trying to solve some problem with Mesos. I have three build servers for Jenkins. Jenkins schedules jobs on them through Mesos.
For now, Mesos loads one agent(slave) as hard as possible, but I want it to spread jobs across all agents..
As I see, it's better to run three jobs on three agents, than on one.
Is it possible to randomise job scheduling?
Or perhaps, I have such scenario. 2 large servers and one mini. I want to schedule Jobs on mini by default, and if it's not enough resources, then proceed to large servers. How can I achieve this goal? Is it possible to set priority for agents(slaves) to specify on which agent I want job to run at first?
The Mesos plugin for Jenkins attempts to build on the most recently built slave (see this method). This means that once it builds on that machine once, as long as that machine still has available spare resources - it'll schedule additional jobs on that machine until it is full. Right now it looks like that isn't optional (I have filed it as a feature request).
Related
I have a scenario where I have N numbers of nodes and N number of tests, and these tests will be distributed to the nodes. My nodes have a label Windows.
Here's an example:
I have a pipeline job that will manage the distribution of the tests to the nodes. I set my pipeline job to run 10 tests on 10 VMs having the label Windows and it will run smoothly. However, one of my requirements is to concurrently run that pipeline job. The problem I might encounter If I have 10 tests on VMs 1-10 in the first run of the job, and run another job for 5 tests for VMs 11-15, given that I am using the Windows label, there might be a possibility that Jenkins will assign the test to VMs 1-10 but should run on VMs 11-15 or vice versa.
The solution I came up with is to dynamically change the label of the VMs from one of the jobs to a unique label that will only be used for that Job. Unfortunately, I still don't know how to do that.
Basically, I just want to logically group my nodes via label on demand in my pipeline script.
I searched all throughout the internet and yet I still wasn't able to find a solution that fits my needs.
Kindly help me with this. Also, I am open to using a different approach.
I'm trying to build a Jenkins declarative pipeline that will build on all agents in parallel.
How can I do this without disabling sandbox?
I have come across this page: https://jenkins.io/blog/2017/09/25/declarative-1/ but it seems repetitive, especially when padded out with my code as nearly all operations are performed almost the same on every node. Is there a way to do this and avoid repeating code?
I suggest that you follow the common pattern described in the referenced article.
By assigning labels identifying the node's operating system and allocating nodes based on these labels, you ensure that the job runs exactly once in each of the different build environments.
A severe drawback of your suggestion to build on all of the available agents (as said, I don't know anything how to actually do that)) would be in the case of one or multiple build agents being offline. So you don't run on Windows, because the server was just rebooting, but your build result is green as nothing failed? Not a good idea, isn't it?
Another benefit of the label-based approach is that you can easily add additional build agents to cope with increased number of builds, e.g., as your team grows. You don't want to build twice on Windows, when you add another build agent with Windows, right?
So I strongly recommend: Assign labels to your build agents and then specify, on which agents your job needs to run.
Is there any difference between I create two slaves, or one slave with two executors on the same Windows server?
Yes, there is a difference: It's about memory consumption and effort of maintenance/administration.
Starting a slave on a system starts a (main) process. This process costs (private) main memory to run and connects to the master.
Each executor is a sub-process of the main process.
It is therefore apparent that running two executors on one slave costs less memory in total compared to running two slaves (with one executor each), as there would be the memory consumption of the main process twice:
2 * Main Processes + 2 * Executors > 1 * Main Process + 2 * Executors
Moreover, administrating a slave is some more effort than just an executor: Whilst an executor has virtually nothing to worry, there are numerous things to configure for a slave. Additionally, the capabilities of the two slaves are anyhow the same (they are running on the same OS as you said), so there is little value-add to also assign it different labels.
In short, if there are no other boundary conditions, which make me do it differently, I always would prefer running two executors on one slave, as this is easier to administrate and some memory is saved.
A slave is a "machine". An executor is an "OS Process" in the slave.
So ideally we always add executors - they do the work and can run in parallel, and the simple theoretic answer to your question is "2 executors on one slave"
In practice we need to add slaves in several use cases:
We need more resources (more cpu, more memory, more "machines")
We need a different setting (Different OSes, Different hardware)
We have global resources that would create a conflict for executors on same machine (shared browser for a UI testing process)
Make the decision based on your use case.
One benefit which immediately comes to my mind for running 1 executor on given node, is to prevent conflicts between processes run at the same time.
On other hand you could prevent job conflicts using existing Jenkins plugins, ie. Heavy Job, Build Blocker.
We need to be able to run a Jenkins job that consumes two slaves. (Or, two jobs, if we can guarantee that they run at the same time, and it's possible for at least one to know what the other is.) The situation is that we have a heavy weight application that we need to run tests against. The tests run on one machine, the application runs on another. It's not practical to have them on the same host.
Right now, we have a Jenkins job that uses a script to kick a dedicated application server up, install the correct version, the correct data, and then run the tests against it. That means that I can't use the dedicated application server to run other tasks, when there aren't the heavy weight testing going on. It also pretty much limits us to one loop. Being able to assign the app server dynamically would allow more of them.
There's clearly no way to do this in the core jenkins, but I'm hoping there's some plugin or hackery to make this possible. The current test build is a maven 2 job, but that's configurable, if we have to wrap it in something else. It's kicked off by the successful completion of another job, which could be changed to start two, or whatever else is required.
I just learned from that the simultaneous allocation of multiple slaves can be done nicely in a pipeline job, by nesting node clauses:
node('label1') {
node('label2') {
// your code here
[...]
}
}
See this question where Mateusz suggested that solution for a similar problem.
Let me see if I understood the problem.
You want to have dynamically choose a slave and start the App Server on it.
When the App server is running on a Slave you do not want it to run any other job.
But when the App server is not running, you want to use that Slave as any other Slave for other jobs.
One way out will be to label the Slaves. And use "Restrict where this project can be run" to make the App Server and the Test Suite run on the machines with Slave label.
Then in the slave nodes, put " # of executors" to 1. This will make sure that at anytime only one Job will run.
Next step will be to create a job to start the App Server and then kick off the Test job once the App Server start job is successful..
If your test job needs to know the server details of the machine where your App server is running then it becomes interesting.
I have a Jenkins master setup with several slaves in our primary data center and one slave in our DR/BC (disaster recovery/business continuity) data center.
I would like jobs to run on the DR slave regularly to ensure it stays up to date with the required software and doesn't get stale, but since our DR center is geographically distant from the resources used in the builds & tests (SAN, DB, etc.) jobs take 4x - 10x longer to run. This is fine for a DR scenario, but painful for everyday life.
It appears that Jenkins sorts the slaves alphabetically for selecting which to run a job on, which is unfortunate because our machine naming convention is based on datacenter location, and the DR slave is always picked first.
Is there a way to specify how Jenkins picks a slave? or a way to specify a "cost" (like a routing cost) of a slave, so that it will be picked less often?
The solution I have settled on:
Configure the DR slave Availability as "Take this slave on-line when in demand and off-line when idle"
Create a "BuildAll" job to launch all of the integration builds at the same time.
Schedule BuildAll to run repeatedly at 3am ( 0-10 3 * * * )
This will force the DR slave to come online, and run random jobs several times which should show if that slave has fallen behind in any required software, patches, etc.