Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Jesse Glick
    @jglick
    For JFR use cases, it makes little sense to use an actual ArtifactManagerFactory implementation plugin—these are only useful for Jenkins servers, especially with agents. For example, stash and unstash would gratuitously use cloud storage! Would be more sensible to have the function packaging simply upload files from $JENKINS_HOME/jobs/job/1/artifacts/ after JFR exits.
    Oleg Nenashev
    @oleg-nenashev
    It still makes sense if you want to archive() things to a permanent storage without updating Pipeline definitions
    Jesse Glick
    @jglick
    @oleg-nenashev no, my point is that if the function packaging uploads things from the stock Jenkins artifacts directory then the same thing happens, without all the unused machinery of an actual JEP-202 implementation.
    What is up with the [$LATEST] prefix here?
    Oleg Nenashev
    @oleg-nenashev
    Yes, many things can be implemented in a different way when you do not need to read the artifacts
    JEP-202 was created for Jenkins, not for the "Cloud Native Jenkins" vision
    Jie Shen
    @gavinfish
    @timja yes, we are planning to work on this part.
    Jesse Glick
    @jglick
    @tomaszsek does https://github.com/odavid/my-bloody-jenkins overlap with JEP-219?
    Jesse Glick
    @jglick

    @timja to clarify https://github.com/jenkinsci/junit-plugin/pull/110#issuecomment-477900044

    we're running our tests in Kubernetes and wanting to report them back to jenkins, not have jenkins collect them

    Does this mean you want your Jenkins-initiated test runner (whatever that is) to upload results directly to some kind of database, then have the Jenkins web UI be able to display those results in its usual way later? If so, this is theoretically possible (not a use case I had considered), though you would still need the Jenkins job to run a dummy junit step so as to add a TestResultAction marker to build.xml (all details would be loaded from the external provider).

