- Zuul-discuss - lists.opendev.org

Rethinking the design of base jobs
by Paul Belanger 19 Jul '18

19 Jul '18

Greetings, I'd like to open a discussion around the idea of changing how base jobs work, today base jobs are stored in a trusted config project. Mostly, this is done because certain aspects of a base jobs need a secret, for example uploading logs to logs.exexmple.org. We also have other tasks in a base job, lets say configure-mirrors, which pre-populates the mirror info on a node from nodepool. As a recap, the main thing to remember about a config project, or trusted job, is we cannot leverage the speculative testing around jobs. We first need to merge a patch, then hope the job continues to work. In openstack-infra, we've create an idea of base / base-test jobs and some instructions[1] on how to interact with them. Now back to configure-mirrors role above, I believe there isn't really a reason for the role to live in a trusted job, nothing it does will affect zuul security (think secrets) or interact with zuul-executor (delagate_localhost). And given that changes to the role need to do the base / base-test dance outlined above, I'd like to make the following suggestion. We keep the idea of base job, but we split it into base (trusted) and base (untrusted). Given that today zuul cannot support 2 jobs with the same name, I think it would look more like. foo-zuul-jobs (untrusted project) - name: base parent: base-minimal config-project (trusted) - name: base-minimal parent: null This means all jobs, now parent to an untrusted project job, with the main benefit that speculative to base are more easily tested. And things that really need to be locked down (eg secrets), still live in the trusted base-minimal job (NOTE: open to renames). In the case of openstack-infra[2], I actually think a lot of the roles in pre.yaml can be moved into untrusted jobs, if we followed this idea. The post-run tasks, are almost all trusted context. Mind you, we don't really need to make this change in zuul, but I'm more thinking we should use it as the default recommendation for new users. I'm looking to start implementing the change in rdoproject.org but wanted to get the idea out before I proposed changes. [1] http://git.openstack.org/cgit/openstack-infra/project-config/tree/zuul.d/jo… [2] http://git.openstack.org/cgit/openstack-infra/project-config/tree/playbooks…

2 2

Potential enhancements to logging
by corvus＠inaugust.com 18 Jul '18

18 Jul '18

Hi, Some recent (and not-so-recent) issues with log hosting in OpenStack Infra have caused a renewed interest in reworking our log storing infrastructure. Some of the motivation and much of the initial work is OpenStack-specific, so I have written a post to the openstack-infra list describing that. Technically, none of what is in there affects the wider Zuul community, but folks reading this may want to read that message for background (or, if you follow our patterns closely, it may affect you as well). That message is here: http://lists.openstack.org/pipermail/openstack-infra/2018-July/006020.html Also, like that message, this one is the result of several conversations, especially with Monty; many of these ideas are his. Essentially, that message describes a process for us to begin hosting our logs in swift instead of on a file server. I'll note here that Zuul is (and will continue to be, at least for the foreseeable future) agnostic about where logs are hosted. Log storage is implemented as roles in base jobs, so any site can store logs where and how they wish. We can help each other by sharing the roles which perform these tasks. OpenStack moving its log storage to swift doesn't mean anyone else needs to -- storing logs on a sever (or even the executor) will still be supported. As would storing in S3, or any other system for which someone is motivated to write a role. That message also describes decommissioning our log-hosting server which happens to provide a number of useful services today. Among the these are dynamic generation of directory indexes, and HTMLification of text logs. Our plan is to pre-generate that data, however, that's not ideal, as it expands and duplicates the amount of data stored, and it also creates more work for the executors. We can expand the role of the Zuul dashboard to take on some of those duties, thereby reducing the storage requirements, and offloading the computation to the browser of the user requesting it. In the process, we can make a better experience for users navigating log files, all without compromising Zuul's independence from backend storage mechanisms. All of what follows should apply equally to file or swift-based storage. Let's start by creating a per-build page in the dashboard. Currently, we show a list of builds, and link to the log locations for those builds. In the future, we should link to a new page with details about the build. Not only so that we can show more info (which we already collect) than can be shown on one line, but also so that we have a place for the additional functionality which follows. If we have the uploader also create and upload a file manifest (say, zuul-manifest.json), then the build page can fetch[1] this file from the log storage and display an index of artifacts for the build. This can allow for access to artifacts directly within the Zuul web user interface, without the discontinuity of hitting an artifact server for the index. Since swift (and other object storage systems) may not be capable of dynamically creating indexes, this allows the functionality to work in all cases and removes the need to pre-generate index pages when using swift. We can also extend Zuul to store additional artifact information about a build. For example, in our docs build jobs, we override the success-url to link to the generated doc content. If we return the build location as additional information about the build, we can highlight that in a prominent place on the build page. In the future, we may want to leave the success-url alone so that users visit the build page and then follow a link to the preview. It would be an extra click compared to what we do today, but it may be a better experience in the long run. Either way would still be an option, of course, according to operator preference. Another important feature we currently have in OpenStack is a piece of middleware we call "os-loganalyze" (or OSLA) which HTMLifies text logs. It dynamically parses log files and creates anchors for each line (for easy hyperlink references) and also highlights and allows filtering by severity. We could implement this in javascript, so that when a user browses from the build page to a text file, the javascript component could fetch[1] the file and perform the HTMLification as OSLA does today. If it can't parse the file, it will just pass it through unchanged. And in the virtual file listing on the build page, we can add a "raw" link for direct download of the file without going through the HTMLification. This would eliminate the need for us to pre-render content when we upload to swift, and by implementing this in a generic manner in javascript in Zuul, more non-OpenStack Zuul users would benefit from this functionality. Any unknown files, or existing .html files, should just be direct links which leave the Zuul dashboard interface (we don't want to attempt to proxy or embed everything). Finally, even today Zuul creates a job-output.json, with the idea that we would eventually use it as a substitute for job-output.txt when reviewing logs. The build page would be a natural launching point for a page which read this json file and created a more interactive and detailed version of the streaming job output. In summary, by moving some of the OpenStack-specific log processing into Zuul-dashboard javascript, we can make it more generic and widely applicable, and at the same time provide better support for multiple log storage systems. -Jim [1] Swift does support specifying CORS headers. Users would need to do so to support this, and users of static file servers would need to do something similar.

3 5

Zuul project update 2018-07-16
by corvus＠inaugust.com 17 Jul '18

17 Jul '18

Hi Last week Tristan fixed issues with direct enqueueing of changes with ambiguous project names, Improved timer trigger to skip un-needed projects, and improving Github Event filter debugging. Tobias fixed a potentially serious bug with contamination of configuration. He's working on improving job freeze performance, and parent jobs which pause and continue to run while their children run (for resources and artifact exchange during a buildset). This week I plan on finishing the container spec. I've also written up some ideas for log storage which I may do some work on. We're waiting on a resolution to an issue with multi-tenancy in the new Javascript code. Once that's resolved, or we revert the new JS code, we should be in position to make a new release. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim

2 1

Building container images in nodepool
by corvus＠inaugust.com 16 Jul '18

16 Jul '18

Hi, I think the container spec is just about ready, but one part of it recently prompted a discussion that I think could use some wider input. Specifically, this is regarding the "container-as-machine" use-case. So think "run pep8 in a container" when you're thinking about this. There's a whole bunch of stuff in the spec about container-native workloads. This isn't it. The issue is whether or not Nodepool should be able to build container images, much as we do today for machines. Note that regardless, Nodepool will be able to use "upstream" container images from the local Kubernetes registry or Docker Hub. So this would be an optional feature (just as Nodepool can use extant cloud images). There are a few reasons to build local images (of any kind): * To provide a git cache. * To have more up to date packages than those provided externally. * To have a backup of a working image if something upstream changes. * To have a standardized image. * To add local requirements. To be fair, the last two might be the same thing said different ways. In an offline discussion prompted by the spec, Monty has suggested that running a git mirror in Kubernetes may obviate the need for using custom images for a git cache [related: applying this to VM images as well could reduce image sizes there; he's planning on writing up thoughts on this more fully later]. So if we take that out of the equation, the question becomes: Are the remaining reasons for building an image locally sufficient to build this functionality into Nodepool? A quick clarification: we're obviously not going to write a new image builder either way. We're talking about having Nodepool run some existing container image process on a periodic schedule and upload the result to the local registry. The main benefit to Nodepool doing this is using the same images across different registries, having a daemon performing the image builds automatically, and the ability to roll-back to previous versions. If you have thoughts on whether this is useful, or if it should be omitted in favor of relying more heavily on base images or external systems, please let me know. I'll try to collect feedback and update the spec early next week. Then we can merge it. :) -Jim

