Subject: Pipeline Storage Performance Work Available For Beta - Want to kick the tires?
Hey all, I've just released a set of plugin betas to the Experimental Update Center. They have enhancements to Pipeline which CAN dramatically reduce I/O use and improve performance. Please given them a try and report back how they work out for you. Please note: to maintain existing behavior, these changes are OPT-IN. You MUST enable them to see a difference (see below). The settings themselves have their own explanations (tooltips and help info), but the below gives more info. *Will it help me?* * Yes, if you are running complex Pipelines or Pipelines with many steps. * Yes, if your Jenkins instance uses NFS, magnetic storage, runs many Pipelines at once, or shows high iowait. * No, if your Pipelines spend almost all their time waiting for shell/batch scripts to run. This isn't a magic "go fast" button for everything (I wish!). * No, if you are not using Pipelines, or your system is loaded down by other factors. *How do I get it?* * You need to be on Jenkins LTS 2.73+ or higher (or a weekly 2.62+) * Enable the experimental update center - instructions here: https://jenkins.io/blog/2013/09/23/experimental-plugins-update-center/ * Check for plugins updates * You should see and install updates for the following plugins, with versions including the word "durability" - Pipeline: API (workflow-api) - Pipeline: Groovy (workflow-cps) - Pipeline: Job (workflow-job) - Pipeline: Supporting APIs (workflow-support) - Pipeline: Multibranch (workflow-multibranch) * Restart the master to use the updated plugins - note: you need all of them to take advantage. *What does it do?* This adds a performance/durability setting for Pipelines. If you use the performance-optimized mode, disk writes are reduced significantly. This lets you improve Pipeline performance greatly (reduce I/O) at some cost to the running Pipelines' ability to survive if Jenkins falls over completely (durability). Stability of Jenkins ITSELF is not changed, nor are there changes to completed Pipelines. We also add the ability to mark Pipelines to NOT resume upon restart (a requested feature) - available under the properties at the top. *How do I USE it?* Durability settings need to be enabled (and will display in the logs when a job begins), either globally or per Pipeline/branch (MultiBranch). Settings take effect the next time the Pipeline runs. There are 3 ways to configure the durability setting: **Globally**, you can choose a durability setting under "Manage Jenkins > Configure System", labelled "Pipeline Speed/Durability Settings". These settings will take effect for Pipelines upon the next run, unless you override them with one of the below settings **Per pipeline job:** at the top of the job configuration, labelled "Custom Pipeline Speed/Durability Level" - this overrides the global setting. Or, use a "properties" step - the setting will apply to the NEXT run after the step is executed (same result). **Per branch for a multibranch project:** configure a custom Branch Property Strategy (under the SCM) and add a property for Custom Pipeline Speed/Durability Level. This overrides the global setting. *What are the settings?* * Performance optimized mode ("PERFORMANCE_OPTIMIZED") - Greatly reduces disk I/O but running Pipelines with lower durability settings may lose runtime data IF they do not finish AND Jenkins is not shut down gracefully. If this happens, they behave like FreeStyle builds (logs, but no steps to visualize). Details at bottom. * Maximum durability ("MAX_SURVIVABILITY") - behaves just like Pipeline did before, slowest option. Use this for running your most critical Pipelines. * Less durable, a bit faster ("SURVIVABLE_NONATOMIC") - Writes data with every step but avoids atomic writes. On some filesytems, especially networked ones (i.e. NFS), this is faster than maximum durability mode, but it carries a small extra risk (details at bottom). *Nitty-gritty details* Remember: worst-case behavior reverts to something like FreeStyle builds -- Pipelines that cannot persist data may not be able to resume or displayed in Blue Ocean/Stage View/etc, but will have logs. Running pipelines with the performance-optimized setting may lose data IF they do not finish AND Jenkins is not shut down gracefully. A "graceful" shutdown is where Jenkins goes through a full shutdown process, such as visiting http://[jenkins-server]/exit or using one of the gentler signals to kill the process. A "dirty" shutdown is where the Jenkins process dies without doing shutdown tasks -- killing a Docker container or using "kill -9" to terminate the Java process will do this. The less-durable/a bit faster setting avoids atomic writes -- what this means is that if the Operating System fails, data that is buffered for writing to disk will not be flushed and will be lost. This is quite rare, but can happen as a result of Docker or virtualization operations that halt the operating system or disconnect storage. Or, y'know, if someone pulls the plug. *Warnings* This has extensive unit testing, plus initial code review and some manual testing BUT it is still a beta, so there's some risk. Take appropriate precautions - back up JENKINS_HOME, don't deploy to a critical Jenkins instance, etc. Think of it as an early Christmas gift, and happy holidays to all! -- You received this message because you are subscribed to the Google Groups "Jenkins Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-dev/17308e33-1387-4174-a328-7976aada79a7%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