    Tim Jacomb
    @timja
    (kubernetes job)
    Yes that would be fine
    Nguyen Le Vu Long
    @longngn
    Hi everyone. I'm Long Nguyen, an aspiring student of Jenkins in Google Summer of Code 2019. Currently I'm making a proposal about integrating Remoting over Kafka plugin with K8s environment. More specifically, the plugin would be able to start Kafka pods and Jenkins agent pods on K8s. This proposal might intersect with some parts of Cloud Native SIG (K8s plugin, Helm chart,...) so I'd like to introduce myself here and see if there'd be any idea or suggestion. Any help is greatly appreciated. Btw here is my proposal, you can comment directly on it: https://docs.google.com/document/d/1Iu0pJLkr_PHQIpMA2BDtQQ-fU4yDVBPe_5OwAbv2y-g/edit?usp=sharing
    (CC @oleg-nenashev @pvtuan10 )
    sarathchandrabhaskara
    @sarathchandrabhaskara
    hi, can someone help me how to run ansible playbooks on Jenkins installed windows 10?
    mathieucoavoux
    @mathieucoavoux
    Hello, @jglick I was reading your last changes to the JEP-207 and notice the FlowNode.id, I am interested to see the code implementation but I am a bit lost with the PR and what have been merged to which branch. Can you help me ? Thanks a lot
    Jesse Glick
    @jglick
    @mathieucoavoux JEP-210 you mean. Everything is long since merged (though “watch mode” for sh etc. remains disabled by default). You can run the reference implementation in pipeline-cloudwatch-logs as an example. Which part are you interested in?
    mathieucoavoux
    @mathieucoavoux
    @jglick I was completely wrong, my first understanding was the FlowNode.id was something new in the Jenkins core while it's FlowNode part of the workflow plugin. ashamed Thanks
    Jesse Glick
    @jglick
    No problem! Let me know if I can point you to anything specific.
    mathieucoavoux
    @mathieucoavoux
    Thanks a lot @jglick . I see that ConsoleNotes is quite easy to use in CloudWatchSender is it planned to add this class into either Jenkins core or workflow plugin to be reused by others?
    Jesse Glick
    @jglick
    @mathieucoavoux possibly to some external logging support plugin as in JEP-212. Work on JEP-210/207/212 was suspended. Do you have some plan to produce another implementation?
    mathieucoavoux
    @mathieucoavoux
    I will copy your and rework on the plugin (based on ES) I've started a while ago to use LogFactory. Indeed I could probably use cloudwatch instead but I am also needing jep-210/207/212. I dont know if I can help somehow for those JEPs? My implementation was quick-and-dirty as I needed to serialize builds in Json and tried to made it compatible with freestyle and pipeline jobs
    Jesse Glick
    @jglick
    @oleg-nenashev had prototyped some work for JEP-212 on https://github.com/oleg-nenashev/external-logging-logstash-plugin but I am not sure of the status. It is not, as written, compatible with JEP-210.
    mathieucoavoux
    @mathieucoavoux
    Ok, yes I remember this from @oleg-nenashev but for my understanding at that time the purpose of this plugin was to send logs to ES but didn't cover the JEP-210 aspect. I understand, for the sake of clarify that it's better to dissociate this logging plugin with another build-storage plugin. Aside from this I think that at the time @oleg-nenashev wrote the plugin the LogFactory wasn't yet implemented and he prefered to use the extlogging plugin but I may be wrong.
    Jesse Glick
    @jglick
    Well, it is complicated. We were working on the same overall design, but from different physical APIs. JEP-210, purely implemented in Pipeline, shipped; 207/212, with hooks in core to also support freestyle projects etc., did not. And then all work on this area was suspended indefinitely. Can I ask again what your goals are here? You said something about “serializing builds in JSON” but I have no idea what that is about.
    mathieucoavoux
    @mathieucoavoux
    Yes, so what I wanted initially was to store all builds + logs in an central storage to be able to put multiple Jenkins masters behind a proxy to have a HA platform. I made a video available here just to see the context. Sorry it's a bit long+ a robot voice for comments. So I understand that efforts made for the logging are also usable in the Jenkins X context and that's very great. But I was trying to see how I can have a custom plugin for a while which is not that good compare to what you have developped but at least align with what has been adopted along the way and when JEP-210 is implemented give up this plugin. Of course if I can help somehow I will.
    Jesse Glick
    @jglick
    JEP-210 and related efforts are not used in Jenkins X.
    Or at least, not in the recommended serverless mode (i.e., --prow). Containers using jenkinsfile-runner or Tekton are run and pod logs are optionally copied to permanent storage (as per jx get storage) when the build completes. Very different from JEP-207/210/212, which deals with regular Jenkins servers that should upload logs of potentially in-progress builds from agents to external storage and (as needed) retrieve from storage to the master.
    Jesse Glick
    @jglick
    Anyway, 210 is usable today (so long as you care only about Pipeline projects), though master-based log display has some structural inefficiencies since the browsing API in Jenkins core (used by both the traditional console and Blue Ocean) is oriented towards byte offsets rather than the opaque cursors you would use in a typical cloud system.
    mathieucoavoux
    @mathieucoavoux
    Sorry I again confused. What I have put in my plugin was the Runs (https://jenkins.io/sigs/cloud-native/pluggable-storage/ ) in addition to external logging. And that the one I had issue with the serialization
    mathieucoavoux
    @mathieucoavoux
    Do you think that the configuration storage and run storage shall/must be different than the logs storage in the future ? For example logs are sent to cloudwatch but runs status are kept in an SQL database?
    Jesse Glick
    @jglick
    @mathieucoavoux yes that would be my expectation. Build metadata is structured and Jenkins needs it in indexed form sometimes; whereas logs are textual, streamed, full-text-searchable, etc.
    mathieucoavoux
    @mathieucoavoux
    Ok thanks :) @jglick
    Oleg Nenashev
    @oleg-nenashev
    @mathieucoavoux @jglick Yes, separate storages would be my expectation as well. or configurations and runs there is JEP-213 (https://github.com/jenkinsci/jep/tree/master/jep/213) which could do the job. But it is stalled as well
    mathieucoavoux
    @mathieucoavoux
    Thank you @oleg-nenashev I've just been through the JEP-213 and JENKINS-41196, I will take sometimes to see what this custom-war-packager is. I will read comments in the jenkinsci/jenkins#2754 accordingly. Thanks
    Oleg Nenashev
    @oleg-nenashev
    @mathieucoavoux yw. Enjoy the exploration
    370672701
    @370672701
    Hello, I'm new to Jenkins,
    Is any way can configure active-standby servers for jenkins, to keep it HA
    thanks advance
    Oleg Nenashev
    @oleg-nenashev
    @370672701 short answer is "no". I have done and presented a lame PoC at some point, but it has never been productized
    If you want scalability && your projects follow GitOps && can be run in Kubernetes, then it makes sense to consider Jenkins X Serverless mode
    Oleg Nenashev
    @oleg-nenashev
    @370672701 you may be also interested in https://go.cloudbees.com/docs/cloudbees-documentation/cje-install-guide/#high-availability . It somewhat implements active-standby
    370672701
    @370672701
    got it, thanks a lot @oleg-nenashev
    Philipp Garbe
    @pgarbe
    Anyone interested in taking over the maintainer for the ecs plugin from me?
    Martin d'Anjou
    @martinda
    @pgarbe are you talking about https://github.com/jenkinsci/amazon-ecs-plugin ?
    We have a GSoC project on this topic. It was meant to add ECS to https://github.com/jenkinsci/external-workspace-manager-plugin but we are coming to the realization that ECS should be its own thing.
    If you have any ideas with regards to the direction ECS should take, please share it with us.
    Here is what the student recently wrote about it: https://groups.google.com/forum/#!topic/jenkinsci-gsoc-all-public/oRHraEPb_cA
    @/all We have a GSoC project on ECS. What are your thoughts on ECS in Jenkins? See this post from the student.
    Andreas Sieferlinger
    @webratz
    The post talks about EFS? Which is unrelated. So whats the goal ECS or EFS?
    But i agree that EFS / shared workspaces could be a interesting topic to be used together with the ECS plugin, especially if you are using multiple containers in a pipeline which each usually start off with a completely fresh filesystem
    Nguyen Le Vu Long
    @longngn
    Hi everyone, nice to meet you. Currently I'm implementing Jenkins Cloud. How can I add a textbox field value to Jenkins' Labels (so that new job could use that label to run on Cloud) ?
    Oleg Nenashev
    @oleg-nenashev
    jenkins-infra/jenkins.io#2387 - blog post for Azure artifact manager
    tahaa karim
    @tahaalibra
    Hi, I am using docker plugin and have a few docker cloud agents, what I want to do now is to be able to choose a label eg. node 'my-label', and also specify slave image to be used. is there any way to achieve this?