6 6

Improving new user experience
by corvus＠inaugust.com 11 Jul '18

11 Jul '18

Hi, Recently Artem has rightly pointed out that the experience for new users wanting to run Zuul is not good. We have some work in progress to improve that, and some efforts which are stalled and could use help. Here's what I think we should do: 1) Continue work on the Zuul From Scratch guide. This part of the documentation ends right after the Zuul services start up, but without telling anyone how to actually get Zuul to do anything useful. It needs to continue to explain: * Configuring a pipeline * What a base job needs to do * How to create a base job from scratch * How to use or copy the existing base jobs in zuul-base-jobs * How to use jobs from zuul-jobs * Configuring a project * Actually reporting on a change or pull request I'll note that the original Zuul From Scratch document focused only on GitHub and Fedora. Obviously there's a lot of options, and folks have added sections on Gerrit and other operating systems. The most important aspect of this document, however, is that it walk someone through the entire process of setting up Zuul for at least *one* configuration. If we can handle more, that's fine. Ideall, we can keep divergent steps to a minimum. For example, we may want to describe how to push a configuration change up once (for Github, and once for Gerrit) and then subsequently refer to that process generally when we discuss how to build pipelines, jobs, etc. Otherwise we'll have to write the entire document twice. 2) Separately, we should make the process of bootstrapping a running system easier. While Zuul From Scratch should describe everything needed to build an entire Zuul system, well, "from scratch", we should also have a system that makes it easy to deploy a working Zuul so someone can at least start playing with it, and perhaps even continue to run it in production. I'll call this the "quickstart", and it should replace the current erroneously titled "quickstart" in our docs. Monty is working on jobs which will automatically build and upload container images of Zuul to Dockerhub. Once those are in place, we can write a helm chart, or Ansible playbook, or something similar to deploy those with a very minimal configuration. The goal should be to get something up and running with the minimum amount of work and also the minimum amount of documentation. If you want to learn how a Zuul works, read Zuul From Scratch. If you want to just start breaking it, use the quickstart. I think any progress we make on these two items will drastically improve the experience for new users. -Jim

3 4

zuul-apb to install Zuul on K8/openshift using ansible-service-broker
by Artem Goncharov 10 Jul '18

10 Jul '18

Hi all, as promised in the IRC I am publishing POC work on preparing zuul-apb. It is intended to help (currently only myself) to install Zuul on top of Openshift (Kubernetes should be supported also, but with limited functionality and not tried so far). The project started to be initially used in my company, where after running zuul on VMs I decided to shift it to using Openshift. It is currently located in my personal repo in github with intention to be moved to corporate repo shortly. But before I do this I wanted to offer it to Zuul community for discussion. If there is enough interest in adopting it, I am open and will be glad to shift development under the Openstack hat. So here are some details: - Repo: https://github.com/gtema/zuul-apb - project is built to make use of source-to-image <https://github.com/openshift/source-to-image> to build own images. This is, however, only possible if you use Openshift (not the S2I tool itself, but presence of BuildConfigs). Example repo <https://github.com/gtema/s2i-zuul> with s2i image scripts. For plain Kubernetes this might be later solved by separate pod invoking s2i inside. - regular project-config repo is also mandatory, since it is used to fetch initial components configuration. It should contain zuul/{zuul.conf,main.yaml}, nodepool{nodepool.yaml} files. It is also planned to watch project-config changes and redeploy configMaps. - gearman certificates and all other secrets are passed as APB parameters (unless you use `dev` plan and bootstrap special variables file inside for dev purposes) and stored as Secret - zookeper cluster is created also part of the project (external cluster might be used, but I decided to initially provide own) - project is not yet used in production, therefore not everything is how it should be. I intend to work on this further. There are also some issues with AnsibleServiceBroker itself, which caused some delays and few iterations with upstream. So for example the Dockerfile bases on my upstream fork due to issues with ansible 2.7 (some bugs in `lookup`, causing line breaks to be broken). When changes upstream are published I will rebase back. - work is based on Tobias's shared Openshift templates - thanks Tobias - currently 2 components (nodepool-builder and zuul-executor) are requiring privileged accounts to work. This is kind of problem with APB approach (and generally using containers), since there is currently no way to deliver this nicely and it poses big security risk to the whole cluster. Therefore there is a manual step required for deployment to grant those privileges. On the other hand nodepool-builder is very poor on running inside of container due to lots of required privileges, security concerns, and bugs in `elements` with respect to i.e. SELinux handling ( https://bugs.launchpad.net/diskimage-builder/+bug/1779273) potential problems with cross-platform builds. Tristian had an idea of using nodepool-launcher to start VM with nodepool-builder (I personally like the approach). Otherwise it is anyway funny - VM is running container, which is building VM image ;-). Additionally Tristian volunteered to work also on zuul-executor not to required privileged account. - together with work on containerization of zuul jobs this might become a full solution of running zuul under K8/Openshift and complement Windmill with a VM installation approach. I would be glad, if you share your thoughts. Especially whether community is interested in having this as an official project. Regards and have a nice weekend, Artem

5 12

Canceling a running job
by Matthieu Huin 10 Jul '18

10 Jul '18

Hello, Yesterday we ran into a case where a periodic job had to be stopped in the middle of its execution. We ended up doing the following: * check the console stream to grab the nodes' IP addresses, from the ansible role that sets up firewall rules (this was a multi-nodes build) * SSH into each node and manually reboot the nodes This is obviously suboptimal, and possibly not pullable on single-node builds. Is there any better way to do this, or at least to get the nodes' IP addresses for any given build? Generally speaking, this use case is rather generic, as one might want to be able to stop a running job for a variety of reasons (free up resources, prevent "something bad" from happening accidentally, etc). Is this something we could add into Zuul? Is there a spec for this or should we start one? Cheers, MHU

4 3

Exciting opportunity to bikeshed over a naming question!
by Monty Taylor 09 Jul '18

09 Jul '18

Hey everybody, I've been working on jobs to build zuul container images. The obvious next step, once the build job works - is to publish them to dockerhub. This is where the bikeshed comes in. A while back, I registered dockerhub orgs for both "zuul" and "zuulci". Which org should we publish container images to? A part of me wants to use the 'zuul' one, since we have it. OTOH, the website is zuul-ci - so the other part of me thinks we should just use that. The end result of this will be docker images that look like: zuul/zuul-base zuul/zuul-executor zuul/zuul-scheduler zuul/zuul-web etc or zuulci/zuul-base zuulci/zuul-executor zuulci/zuul-scheduler zuulci/zuul-web etc Thoughts? Monty

5 6

Zuul project update 2018-07-09
by corvus＠inaugust.com 09 Jul '18

09 Jul '18

Hi, Last week Shrews tracked down and fixed a Nodepool bug preventing the launcher from finding unmanaged images by ID[1]. Paul is working on a series of changes to enable jobs to specify that specific child jobs should be skipped[2][3]. I have been "helping" Tobias fix a bug related to dynamic configuration. Discussion is continuing on the container spec, but I think it's getting very close to approval. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim [1] https://review.openstack.org/579664 [2] https://review.openstack.org/578181 [3] https://review.openstack.org/578230 [4] https://review.openstack.org/560136

1 0

Dashboard routing proposals
by Tristan Cacqueray 05 Jul '18

05 Jul '18

Hi, with the recent angular6 refactor, the web interface components routing slightly changed as it is now managed by the navigation javascript component. Unfortunately this broke multi-tenant setup and we need an extra fixes to support it. Here are some options: A) Make the navigation component call a relative 'api/info' endpoint onInit to be able to introspect the deployment mode: the returned json may contains a 'tenant' value to indicate this is a white label setup. B) Always assume white-label and document that multi-tenant needs to redirect the entrypoint to 't/tenants.html' C) Always assume multi-tenant and document that white-label needs a rewrite rule to redirect the entry-point to 'status.html' D) Rewrite the navigation endpoint to rely on a '#!' anchor, this would be a mix of the first option (A) to introspect the setup mode, but using a single 'index.html' file instead. I would prefer D because the current design makes multi-tenant setup looks special with the needed '/t/' path and it feels like a second-class citizen. On the other hand, B or C sounds simple enough and this wouldn't delay much the next release. Regards, -Tristan

2 2

Zuul project update 2018-07-02
by corvus＠inaugust.com 02 Jul '18

02 Jul '18

Hi, This past week has seen: Monty upgrade the zuul dashboard to Angular 6 and typescript. It seems like that's mostly working now, though there may still be an issue in the multi-tenant hosting configuration. Mohammed fixed status page to not show empty headers when filtering for a specific project (a long requested feature)! Tristan found more issues with latest Nodepool master version regarding multiple pool using different labels (fixed in [1] by David), worked on changes to the log-inventory role[2], and contributed to the Angular 6 work. He investigated a couple of exception being raised in the github driver, and started to work on improving logging by promoting important log messages from debug to info[3]. Started to work on selenium tests[4]. I started looking into adding support for leaving in-line comments. I'll continue to work a bit on that this week, but mostly focus on finishing up the container spec and hopefully getting it merged. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim [1] https://review.openstack.org/578831 [2] https://review.openstack.org/#/q/topic:log-inventory-fixes [3] https://review.openstack.org/#/q/topic:improve-logging [4] https://review.openstack.org/579039

1 0

Deprecation: zuul-info/inventory.yaml creation moved from validate-host to log-inventory role (v2)
by Ian Wienand 01 Jul '18

