had limited resources (people and equipment) available to do performance evaluation and tuning. There are definitely kernel and HAProxy tuning settings we have evaluated and added to the Amphora driver, but I know there is more work that can be done. If you are interested in help us with this work, please let us know. > > Michael > > P.S. Here are just a few considerations that can/will impact the performance of an Octavia Amphora load balancer: > > Hardware used for the compute nodes > Network Interface Cards (NICs) used in the compute nodes Number of > network ports enabled on the compute hosts Network switch > configurations (Jumbo frames, and so on) Cloud network topology > (leaf‐spine, fat‐tree, and so on) The OpenStack Neutron networking > configuration (ML2 and ML3 drivers) Tenant networking configuration > (VXLAN, VLANS, GRE, and so on) Colocation of applications and Octavia > amphorae Over subscription of the compute and networking resources > Protocols being load balanced Configuration settings used when > creating the load balancer (connection limits, and so on) Version of > OpenStack services (nova, neutron, and so on) Version of OpenStack > Octavia Flavor of the OpenStack Octavia load balancer OS and > hypervisor versions used Deployed security mitigations (Spectre, > Meltdown, and so on) Customer application performance Health of the > customer application > > On Fri, Jul 19, 2019 at 8:52 AM Singh, Prabhjit <Prabhjit.Singh22(a)t-mobile.com> wrote: > > > > Hi > > > > > > > > I have been trying to test Octavia with some traffic generators and > > my tests are inconclusive. Appreciate your inputs on the following > > > > > > > > It would be really nice to have some performance numbers that you guys have been able to achieve for this to be termed as carrier grade. > > Would also appreciate if you could share any inputs on performance > > tuning Octavia Any recommended flavor sizes for spinning up Amphorae, the default size of 1 core, 2 Gb disk and 1 Gig RAM does not seem enough. > > Also I noticed when the Amphorae are spun up, at one time only one > > master is talking to the backend servers and has one IP that its > > using, it has to run out of ports after 64000 TCP concurrent > > sessions, id there a way to add more IPs or is this the limitation > > If I needed some help with Octavia and some guidance around > > performance tuning can someone from the community help > > > > > > > > Thanks & Regards > > > > > > > > Prabhjit Singh > > > > > > > > > > > >

2 1

[tc] Campaign Question: Treating the Problem, not just the symptoms- Burnout, No Polling, etc
by Kendall Nelson 07 Sep '19

07 Sep '19

Hello :) Wanted to split the question Chris Dent asked here[1] into its own thread so people down the road and those tracking it now can find more easily. To kind of rephrase for everyone (Chris, correct me if I am wrong or not getting all of it): What do you think we, as a community, can do about the lack of candidates for roles like TC or PTL? How can we adjust, as a community, to make our governance structures fit better? In what wasy can we address and prevent burnout? I think JP[2] and Jay[3] already started to enumerate good ideas on the other thread, so to summarize/ expand/ add to their lists: - Reducing the number of TC members to 9, and maybe someday down to 7. When we were having polls for every election (maybe not every project) it was at a time where the electorate (and theoretically the number of possible candidates) was also huge. Since we have move past the hype curve and stabilized as a project, the number of polls we've (I say we because I used to be and still plan to help with elections) had to make have decreased. It seems to be a matter of proportions. - Continuing to improve education and onboarding process. Agreed 100%, but this should be an ongoing focus for everyone too- every contributor TC, PTL, or otherwise. The best way to get more people involved faster is a lower barrier to entry, but we all know that. Yes some things like gerrit and IRC are hard for people to get past and likely won't be changing for our community any time soon, but there are things like that with every community (I don't know if you have ever tried to push patches to k8s but their tagging of PRs is something they are working on making less complicated and better documented). Breaking down the onboarding process we have at the moment into smaller modules and clearly documenting the progression through those modules for new comers to easily find and work through is important. Also, though, having that be the only place that we, as a community, point to (meaning no duplicate information in multiple places like we have today) when new contributors have issues. - Better documentation of tribal knowledge. I proposed as a community goal for the U release[4], to formalize project specific onboarding information (some teams have already done this) and project specific guides for PTLs (I know we already have the broad strokes for all PTLs documented fairly well, but there's always project specific stuff) so that when there is a turn over mid release, its easier for someone new to step up. - Utilize the user survey to gather info about how/why contribution is happening or why they aren't contributing if that's the case. There are already several questions there from the TC about this topic in the survey, but perhaps they can be re-framed if we aren't getting the info we want from them. As a reminder, here they are: -- To which projects does your organization contribute maintenance resources, such as patches for bug fixes and code reviews on master or stable branches? -- What prevents you or your organization from contributing more maintenance resources, or makes contributing difficult? -- How do members of your organization contribute to OpenStack? I think the real issue is getting larger vendors of OpenStack to get their users to take the user survey. We have a pretty solid reach as it is, but there are a lot of people using OpenStack that don't take the survey that we don't know about even because they are confidential (their results can still be confidential if they take the survey). - Longer release cycle. I know this has come up a dozen or more times (and I'm a little sorry for bringing it up again), but I think OpenStack has stabilized enough that 6 months is a little short and now may finally be the time to lengthen things a bit. 9 months might be a better fit. With longer release cycles comes more time to get work done as well which I've heard has been a complaint of more part time contributors when this discussion has come up in the past. - Co-PTL type position? I've noticed and talked to several PTLs on a variety of projects that need a little extra help with the role. They either don't feel like they have all the experience they need to be PTL yet and so they want the previous PTL to help out still or maybe they want to do it, but there are enough variables in their day to day work (or lack of overlap tz wise with most of the other contributors to that project), that having a backup person to help out and backfill when they need help. - Talking to each other. I really honestly think just talking to one another could help too. When you find yourself in a conversation with someone about how unmotivated they are because they have a ton of work to do. You might offer to take something off their plate. Or help them see maybe they need to not take on anything new till some other work gets wrapped up. We are a community that succeeds together, so if you see someone burning themselves out do what you can to help lighten their load (helping directly is great, but there are plenty of other people in our community that you could call on to help too). Hopefully goes without saying, but don't burn yourself out trying to help someone else either. Some of these things are more actionable, others are still high level and need to have concrete actions tied to them, but I think there are plenty of things we can do to make progress here. -Kendall (diablo_rojo) [1] http://lists.openstack.org/pipermail/openstack-discuss/2019-September/00908… [2] http://lists.openstack.org/pipermail/openstack-discuss/2019-September/00908… [3] http://lists.openstack.org/pipermail/openstack-discuss/2019-September/00910… [4] https://etherpad.openstack.org/p/PVG-u-series-goals