01 Jul '18

Hello (again), This is a follow-up to a prior mail http://lists.zuul-ci.org/pipermail/zuul-discuss/2018-June/000446.html --- The "validate-host" role [1] currently creates the "inventory.yaml" [2] file, which contains the ansible inventory information for the job. For some history, the role was originally called "debug-ansible" but renamed to "validate-host". Since it collects other ansible information, saving the inventory was added to the role shortly after the rename. As copying this inventory information requires access to the executor context, it must be run as a privileged job. However, the role also includes quite a number of other operations that do not have similar requirements. Due to not currently having an automated CI testing framework for privileged jobs, this leaves changes to these unrelated parts (like [3]) with insufficient testing coverage. The creation of the inventory file has been moved into a separate role "log-inventory" [4]. We now wish to remove the inventory creation from the "validate-host" role [5]. This will lower the surface-area of privileged jobs and also enable more straight-forward testing of the rest of the "validate-host" role [6]. For the OpenStack ecosystem, the new role will be added to the base job [7] so this should be entirely transparent. If you build your own base templates you will need to add the "log-inventory" role to continue collecting the inventory.yaml. You can do this *now*, and thus be unaffected when the creation is removed from "validate-host", with one caveat: Note that testing uncovered an issue with the original role -- the "inventory_file" variable is not set by Ansible for localhost only jobs. Thus "log-inventory" will not break any jobs where "validate-host" was working, but you should be careful not to additionally introduce it into localhost (that is to say, executor-only) jobs without considering this. Longer term, the issue has been fixed and "log-inventory" is safe to use in all contexts. A separate variable for the inventory file has been added to the executor context with [8] and is now conditionally used in "log-inventory" with [9] (thus if you are using zuul from master, you are unaffected). Per the deprecation policy for requiring new zuul features [10], this new variable will be used unconditionally to copy the file with [11] around 4 weeks after the next zuul release. Taking into consideration: * reviews posted for several weeks * pending debugging of gate issues with "validate-host" * the informational/debugging nature of the output * the relative ease of adding the new role to maintain the status quo * interaction with current users of the roles I would suggest we merge [7] - addition of "log-inventory" to base jobs in project-config [5] - removal of inventory copy from "validate-host" To complete the initial part of this transition. Thank you -i [1] https://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/validat… [2] see any job results zuul-info/inventory.yaml [3] https://review.openstack.org/576351 [4] https://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/log-inv… [5] https://review.openstack.org/563790 [6] https://review.openstack.org/563788 [7] https://review.openstack.org/563789 [8] https://review.openstack.org/578235 [9] https://review.openstack.org/578234 [10] https://zuul-ci.org/docs/zuul-jobs/deprecation.html [11] https://review.openstack.org/578238

1 0

Removing the add_host exclusion
by Monty Taylor 30 Jun '18

30 Jun '18

Hey everybody, While working on the OpenStack Infra spec for updating our config management to use Ansible, which involves some pieces of using Zuul for CD, I ran in to an issue related to add_host - which we currently prevent people from using. I'd like to use add_host in jobs in a non-trusted repo - but add_host is on the list of things we block for security reasons. I **think** we're blocking it as a hold-over from the time before we had per-job ssh keys (we were worried about a user using add_host to connect to another job's nodes) Since we do per-build keys now, that vector is gone - and I can't come up with any more reasons why we should be blocking add_host from use. It can't be used to attack anything that we don't already protect from in a different place (like add_host with a connection_type: local would still get blocked on usage by the 'can't execute on localhost' exclusions) So I'd like to relax the exclusion - unless someone has a reason we should keep it? Thanks! Monty

3 2

Matching tags against branches
by corvus＠inaugust.com 29 Jun '18

29 Jun '18

Hi, Recently a discussion came up in OpenStack about moving some job invocations out of our global config-project and into individual project repos. Some of the jobs in question run in our "release" pipeline, which is triggered by tag events. These pipelines have Tag items enqueued into them, rather than Change items. Tags have a ref (the name of the tag), but they don't have a branch. This is because, in git, a tag may point to a commit that is on one branch, many branches, or no branches at all. Moreover, because of this, none of the code review systems attempt to provide "branch" information for tags. The current behavior in Zuul is that explicit branch matchers only match branches or refs. So you can define a job with "branches: stable", and it only run on changes to the "stable" branch. Or you can define a job with "branches: 3\..*" and it will only run on 3.x tags. Or you can say "branches: [stable, 3\..*]" and it will run on both. Implicit branch matchers match branches if the item has one (eg, a Change). But if the item does not (eg, a Tag), they always match. This means that a project can add a job to a tag-based pipeline in-repo, and it will work. The project stanza will get an implicit branch matcher, but that will always match the tag item, so the job will always run. But this means that ever project stanza for the project matches. So if the stable branch says "run foo on release" and the master branch says "run bar on release" both foo and bar will run on release. There is no way to implicitly have different jobs running on tags "from different branches" (I put that in scare quotes because that's an artificial construct that we're placing on git). Of course this can be done with explicit branch matchers, but determining which version numbers correspond to which branches and building those regexes can be high maintenance. If we want to make this a bit more intuitive, there are two approaches we could take. One would be to perform a dynamic reconfiguration for all tag items to determine the actual in-repo content for that tag, and use it to the exclusion of all other in-repo content from that project. That would produce the behavior of running whatever was defined in that commit. This is very intuitive for folks coming from other CI systems, however, it's rather resource intensive for Zuul, and could pose problems if, for example, a stable branch project-pipeline used a job defined on the master branch. The other approach would be to attempt to create the fiction that tags have branches. We could ask the mergers to run "git branch --contains" on the tagged commit and find out which branches it is in. We can then compare that set of branches to any explicit and implicit branch matchers. If any of the branches match, we'll consider the Tag item to have matched. This will probably do the expected thing in the typical case. There are still some edge cases, but they generally degrade to the situation we have today, in that multiple versions of a job will all be run. Of course, in any case (now or with this change) full control can always be achieved in a config-project with explicit branch matchers. I've begun implementing this idea, mostly as an exercise in exploration. I'm tentatively inclined to think that it's an improvement and we should consider proceeding with it. But I'd really like other folks to think about whether it's a good idea, and whether it might produce more problems than it solves. Here's the initial implementation; it's lacking only documentation at this point. Some of the edge cases are described in the commit message, so I refer you to that for further reading: https://review.openstack.org/578557 Thanks, Jim

3 4

Deprecation: zuul-info/inventory.yaml creation moved from validate-host to log-inventory role
by Ian Wienand 26 Jun '18

26 Jun '18

Hello, The "validate-host" role [1] currently creates the "inventory.yaml" [2] file, which contains the ansible inventory information for the job. For some history, the role was originally called "debug-ansible" but renamed to "validate-host". Since it collects other ansible information, saving the inventory was added to the role shortly after the rename. As copying this inventory information requires access to the executor context, it must be run as a privileged job. However, the role also includes quite a number of other operations that do not have similar requirements. Due to not currently having an automated CI testing framework for privileged jobs, this leaves changes to these unrelated parts (like [3]) with insufficient testing coverage. The creation of the inventory file has been moved into a separate role "log-inventory" [4]. We now wish to remove the inventory creation from the "validate-host" role [5]. This will lower the surface-area of privileged jobs and also enable more straight-forward testing of the rest of the "validate-host" role [6]. For the OpenStack ecosystem, the new role will be added to the base job [7] so this should be entirely transparent. However, if you build your own base templates, you will need to add the "log-inventory" role to continue creating the inventory.yaml. Note that you can do this *now*, and thus be unaffected when the creation is removed from "validate-host". Per the deprecation policy [8], but taking into consideration * reviews posted for several weeks * pending debugging of gate issues with "validate-host" * the informational/debugging nature of the output * the relative ease of adding the new role to maintain the status quo I would suggest we merge these changes at the start of next week, around the 25th June Thank you -i [1] https://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/validat… [2] see any job results zuul-info/inventory.yaml [3] https://review.openstack.org/576351 [4] https://git.openstack.org/cgit/openstack-infra/zuul-jobs/tree/roles/log-inv… [5] https://review.openstack.org/563790 [6] https://review.openstack.org/563788 [7] https://review.openstack.org/563789 [8] https://zuul-ci.org/docs/zuul-jobs/deprecation.html

3 4

Zuul project update 2018-06-26
by corvus＠inaugust.com 26 Jun '18

26 Jun '18

Hi, I was away giving a talk on Zuul last week. It looks like it's been a week of bugfixes: Tristan found a bunch of issues after upgrading their deployment: * Zuul refresh_deps doesn't work well with gerrit 2.14 schema change (change/patchset number are now int): https://review.openstack.org/577154 * Zuul configloader doesn't skip project that include: [] https://review.openstack.org/576016 * Nodepool repeatedly creates new deleting node when OpenStack provider fails to remove an instance: https://review.openstack.org/576288 He also split the initial connection loop out of the zk-retry patches: https://review.openstack.org/576047 and https://review.openstack.org/576048 . This make Zuul restart properly if the control plane reboots. Rebased runC, k8s and openshift nodepool drivers in preparation of the upcoming next release of Nodepool. Tobias worked on tenant reconfigure behavior[1] and on possible multi-enqueue/missing-dequeue issue in the github driver. [1] https://review.openstack.org/#/q/project:openstack-infra/zuul+branch:master… Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim

1 0

Re: Unintended variable disclosure on Ansible unreachable tasks despite no_log
by Jeremy Stanley 19 Jun '18

19 Jun '18

On 2018-06-18 13:47:24 +0000 (+0000), Tobias.Henkel(a)bmw.de wrote: > We discovered that if nodes become offline during the build the > no_log attribute of a task got ignored. If the unreachable error > occurred in a task used with a loop variable (ex: with_items), the > contents of the loop items would be printed in the console. This > could lead to accidentally leaking credentials or secrets. [...] MITRE has assigned CVE-2018-12557 to this vulnerability. -- Jeremy Stanley

1 0

Zuul project update 2018-06-18
by corvus＠inaugust.com 18 Jun '18

18 Jun '18

Hi, TristanC has started experimenting with graphQL. The idea is to see how it can be used to simplify the REST apis as well as enable powerful UI using appolo, in particular for pagination. Worked on extra-vars, ansible-tags and ansible-skip-tags job attributes. Found a couple more issues with the Angular6 patch when working on the new project and job page. I worked with Tobias and Clark on fixing a log streaming problem due to the ansible 2.5 upgrade. This week I'm traveling to China to deliver a talk about zuul at OpenInfra Days. I won't be around in channel as much, but should be able to review changes while in transit. Shrews completed static driver rewrite to pre-register nodes. Awaiting review/merge. Identified and fixed an issue where paused request handlers were not removed from active handlers if an exception from a provider operation occured. This potentially caused leaked nodes that remained ready+locked until a launcher restart. Tobiash worked on log streaming problem as mentioned above. Fixed ignored no_log caused by unrechable nodes (a security issue). Fixed caching of empty config. This week: work on improving tenant reconfigure behavior with github. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim

1 0

Zuul participation in OpenCI
by corvus＠inaugust.com 15 Jun '18

15 Jun '18

Hi, A number of us have been involved in the OpenCI effort: https://openci.io/ I think those of us who are, are bringing our experience with Zuul to the group, so it's fair to characterize the Zuul project as participating. It has been suggested that our logo could be added to the OpenCI web page under the "Practitioners from the following Communities" section. Does anyone think this is a good or bad idea? Please feel free to reply to me publicly or privately with any feedback. -Jim

4 3

September PTG attendance
by corvus＠inaugust.com 14 Jun '18

14 Jun '18

Hi, So that folks can start planning for the next PTG, I'd like to find out if we think it's valuable to have a formal Zuul track there, and if folks are planning to attend. I've set up an etherpad to collect information: https://etherpad.openstack.org/p/zuul-denver-ptg-2018 If you could fill it out and indicate whether you: plan to attend, or may attend if there's a standalone Zuul track, or if you won't be able to attend at all. Also, if you have a topic you would like to discuss in person at the PTG, please describe it there. I'll try to collect these within a couple days and we can have a better idea of whether we should make an effort to have a Zuul track, or skip this event in favor of working through our existing backlog. -Jim

1 0

Adding tags, extra-vars files and multiple run to job
by Tristan Cacqueray 14 Jun '18

14 Jun '18

Hi, TripleO folks need some features to be able to re-use deployment playbook natively from a Zuul job. Would it be possible to: * https://storyboard.openstack.org/#!/story/2001353 Adds "tags" and "skip-tags" job attributes that would be used as "--tags" and "--skip-tags" ansible-playbook command line arguments. * https://storyboard.openstack.org/#!/story/2001354 Adds a "extra-vars" job attribute that would be used as "--extra-vars @file.yaml" ansible-playbook command line arguments * https://storyboard.openstack.org/#!/story/2002543 Make the "run" job attribute a list of playbook and make the run phase chain ansible-playbook executions. It seems like those features are simple to implement, so if there are no objections I'd like to work on them. Regards, -Tristan

5 6

Zuul project update 2018-06-12
by corvus＠inaugust.com 12 Jun '18

12 Jun '18

Hi, We landed the changes to upgrade to Ansible 2.5. Things mostly seem to be going well, but we're currently working through a few issues described in story 2002528. Tristan has rebased the extra zuul-web routes with the new cherrypy interface: https://review.openstack.org/#/q/topic:zuul-web-routes and rebased the extra webui pages with the new angular6 interface: https://review.openstack.org/#/q/topic:zuul-ui-pages Monty seems to be updating the angular6 patches, which I guess prompted Tristan's update. Shrews reports that the multi-label changes have merged and he is ready to focus on static driver work now. He also Reviewed and carried Jim's driver API changes through to merge. (Thanks!) Tobias has a number of bug fixes in review. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email -Jim

1 0

Multi-tenancy architecture
by Mohammed Naser 05 Jun '18

05 Jun '18

Hi everyone: We're evaluating running Zuul in a multi-tenant environment by leveraging the following feature: https://docs.openstack.org/infra/zuul/admin/components.html#attr-scheduler.… Now, the idea is to make it as self-serve as possible. I've been toying around with a few ideas in terms of implementing tenants and I've come to the following simplistic idea: - Every 'user' has a 'tenant' inside this environment - All projects share the same `config-projects` which contains the base job and publishing artifact and logs roles (config-project is not controlled by tenant) - `untrusted-projects` are all dynamically loaded, provided by the user This would mean that user would supply a project using $method, and as long as they drop in a .zuul.yml file that is appropriate, things will get going automatically (implying that a basic check/gate/post pipeline is in place for all users). The only challenge with this at the moment is that multi-tenancy in Nodepool doesn't seem to exist, which means that everyone will be using the same pool of resources, rather than possibly having every tenant run their jobs inside their own project. Is there any plans to look into addressing the nodepool issue (or maybe ways to workaround it?) and as a more general question, is this an "ok" way of doing things? Your responses are appreciated! Thanks, Mohammed

3 4

Upgrade to Ansible 2.5
by corvus＠inaugust.com 05 Jun '18

05 Jun '18

Hi, We have some patches ready to upgrade Zuul to use Ansible 2.5, but there is some concern about whether there may be incompatibilities with playbooks and roles in zuul-jobs. I know some folks are running with local patches to use later Ansible versions -- if you are one of them, and you are using zuul-jobs, please let me know so we have an idea of what to expect. How do folks feel about merging the upgrade-to-2.5 changes and releasing those as Zuul 3.1.0 -- say, next Thursday, May 10? If we do that, and we find that there are incompatibilities, we will need to begin merging fixes to zuul-jobs immediately. Hopefully, however, those fixes can use the "ansible_version" variable in order to be backwards compatible, so zuul-jobs continues to work for folks who haven't upgraded yet. Finally, we hope to soon have a spec detailing multi-version support for Zuul, so that future upgrades can be handled more gracefully. -Jim

5 8

Zuul project update 2018-06-04
by corvus＠inaugust.com 04 Jun '18

04 Jun '18

Hi, I skipped a few of these due to the Summit and related discombobulation, but it's back! The Summit and OpenDev was really good. We introduced Zuul to a lot of folks. A number of Zuul devs managed to find some time to talk in person about the container spec. I've pushed another revision of that, hopefully it's close to ready. Artem Goncharov (gtema) has started implementation of nodepool security group support. Tristan followed up on summit discussion around container and worked on a prototype implementation of an openshift resource provider[1]. He also started some howtos, recommended practice documentation about PTI[2], renamed the proposed OCI driver to RUNC, and added documentation to nodepool builder about setting diskimage env-vars in secure configuration. I finished the patches to convert zuul-web to cherrypy, those merged at the end of last week. I also pushed up two changes to implement features I've been promising for a while: overridable file matchers, and a new pipeline manager ("supercedent"). I'm hoping we merge the switch to Ansible 2.5 this week. Shrews has been working on refactoring the nodepool API to better support our different drivers, and especially the multi-label support needed by the static driver. Monty has been refreshing his angular v1 -> v5 and related patches. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim [1] https://review.openstack.org/#/q/topic:openshift-zuul-build-resource [2] https://review.openstack.org/571420

1 0

Behavior change for file matchers
by corvus＠inaugust.com 01 Jun '18

01 Jun '18

Hi, A while back we had a discussion about changing the behavior of file matchers to be more intuitive. The consensus seemed to be that treating them as normal overridable attributes would be the best solution. In particular, that would allow folks to more easily define a file matcher in a job, and then override it at the project-template or project level. See this thread for the full context: http://lists.zuul-ci.org/pipermail/zuul-discuss/2018-May/000397.html I've implemented Proposal 2 from that message: https://review.openstack.org/571745 I think we should merge that, and, after we're satisfied with it, release version 3.1 with it, to signify that it's an end-user visible behavior change. It will cause changes in which jobs are run, in ways that would be difficult to enumerate ahead of time. But it should only do so in more complex situations, and hopefully the folks in those situations will be relieved that the solution is easier to comprehend than the status quo. Along with that change I have also proposed a followup which causes files and irrelevant-files to become mutually exclusive options: https://review.openstack.org/571804 If one is used in a parent and the other in a child, the attribute used in the child wins and clears that which was set in the parent. This makes them easier to reason about, though it does so largely by making them less useful. I suspect that the class of things that it lets us express (example in that commit message and in the previously linked thread) can be expressed today with a single regular expression. However, we're looking to move these matchers to use re2 which lacks negative lookaheads. That may mean that if we merge both this change and the re2 switch, we may not be able to express that configuration at all anymore. It may still be worth merging this change to save Zuul admins from pulling their hair out when people ask about the interaction of the two options. Or we may want to keep the ability to use both because, when used well, it does express something unique and succinct. I'd like for folks to consider this change carefully and provide as much feedback as possible before we decide what to do. If we do decide to merge it, it probably makes sense to do so in the same release (3.1) as the previous change. -Jim