6 9

[neutron] CI issues
by Slawek Kaplonski 07 Sep '19

07 Sep '19

Hi neutrinos, We are currently having some issues in our gate. Please see [1], [2] and [3] for details. If Your Neutron patch failed on neutron-functional, neutron-functional-python27 or networking-ovn-tempest-dsvm-ovs-release jobs, please don’t recheck before all those issues will be solved. Recheck will not help and You will only use infra resources. [1] https://bugs.launchpad.net/neutron/+bug/1842659 [2] https://bugs.launchpad.net/neutron/+bug/1842482 [3] https://bugs.launchpad.net/bugs/1842657 — Slawek Kaplonski Senior software engineer Red Hat

1 1

[all][research] Survey for Openstack developers =)
by Farida El Zanaty 06 Sep '19

06 Sep '19

Hi!I am Farida from McGill University. I am trying to learn more about code reviews in the Openstack community, as I have been studying Openstack projects for a while. Please help me understand your perspective on design discussions during code reviews by filling up this 10-minute survey: https://forms.gle/Hhn191f6cxF5hVgG8 Survey participants will also be entered into a raffle for a $50 Amazon gift card. A little bit of context: Under the supervision of Prof. Shane McIntosh, my research aims to study design discussions that occur between developers during code reviews. Last year, we published a study about the frequency and types of such discussions that occur in OpenStack Nova and Neutron (http://rebels.ece.mcgill.ca/papers/esem2018_elzanaty.pdf<https://slack-redir.net/link?url=http%3A%2F%2Frebels.ece.mcgill.ca%2Fpapers…>).We are reaching out to Openstack developers to better understand their perspectives on design discussions during code reviews. Survey: https://forms.gle/Hhn191f6cxF5hVgG8<https://slack-redir.net/link?url=https%3A%2F%2Fforms.gle%2FHhn191f6cxF5hVgG8>Thanks for your time, Farida El-Zanaty =)

1 0

[all][tc] PDF Community Goal Update
by Akihiro Motoki 06 Sep '19

06 Sep '19

Hi all, I would like to share the status of the PDF community goal. We made great progresses these two weeks. Highlights ---------- It is time for working on the PDF community goal in your project! - openstack-tox-docs job supports PDF build now. AJaeger and I worked on the job support of PDF doc build and publishing and have completed it. - We confirmed PDF build with the sphinx latex builder works for several major projects like nova, cinder and neutron. The approach with the latex builder should work for most projects. - First PDF doc is now published from the i18n project. How to get started ------------------ - "How to get started" section in the PDF goal etherpad [1] explains the minimum steps. You can find useful examples there too. - To build PDF docs locally, you need to install LaTex related packages. See "To test locally" in the etherpad [1]. - If you hit problems during PDF build, check the common problems etherpad [2]. We are collecting knowledges there. - If you have questions, feel free to ask #openstack-doc IRC channel. Also Please sign up your name to "Project volunteers" in [1]. Useful links ------------ [1] https://etherpad.openstack.org/p/train-pdf-support-goal [2] https://etherpad.openstack.org/p/pdf-goal-train-common-problems [3] Ongoing reviews: https://review.opendev.org/#/q/topic:build-pdf-docs+(status:open+OR+status:… Thanks, Akihiro Motoki (amotoki)

9 15