1 0

Templates in job vars
by Ian Wienand 29 May '18

29 May '18

Hi, In templating out some diskimage-builder jobs, there is a situation where we pass a list of tests to run, but some platforms can't run various tests. So something like --- - job: name: base abstract: yes ... vars: dib_functests: - gentoo - debian - job: name: test-on-bionic parent: base - job: name: test-on-xenial parent: base vars: dib_functests: - gentoo - debian - suse # also run suse, redefine whole list --- That the base list of tests is repeated to override the dib_functests is a little ugly. I would prefer if the base tests that all platforms can run was defined only once (so if it's updated, you can't miss one). To my mind, it would be better if I could jinja on the variable, so I could add (or remove) a test; above it would become like: # also run opensuse tests on xenial vars: dib_functests: "{{ dib_functests + ['opensuse'] }}" I ended up just appending a second variable [1], but it feels a bit hacky (if you had more levels, you'd need to keep appending more and more variables). Seeing as I'm pretty sure jinja templates are Turing complete, I can also see issues with expanding each vars: entry with it :) Terrible idea? Already better ways I don't know about? Cheers, -i [1] https://review.openstack.org/#/c/566779/60/roles/dib-functests/defaults/mai…

2 1

Zuul project update 2018-05-14
by corvus＠inaugust.com 14 May '18

14 May '18

Hi, Shrews is beginning implementation of multi-label support in nodepool as part of his continuing work on better integrating the static driver. I spent most of last week and will continue to spend most of this week preparing for the upcoming OpenStack Summit and OpenDev events. However, at the end of the week, I needed a break from that, so I started work on replacing our use of aiohttp with cherrypy. Aiohttp (and related libraries) have broken us a few times recently by rapidly dropping support for what they consider "older" python versions -- yet versions that are still under support on the operating systems we use. Also, several of us have found the python async implementation unnecessarily complex and verbose. Cherrypy supports a thread-based model which fits in well with the rest of Zuul. I don't plan on sending this email next week due to the Summit, so expect the next installment on May 28. Please help write that email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim

1 0

Mission statement and tagline
by corvus＠inaugust.com 09 May '18

09 May '18

Hi, The Foundation marketing team has indicated that it would be good to have a mission statement and tagline. The mission statement can be used to help convey what we're trying to accomplish as a community, and the tagline will likely feature prominently on the website and other material. I'd like us all to be involved in creating these, but we need to have the process finished by May 10 in order to be able to update marketing material before the Summit/OpenDev. I propose the following process: * We make (and refine) suggestions this week. Add suggestions to https://etherpad.openstack.org/p/zuul-mission and discuss in the etherpad or here on the mailing list. Feedback from the zuul-core team and the Foundation marketing team (some of whom are subscribed to this list) will be important to formulate good candidates. * Next Monday, the zuul-core team and Foundation marketing team veto any suggestions which are inappropriate. I don't think it's fair to have a mission statement the core team doesn't agree with, and we should trust the marketing folks as domain experts to help us avoid pitfalls. Email me with veto requests; I'll produce a final list. * If there isn't a clear favorite, well have a condorcet election with the remaining candidates, to be complete by May 10. I'll run the election. * If we've vetoed everything, we'll have a special IRC meeting to come up with *something*. To start things off, I've populated the etherpad with a mission statement suggestion from Clark, and some of the favored tagline suggestions from earlier brainstorming. -Jim

2 2

Zuul Stickers
by James Cole 08 May '18

08 May '18

Hey everyone! I just wanted to share a pic of the stickers we’re giving away at the Summit in Vancouver. A link is below in case you can’t see the attachment. https://www.dropbox.com/s/dwb8df3wuqey5i6/Zuul_Stickers.jpg?dl=0 Come grab one in Vancouver! James Cole Graphic Designer OpenStack Foundation

1 0

Polling users of Nodepool's static node driver
by David Shrewsbury 08 May '18

08 May '18

Hi, Is anyone currently using the static driver? And if so, are you making us of the multi-label feature it supports for the static nodes? I ask because I've been trying to change the behavior of the driver to pre-populate ZooKeeper with the static nodes (more in line with how the OpenStack driver works), rather than only having the ZK nodes appear when the node is actually in-use. Doing so fundamentally changes how the static driver works, and this deep dive of the code has revealed this multi-label usage which I think many of us (certainly me) have missed. Changing the driver to operate more like how we expect will mean changing nodepool to support multiple labels across *all* drivers. That's not a simple change, and affects other things, such as statsd reporting (which reports per-label stats). (And if you were wondering how the static driver was getting away with supporting multiple labels before, it was the fact that it *didn't* pre-register its nodes with ZK that allowed that.) One one hand, I can see how supporting multiple labels might be useful (e.g., a node labeled with both ubuntu-xenial and ubuntu-latest). But I don't see a great benefit to doing so. If no one is depending on this multi-label behavior, I'd like to remove it from the static driver. -Dave -- David Shrewsbury (Shrews)

4 6

Zuul project update 2018-05-07
by corvus＠inaugust.com 07 May '18

07 May '18

Hi, This past week has seen the following: Shrews began attempting to fix nodepool static driver to pre-register static nodes in ZooKeeper. This led to the discussion in [1] I pushed patches to add our new logo to the website and docs, as well as some other changes to the website suggested by the marketing team. The OpenDev conference and OpenStack summit are fast approaching. I have a lot to prepare before then -- both in terms of preparing for talks as well as polishing the website and documentation as much as possible, so I'm probably going to focus on those for the next two weeks rather than take up any significant new development work. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim [1] http://lists.zuul-ci.org/pipermail/zuul-discuss/2018-May/000197.html

1 0

Re: [openstack-dev] Overriding project-templates in Zuul
by Joshua Hesketh 02 May '18

02 May '18

> > I think in actuality, both operations would end up as intersections: > > ================ ======== ======= ======= > Matcher Template Project Result > ================ ======== ======= ======= > files AB BC B > irrelevant-files AB BC B > ================ ======== ======= ======= > > So with the "combine" method, it's always possible to further restrict > where the job runs, but never to expand it. Ignoring the 'files' above, in the example of 'irrelevant-files' haven't you just combined the results to expand when it runs? ie, A and C are /not/ excluded and therefore the job will run when there are changes to A or C? I would expect the table to be something like: ================ ======== ======= ======= Matcher Template Project Result ================ ======== ======= ======= files AB BC B irrelevant-files AB BC ABC ================ ======== ======= ======= > > So a job with "files: tests/" and "irrelevant-files: docs/" would do > > whatever it is that happens when you specify both. > > In this case, I'm pretty sure that would mean it reduces to just "files: > tests/", but I've never claimed to understand irrelevant-files and I > won't start now. > Yes, I think you are right that this would reduce to that. However, what about the use case of: files: tests/* irrelevant-files: tests/docs/* I could see a use case where both of those would be helpful. Yes you could describe that as one regex but to the end user the above may be expected to work. Unless we make the two options mutually exclusive I feel like this is a feature we should support. (That said, it's likely a separate feature/functionality than what is being described now). Anyway, I feel like Proposal #2 is more how I would expect the system to behave. I can see an argument for combining the results (and feel like you could evaulate that at the end after combining the branch-matching variants) to give something like: ================ ======== ======= ======= Matcher Template Project Result ================ ======== ======= ======= files AB BC ABC irrelevant-files AB BC ABC ================ ======== ======= ======= However, that gives the user no way to remove a previously listed option. Thus overwriting may be the better solution (ie proposal #2 as written) unless we want to explore the option of allowing a syntax that says "extend" or "overwrite". Yours in hoping that made sense, Josh

2 1

Re: [openstack-dev] Overriding project-templates in Zuul
by corvus＠inaugust.com 01 May '18

01 May '18

Joshua Hesketh <joshua.hesketh(a)gmail.com> writes: > I might be misunderstanding at which point a job is chosen to be ran and > therefore when it's too late to dissuade it. However, if possible, would it > make more sense for the project-local copy of a job to overwrite the > supplied files and irrelevant-files? This would allow a project to run a > job when it otherwise doesn't match. Imagine that a project with branches has a job added via a template. project-config/zuul.yaml@master: - job: name: my-job vars: {jobvar: true} - project-template: name: myjobs check: jobs: - my-job: vars: {templatevar: true} project/zuul.yaml@master: - project: templates: - myjobs check: jobs: - my-job: vars: {projectvar: true} project/zuul.yaml@stable: - project: templates: - myjobs check: jobs: - my-job: vars: {projectvar: true} The resulting project config is: - project: jobs: - my-job (branches: master; project-local job) - my-job (branches: master; project-template job) - my-job (branches: stable; project-local job) - my-job (branches: stable; project-template job) When Zuul decides what to run, it goes through each of those in order, evaluates their matchers, and pulls in parents and their variants for each that matches. So a change on the master branch would collect the following variants to apply: my-job (branch: master; project-local job) my-job (job) base (job) my-job (branch: master; project-template job) my-job (job) base (job) It would then apply them in this order: base (job) my-job (job) my-job (branch: master; project-template job) my-job (branch: master; project-local job) To further restrict a project-local job with a "files:" matcher would cause the project-local job not to match, but the project-template job will still match, so the job gets run. That's the situation we have today, which is what I meant by "it's too late to dissuade it". Regarding the suggestion to overwrite it, we would need to decide which of the possible variants to overwrite. Keep in mind that there are 3 independent matchers operating on all the variants (branches, files, irrelevant-files). Does a project-local job with a "files:" matcher overwrite all of the variants? Just the ones which match the same branch? That would probably be the most reasonable thing to do. In my mind, that effectively splits the matchers into two categories: branch matchers, and file matchers. And they would behave differently. Zuul could collect the variants as above, considering only the branch matchers. It could then apply all of the variants in the normal manner, treating files and irrelevant-files as normal attributes which can be overwritten. Then, once it has composed the job to run based on all the matching variants, it would only *then* evaluate the files matchers. If they don't match, then it would not run the job after all. I think that's a very reasonable way to solve the issue as well, and I believe it would match people's expectations. Ultimately, the outcome will be very similar to the proposal I made except that rather than being combined, the matchers will be overwritten. That means that if you want to expand the set of irrelevant-files for a job, you would have to copy the set from the parent. There's one other aspect to consider -- it's possible to create a job like this: - job: name: doc-job - jobs: name: doc-job files: docs/index.rst vars: {rebuild_index: true} Which means: there's a normal docs job with no variables, but if docs/index.rst is changed, set the rebuild_index variable to true. Either approach (combine vs overwrite) eliminates the ability to do this within a project or project-template stanza. But the "combine" approach still lets us do this at the job level. We could still support this in the overwrite approach, however, I think it might be simpler to work with if we eliminated it as well and just always treated files and irrelevant-files matchers as overwriteable attributes. It would no longer be possible to implement the above example, but I'm not sure it's that useful anyway? > What happens when something is in both files and irrelevant-files? If the > project-template is trying to say A is in 'files', but the project-local > says A is in 'irrelevant-files', should that overwrite it? I think my statement and table below was erroneous: >> This effectively causes the "files" and "irrelevant-files" attributes on >> all of the project-local job definitions matching a given branch to be >> combined. The combination of multiple files matchers behaves as a >> union, and irrelevant-files matchers as an intersection. >> >> ================ ======== ======= ======= >> Matcher Template Project Result >> ================ ======== ======= ======= >> files AB BC ABC >> irrelevant-files AB BC B >> ================ ======== ======= ======= I think in actuality, both operations would end up as intersections: ================ ======== ======= ======= Matcher Template Project Result ================ ======== ======= ======= files AB BC B irrelevant-files AB BC B ================ ======== ======= ======= So with the "combine" method, it's always possible to further restrict where the job runs, but never to expand it. As to your question about what if something appears in both -- in my "combine" proposal, I would continue to treat them independently, so if a project-local variant has an "irrelevant-files:" matcher, and a project-template variant has a "files:" matcher, then the usual nonsense would happen: they'd both have to match. So a job with "files: tests/" and "irrelevant-files: docs/" would never run because it's impossible to satisfy both. That actually matches some current behavior we have -- a job must match a "real" job definition as well as at least one project-template or project-local job definition, so a "job" with "files: tests/" and a project-template with "irrelevant-files: docs/" will behave the same way today. However, if we go with the "overwrite" proposal, since we're altering the behavior of the non-branch matchers anyway, that might be a fine time to pair them up and treat them as a unit. Then, whichever is the latest would win. So a job with "files" and a project-template with "irrelevant-files" would end up only having an irrelevant-files attribute. Then a further project-local stanza with only "files" would eliminate the irrelevant-files attribute leaving only the files value. Okay, let's summarize: Proposal 1: All project-template and project-local job variants matching the item's branch must also match the item. * Files and irrelevant-files on project-template and project stanzas are essentially combined in a set intersection. * It's possible to further reduce the scope of jobs, but not expand. * Files and irrelevant-files are still independent matchers, and if both are present, both must match. * It's not possible to alter a job attribute by adding a project-local variant with only a files matcher (it would cause the whole job to run or not run). But it's still possible to do that in the main job definition itself. Proposal 2: Files and irrelevant-files are treated as overwriteable attributes and evaluated after branch-matching variants are combined. * Files and irrelevant-files are overwritten, so the last value encountered when combining all the matching variants (looking only at branches) wins. * Files and irrelevant-files will be treated as a pair, so that if "irrelevant-files" appears, it will erase a previous "files" attribute. * It's possible to both reduce and expand the scope of jobs, but the user may need to manually copy values from a parent or other variant in order to do so. * It will no longer be possible to alter a job attribute by adding a variant with only a files matcher -- in all cases files and irrelevant-files are used solely to determine whether the job is run, not to determine whether to apply a variant. I think both would be good solutions to the problem. The key points for me are whether we want to keep the "alter a job attribute with variant with a files matcher" functionality (the "rebuild_index" example from above), and whether the additional control of overwriting the matchers (at the cost of redundancy in configuration) is preferable to combining the matchers. -Jim

1 1

Zuul Logo
by James Cole 30 Apr '18

30 Apr '18

Hello Zuul Enthusiasts, I’m James, a graphic designer with the OpenStack Foundation, and I’ve been working with Jim and the Zuul-core team to design an official Zuul logo. As you probably know, the goal has been to create something unique that can be used across a variety of mediums, such as an avatar on GitHub, on the website, as badge, and on other types of Zuul communications. After a few iterations, the team is leaning towards the concept featured in the attached PDF. The symbol in the logo was inspired by the geometry of the gate in Ghostbusters, since Zuul focuses on project gating and the name “Zuul” originated from the Zuul the Gatekeeper in the movie. More generally, the logo consists of a triangular, linework shape with the Zuul word mark placed underneath. The logo can be used a number of different ways and still maintain legibility at smaller sizes. The mockups in the attached PDF show a few different use cases, and offers some supporting typography and color information at the end. In case the email attachment did not make it to you, here is a link to view the document on Dropbox <https://www.dropbox.com/s/vv4yszyev1q9pbs/Zuul_Logo.pdf?dl=0>. We need to finalize something soon, since Zuul will be featured at the OpenStack Summit in May. So, we would love to hear what you think about this idea, if there is a detail we missed, or if you have any other feedback. Thank you! James Cole Graphic Designer OpenStack Foundation

10 22

Overriding project-templates
by corvus＠inaugust.com 30 Apr '18

30 Apr '18

Hi, We tried to make the Zuul v3 configuration language as intuitive as possible, and incorporated a lot that we learned from our years running Zuul v2. One thing that we didn't anticipate was how folks would end up wanting to use a job in both project-templates *and* local project stanzas. Essentially, we had assumed that if you wanted to control how a job was run, you would add it to a project stanza directly rather that use a project-template. It's easy to do so if you use one or the other. However, it turns out there are lots of good reasons to use both. For example, in a project-template we may want to establish a recommended way to run a job, or that a job should always be run with a set of related jobs. Yet a project may still want to indicate that job should only run on certain changes in that specific repo. To be very specific -- a very commonly expressed frustration is that a project can't specify a "files" or "irrelevant-files" matcher to override a job that appears in a project-template. Reconciling those is difficult, largely because once Zuul decides to run a job (for example, by a declaration in a project-template) it is impossible to dissuade it from running that job by adding any extra configuration to a project. We need to tread carefully when fixing this, because quite a number of related concepts could be affected. For instance, we need to preserve branch independence (a change to stop running a job in one branch shouldn't affect others). And we need to preserve the ability for job variants to layer on to each other (a project-local variant should still be able to alter a variant in a project-template). I propose that we remedy this by making a small change to how Zuul determines that a job should run: When a job appears multiple times on a project (for instance if it appears in a project-template and also on the project itself), all of the project-local variants which match the item's branch must also match the item in order for the job to run. In other words, if a job appears in a project-template used by a project and on the project, then both must match. This effectively causes the "files" and "irrelevant-files" attributes on all of the project-local job definitions matching a given branch to be combined. The combination of multiple files matchers behaves as a union, and irrelevant-files matchers as an intersection. ================ ======== ======= ======= Matcher Template Project Result ================ ======== ======= ======= files AB BC ABC irrelevant-files AB BC B ================ ======== ======= ======= I believe this will address the shortcoming identified above, but before we get too far in implementing it, I'd like to ask folks to take a moment and evaluate whether it will address the issues you've seen, or if you foresee any problems which I haven't anticipated. Thanks, Jim

1 0

Zuul project update 2018-04-30
by corvus＠inaugust.com 30 Apr '18

30 Apr '18

Hi, This past week has seen the following: We released Zuul 3.0.2. Note that we added a dependency on fb-re2, a python wrapper for the re2 library, which is now a binary dependency. The long-term plan is to start moving all of our regular expression parsing to re2, mostly because it is resistant to denial-of-service attacks. Clint has some patches in progress to do that. But in the mean time, any new uses of regular expressions should use re2 so that we don't create additional debt. We plan on issuing a Nodepool release this week with several accumulated bugfixes and improvements. Tobiash improved the merger so that it is more able to recover from an unclean shutdown. He also has a change [1] in review to add regular expression support to the project stanza so that you can, for example, specify that a job should run on all projects. Last week (and over the weekend) we merged my work to reduce memory consumption for configuration changes. It's now reasonable to consider configuration changes as not any more expensive than normal changes. The CPU time needed to calculate configuration changes is also reduced. This week I plan on returning to reviewing our outstanding specs. Tristan has been working on adding MQTT schema documentation. He also has been working on adding OpenAPI documentation but couldn't find a sphinx doc generator or how to embed swagger-ui in zuul-web. It sounds like we could use some investigation in to how to best handle documentation for the REST API. He also has a patch [2] to support secure diskimage env-vars in nodepool. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim [1] https://review.openstack.org/535713 [2] https://review.openstack.org/564687

1 0

Proposed updates to website to empower our non-coder friends
by Monty Taylor 26 Apr '18

26 Apr '18

Hey everybody! As part of the work with the Foundation staff on getting Zuul set up, we've chatted a bit about how to manage the website content. Obviously from the dev side, we find it important to be able to edit content like source code, propose those changes to gerrit and to review each other's changes. There are, however, folks out there who would like to help us do things like marketing who are not crazy about the concept of editing raw html. SO - we went and did a little bit of exploration to see what we could find. Actually, what really happened is that we agreed we'd just go with the html thing we have currently for now and circle back later to figure out a CMS + code review system that would work for everyone. But then I had a weekend and a personal project that had pretty much the same set of requirements. End result - what I'd like to propose is that we adopt a combination of using gatsby [1] as a static site generator and netlify-cms [2] as a web-based CMS interface for editing content and submitting changes to gerrit. Both gatsby and netlify-cms are completely client-side - there are no server side components to run. I found them by looking at https://headlesscms.org/ and https://www.staticgen.com/ gatsby ------ gatsby is a Javascript/React-based static site generator that allows managing content in markdown files. It has netlify-cms plugin support, and the build toolchain works very similarly to the new javascript toolchain for zuul-dashboard. I've got a first rough-draft of this part done that I'll push up for review today or tomorrow. netlify-cms ----------- netlify-cms is a Javascript/React-based CMS for static site generators. It has support for either pushing changes directly in to git or for submitting them as proposed changes. It currently basically only supports github - but the github interface is broken out into a driver layer. In fact, they have two different options- one that talks to github directly and one that talks to a netlify git-gateway service, so the plugin layer isn't just theoretical. We'll need to write a gerrit driver, but that shouldn't be terribly hard. Proposed Plan ------------- We can keep this pretty simple. Step one- migrate existing site from static html to gatsby + zuul jobs. Step two - write gerrit driver for netlify-cms. Step three - add netlify-cms to website. Why not use ... --------------- There are many options out there for static site generation. I like gatsby because it's pure-javascript rather than having a go, ruby or python component. I've personally come to really like keeping web things in javascript ... especially since our zuul-dashboard is doing javascript build tooling now too. There is also tight integration support with netlify-cms (which is also a good choice for the same reason) - and it was super easy for me to get up and going with one of their examples. [1] https://www.gatsbyjs.org/ [2] https://www.netlifycms.org/

3 3

Zuul project update 2018-04-23
by corvus＠inaugust.com 26 Apr '18

26 Apr '18

Hi, This past week has seen the following: Shrews started testing Ansible 2.5 and has related changes up for review[1]. I made a video[2] to upload to the website which explains the gating algorithm. I expect it to be merged soon. I've also resumed work on the configuration loading refactor which will allow us to use less memory for dynamic configuration changes. I plan on continuing that this week. Tobiash has been improving scalability with parent-change-enqueued event and many github app installations[3]. Part or that stack improves github testing. Next week, he plans on working on weak job dependencies (jobs that still run even if their parents are skipped -- so that Zuul can decide that an image build job does or doesn't need to run), along with finishing up GitHub branch status protection handling. Clarkb has (with Shrews) made zuul-web more asynchronous. He points out that we've merged a number of bugfixes, including updates to zuul-changes and encrypt_secret. We've alo removed the zuul-cloner command. He also asks whether it's time to make another release. TristanC is working on `allowed-triggers` and `allowed-reporters` tenant settings to better support a multi-tenant environment with a multitude of connections. The nodepool driver work mentioned last week has merged. Another change[4] is proposed to make the OpenStack NodeLauncher generic. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim [1] https://review.openstack.org/#/q/topic:zuul-ansible-2.5 [2] https://review.openstack.org/562005 [3] https://review.openstack.org/#/q/topic:parent-change-enqueued+(status:open+… [4] https://review.openstack.org/535555

2 1

Re: Zuul + OpenStack Foundation update
by Lauren Sell 24 Apr '18

24 Apr '18

Hi everyone, Thanks very much Jim for providing the update on our plans heading into the Vancouver Summit. I want to sincerely apologize for the miscommunication around the board meeting. We presented an update on making Zuul an independent project at the Gold Member meeting yesterday and again at the Board Meeting today, but unfortunately the entire presentation today was scheduled during the executive session and not able to be heard by the public. It was a definitely oversight and miscommunication. I want to share the deck we presented here and am happy to do a call for those who want to run through it, or simply chat about it in IRC. We welcome any feedback on the plans: https://docs.google.com/presentation/d/1iR_Q0ntV5BAs1TNjN-b1JBnsfahb2hI4IxM… The primary feedback we received during the gold member meeting was trying to understand where Zuul would fit under the OpenStack Foundation (ie. is it still part of OpenStack, or an adjacent project with independent governance?) and where Zuul fits in the CI/CD landscape (ie. what’s competitive and complementary). For people who are newer to the space, there has been a lot more activity recently around XCI and the OpenCI initiative and of course existing tools like Jenkins and new ones like Spinnaker, and they need help parsing what’s an effort by a group of people versus what’s actually software and when you would use which software :) Hope that is useful, and again, apologies for the miscommunication. I’m now subscribed to the mailing list and am happy to discuss here :) Thanks, Lauren PS If you can’t view the deck in google docs, I’m happy to send over a PDF. I just didn’t want to send an attachment to the mailing list... >> Begin forwarded message: >> >> From: corvus(a)inaugust.com <mailto:corvus@inaugust.com> (James E. Blair) >> Subject: Zuul + OpenStack Foundation update >> Date: April 11, 2018 at 5:28:08 PM CDT >> To: Zuul Discussion List <zuul-discuss(a)lists.zuul-ci.org <mailto:zuul-discuss@lists.zuul-ci.org>> >> >> Hi, >> >> Several of us on the zuul-core team have been working with the OpenStack >> Foundation staff on establishing and launching Zuul as a top-level >> project under the Foundation's umbrella. >> >> As a free/open-source project, and one that follows OpenStack's "four >> opens" we generally strive to use public channels for collaboration. >> However, when making a change like this, that is not always practical, >> especially in the early stages. I think we'll get there soon, but in >> the mean time, I ask you to accept Monty, Clark, Jeremy and I as >> delegates of the project to work with the Foundation on this. >> >> I'll try to keep everyone updated as best as possible with what we're >> working on, though sometimes those updates may need to be vague. Please >> feel free to contact me or the others privately if you have specific >> concerns you want to ensure we address. Here are updates from our most >> recent discussions: >> >> * There will be discussion about Zuul at the OpenStack board of >> directors meeting on April 24. That will be a phone call to which >> anyone is welcome to listen. >> >> * The Foundation will promote Zuul at the upcoming OpenStack Summit in >> Vancouver. It will be officially announced as a Foundation project on >> the keynote stage and via press release. >> >> * We are working with the Foundation's design team on a logo. We want >> to have something that can be adapted to work as an avatar (i.e., on >> GitHub), on the website, as a badge, and on slide decks. >> >> * The Foundation marketing team has begun a review of the website >> content and will be assisting with changes. Included in this is an >> effort to collect user stories for Zuul. >> >> * We're looking into implementing a system to manage the website content >> which is both Gerrit-friendly and has a workflow suitable for those >> unaccustomed to using Git. Monty has volunteered to research and >> report back with some options to explore. >> >> -Jim >> >> _______________________________________________ >> Zuul-discuss mailing list >> Zuul-discuss(a)lists.zuul-ci.org <mailto:Zuul-discuss@lists.zuul-ci.org> >> http://lists.zuul-ci.org/cgi-bin/mailman/listinfo/zuul-discuss >

1 0

Simplified DSL
by Tristan Cacqueray 21 Apr '18

21 Apr '18

Hello folks, Some users may find job written in Ansible hard to reproduce, and I was wondering if we could add a new "exec" job attribute that would simply be shell task to be executed on a single node. This would let us write zuul.yaml file more similar to other in-repo ci system. For example a zuul.yaml could be: - job: name: my-job exec: - mvn build - mvn test - project: check: jobs: - my-job gate: jobs: - my-job Moreover we could go a step further and streamline that full description in a single stanza: - ci-job: - mvn build - mvn test Which could be translated into a unique job name attached to the check and gate pipeline by default. What do you think? -Tristan

8 9

Question regarding using etcd instead of Zookeeper
by Tristan Cacqueray 21 Apr '18

21 Apr '18

Hello folks, From a Kubernetes operator point of view, the Zookeeper component sounds redundant with the already running etcd service. Moreover etcd seems more active and it shows better overall performance. So my question is why Zookeeper was picked in the first place, and would it be possible to support/migrate to etcd in the future? Thanks in advance, -Tristan

9 11

Reverse of required-projects.override-checkout?
by Mike Pagel 20 Apr '18

20 Apr '18

Hi all, I am using zuul now for a while at work and I am struggling to properly configure the following use-case. Hope I don't miss the obvious. I have two repositories A and B. A is an internal, B is an external project from another department. We define our jobs in A's zuul.yaml. B is using different branch names (e.g., because they deliver to other teams as well). A/zuul.yaml (master branch): # build the "master-a" branch of B for A's "master": job: required-projects: name: B override-checkout: master-a Similarly, when we have a maintenance branch "v1.0" in A, B might call it "v1.0-a": A/zuul.yaml (v1.0 branch): # build the "v1.0-a" branch of B for A's "v1.0": job: required-projects: name: B override-checkout: v1.0-a Now, without duplication of these job definitions, how would I run the correct job variant when checking a change in B, let's say in the "v1.0-a" branch, where I cannot fall back to "master" as default branch? AnsibleJob.resolveBranch() has some logic, but I think that's still the forward reference. Could I use job.branches for this, as in: A/zuul.yaml (v1.0 branch): # build this "v1.0" branch in A if a change occurred in (any) "v1.0-a" branch: job: branches: v1.0-a required-projects: name: B override-checkout: v1.0-a I see code in and calls to model.Job.changeMatches() that could suggest this behavior. Or would it try to als check out A's "v1.0-a" branch, which is non-existent? Any hints are greatly appreciated. Mike

2 3

GitHub OAuth application
by Tristan Cacqueray 19 Apr '18

19 Apr '18

Hello folks, Would it be possible to adapt the GitHub driver to use user's access instead of the Zuul unique app key/api token? Ideally, users would register on the scheduler and get redirected to grant access to their GitHub identity through an OAuth app. Then zuul could manage the project's hook. That would be the easiest workflow from an user point of view. Though this would require some sort of user management. Another solution would be to create the access_token out of bound and extend the driver configuration to support using multiple tokens to be used per projects or oganizations. If this can be implemented without a major rewrite of the driver, that sounds like something worthy as it would enable new use cases and it may mitigate rate limiting issue for large deployment. Regards, -Tristan

2 2

Zuul project update 2018-04-16
by corvus＠inaugust.com 16 Apr '18

16 Apr '18

Hi, Last week, we agreed to suspend weekly IRC meetings and instead try to conduct more of our business over email, or in ad-hoc IRC meetings as needed. In an attempt to keep the cadence provided by the meetings, we're going to send out a weekly email update. Hopefully this will help us all keep up with what's going on in the project. Please feel free to reply to this email with further discussion. Last week, we released Zuul 3.0.1. Shrews fixed a common bug seen in long streaming involving multi-byte characters across read boundaries, and is continuing work on making zuul-web more async. Also in progress: improvements to the zuul-from-scratch document. [1] I posted a spec about container support [2]. It's more of a starting point for discussion than a complete plan at this point. It's collected a number of comments so far, and will certainly be revised several times before it is finalized. This week, I plan to work on resuming the work I started before the 3.0.0 release to reduce the memory used by dynamic configuration changes. Mordred has switched us to using nodeenv for javascript dependencies which makes it easier to run the tests locally and means we can use the standard tox-py35 job. He also upgraded webpack to version 4. This week has us moving the dashboard from angularjs v2 to angular.io v5. [3] Tristan has updated the MQTT driver which should be ready now and updated several jobs in zuul-jobs (ansible-lint, review, spec and upload-to-galaxy). In nodepool, the run_handler refactor (which is intended to make adding more drivers easier) is ready for review. He also updated zuul-changes.py to support the new zuul-web api routes. Tobiash fixed a number of bugs, and has patches up to refactor the executor's governor into a sensor API for easier extension. He added support for postgres (we run tests on PG and mysql now), and regex matching of github status. This week he plans on trying out Ansible 2.5, improving scalability of parent-change-enqueued event across many github app installations, and adding tests to status protection handling. Please help write next week's email by updating this etherpad: https://etherpad.openstack.org/p/zuul-update-email Thanks, Jim [1] https://review.openstack.org/#/q/status:open+project:openstack-infra/zuul+b… [2] https://review.openstack.org/560136 [3] https://review.openstack.org/#/q/status:open+project:openstack-infra/zuul+b…

1 0

Next Zuul release (3.0.1?)
by corvus＠inaugust.com 11 Apr '18

11 Apr '18

Hi, We have some mildly-urgent bugfixes merged or ready to merge in Zuul, so I'd like to make a release sometime in the next few days. By mildly-urgent, I mean that the sooner we release the fixes, the less problems people will have in the future (either by creating jobs relying on wrong behavior, or system configurations which are problematic). The items I have on the list are: * Roles checkout fix (merged) * Github3.py version bump (ready to merge) * Postgress fixes/support (merging as I write this) We've also merged some other bugfixes, and may slip some others in during the remaining time. Are there any other items of similar urgency to the three above that we need to release ASAP? Once we're satisfied (probably tomorrow?) I'll release 3.0.1. I think that since we're only focusing on bugfixes, the micro version bump is appropriate. I don't think Nodepool needs a corresponding release. -Jim

4 3

Zuul + OpenStack Foundation update
by corvus＠inaugust.com 11 Apr '18

11 Apr '18

Hi, Several of us on the zuul-core team have been working with the OpenStack Foundation staff on establishing and launching Zuul as a top-level project under the Foundation's umbrella. As a free/open-source project, and one that follows OpenStack's "four opens" we generally strive to use public channels for collaboration. However, when making a change like this, that is not always practical, especially in the early stages. I think we'll get there soon, but in the mean time, I ask you to accept Monty, Clark, Jeremy and I as delegates of the project to work with the Foundation on this. I'll try to keep everyone updated as best as possible with what we're working on, though sometimes those updates may need to be vague. Please feel free to contact me or the others privately if you have specific concerns you want to ensure we address. Here are updates from our most recent discussions: * There will be discussion about Zuul at the OpenStack board of directors meeting on April 24. That will be a phone call to which anyone is welcome to listen. * The Foundation will promote Zuul at the upcoming OpenStack Summit in Vancouver. It will be officially announced as a Foundation project on the keynote stage and via press release. * We are working with the Foundation's design team on a logo. We want to have something that can be adapted to work as an avatar (i.e., on GitHub), on the website, as a badge, and on slide decks. * The Foundation marketing team has begun a review of the website content and will be assisting with changes. Included in this is an effort to collect user stories for Zuul. * We're looking into implementing a system to manage the website content which is both Gerrit-friendly and has a workflow suitable for those unaccustomed to using Git. Monty has volunteered to research and report back with some options to explore. -Jim

1 0

Deprecation notice: Remove support for extensions as lists in zuul-jobs/roles/stage-output - change https://review.openstack.org/540890
by Andreas Jaeger 11 Apr '18

11 Apr '18

Andrea Frittoli has updated roles/stage-output in zuul-jobs to handle dicts instead of lists for extensions - and updated all usage in OpenStack itself. His review https://review.openstack.org/540890 removes now handling of lists. Please check that none of your playbooks and jobs uses extensions with stage-output and update them. I put a WIP on the change and will remove it in two weeks so that we can merge, please update your files as needed and feel free to review the change, Andreas -- Andreas Jaeger aj(a){suse.com,opensuse.org} Twitter: jaegerandi SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) GPG fingerprint = 93A3 365E CE47 B889 DF7F FED1 389A 563C C272 A126

1 0

Spec for using containers as Zuul build resources
by corvus＠inaugust.com 11 Apr '18

11 Apr '18

Hi, We've all discussed using containers in Zuul for some time, but it's been difficult to get on the same page. Partly because of differing use cases and assumptions, and partly because Zuul brings some unique things to the table which need to be integrated into the container mindset. I've uploaded the first draft of a spec which I hope will address this and set a path for us to successfully integrate containers: https://review.openstack.org/560136 It's long on exposition and short on work items at this point. I think first it's useful for us to look at it from a distance and see if it handles the use cases we think we need to cover, and if the general approaches to workflow will work. Then, we should focus on the details, and expand it to ensure that we describe the actual processes we're going to implement. And finally, it should grow a list of work items so we can get things done. If you haven't reviewed a spec like this before, know this: it's a very iterative process. Please leave detailed inline review comments in Gerrit, where we may engage in a back-and-forth discussion for a while. If it gets awkward to have that discussion in Gerrit, we can move it to the mailing list, but in general, it's preferable to keep the discussion in Gerrit where the context and history are all together. After some discussion, I will update the spec with suggested changes and we'll go through the process again. The goal is to achieve consensus during the review process and merge a polished result, not to merge the change ASAP. This also establishes a specs section of the Zuul developer documentation. We had a brief discussion in an earlier IRC meeting about how to handle specs now that openstack-infra/infra-specs is no longer appropriate. I think putting them directly in the Zuul repo is worth trying. The main danger there is accumulating too much documentation which does not reflect the current state of the project, but this includes big red warnings, so maybe that will help. It also looks like a significant amount of this work may end up happening in the nodepool repo as opposed to zuul, but as long as we don't have a dedicated place for cross-project specs, I think the zuul repo is a reasonable place to start. -Jim

2 1

Zuul meeting time
by Tobias.Henkel＠bmw.de 09 Apr '18

09 Apr '18

Hi zuul community, Currently the zuul meeting is at 22UTC which is currently 23:00 localtime or soon 0:00 within summer time at my location. That makes it impossible for me to participate regularly. However I’d really like to attend the zuul meeting. So I’d like to ask zuul folks about your opinion about shifting it around 20ish UTC. I wouldn’t have any strong opinion about the day. Thanks and kind regards Tobias

9 15