Is it possible to specify the configurations I want for a single dataflow worker? It seems like the default contains 4 cores and 15GB of memory, which is more than enough. How can I down size it or is this the smallest unit of a worker being offered?
Per the Workers section of the Dataflow > How-to Guides > Deploying a Pipeline page, you can specify a custom machine type (with different cores or memory) with the --worker_machine_type option.
You can also see the other Dataflow worker-related options in the docs/source code for the WorkerOptions class, which parses the various worker-related command-line options. Some of the other options listed here include: disk_size_gb, worker_disk_type.
Tangentially related: Additional GCP-related Dataflow options are handled by the GoogleCloudOptions class.
Related
I am just looking into using GCP for cloud computing stuff. So far I have been using AWS and the boto3 library and was trying to use the google python client API for launching instances.
So an example I came across was from their docs here. The instance machine type is specified as:
machine_type = "zones/%s/machineTypes/n1-standard-1" % zone
and then it passed to the configuration as:
config = {
'name': name,
'machineType': machine_type,
....
I wonder how does one go about specifying machines with GPU and custom RAM and processors etc. from the python API?
The Python API is basically a wrapper around the REST API, so in the example code you are using, the config object is being built using the same schema as would be passed in the insert request.
Reading that document shows that the guestAccelerators structure is the relevant one for GPUs.
Custom RAM and CPUs are more interesting. There is a format for specifying a custom machine type name (you can see it in the gcloud documentation for creating a machine type). The format is:
[GENERATION]custom-[NUMBER_OF_CPUs]-[RAM_IN_MB]
Generation refers to the "n1" or "n2" in the predefined names. For n1, this block is empty, for n2, the prefix is "n2-". That said, experimenting with gcloud seems to indicate that "n1-" as a prefix also works as you would expect.
So, for a 1 CPU n1 machine with 5GB of ram, it would be a custom-1-5120. This is what you would replace the n1-standard-1 in your example with.
You are, of course, subject to the limits of how to specify a custom machine such as the fact that RAM must be a multiple of 256MB.
Finally, there's a neat little feature at the bottom of the console "create instance" page:
Clicking on the relevant link will show you the exact REST object you need to create the machine you have defined in the console at that very moment, so it can be very useful to see how a particular parameter is used.
You can create a Compute Engine instance using the Compute Engine API. Specifically, we can use the insert API request. This accepts a JSON payload in a REST request that describes the desired VM instance that you desire. A full specification of the request is found in the docs. It includes:
machineType - specs of different (common) machines including CPUs and memory
disks - specs of disks to be added including size and type
guestAccelerators - specs for GPUs to add
many more options ...
One can also create a template description of the machine structure you want and simplify the creation of an instance by naming the template to use and thereby abstracting the configuration details out of code and into configuration.
Beyond using REST requests (which can be passed from a python), you also have the capability to create Compute Engines from:
GCP Console - web interface
gcloud - command line (which I suspect can also be driven from within Python)
Deployment Manager - configuration driven deployment which includes Python as a template language
Terraform - popular environment for creating Infrastructure as Code environments
I am trying to run a CI/CD on my codebase, but in order to run my tests, I need a GPU-enabled VM (to produce deep learning results).
However, the only configurable machine option I see is the machine type (number of cores and memory). I don't see an option for adding an accelerator type (GPU).
Is there a way to attach a GPU to the build VM, and if not, is there another method for triggering a test on another GPU enabled VM?
Thanks!
Google Cloud Build doesn't provide machine types equipped with GPUs at the moment. One option though is to use the remote-builder cloud builder. It allows you to run your builds on a Compute Engine instances running in your project. You can use the INSTANCE_ARGS option to customize the instance to fit your specific needs, adding one or more GPUs in this case. You can have a look here for some example configs. You can use any flag available with the gcloud compute instances create command, including the --accelerator flag for GPUs.
I am really hoping to use Presto in an ETL pipeline on AWS EMR, but I am having trouble configuring it to fully utilize the cluster's resources. This cluster would exist solely for this one query, and nothing more, then die. Thus, I would like to claim the maximum available memory for each node and the one query by increasing query.max-memory-per-node and query.max-memory. I can do this when I'm configuring the cluster by adding these settings in the "Edit software settings" box of the cluster creation view in the AWS console. But the Presto server doesn't start, reporting in the server.log file an IllegalArgumentException, saying that max-memory-per-node exceeds the useable heap space (which, by default, is far too small for my instance type and use case).
I have tried to use the session setting set session resource_overcommit=true, but that only seems to override query.max-memory, not query.max-memory-per-node, because in the Presto UI, I see that very little of the available memory on each node is being used for the query.
Through Google, I've been led to believe that I need to also increase the JVM heap size by changing the -Xmx and -Xms properties in /etc/presto/conf/jvm.config, but it says here (http://docs.aws.amazon.com/emr/latest/ReleaseGuide/emr-presto.html) that it is not possible to alter the JVM settings in the cluster creation phase.
To change these properties after the EMR cluster is active and the Presto server has been started, do I really have to manually ssh into each node and alter jvm.config and config.properties, and restart the Presto server? While I realize it'd be possible to manually install Presto with a custom configuration on an EMR cluster through a bootstrap script or something, this would really be a deal-breaker.
Is there something I'm missing here? Is there not an easier way to make Presto allocate all of a cluster to one query?
As advertised, increasing query.max-memory-per-node, and also by necessity the -Xmx property, indeed cannot be achieved on EMR until after Presto has already started with the default options. To increase these, the jvm.config and config.properties found in /etc/presto/conf/ have to be changed, and the Presto server restarted on each node (core and coordinator).
One can do this with a bootstrap script using commands like
sudo sed -i "s/query.max-memory-per-node=.*GB/query.max-memory-per-node=20GB/g" /etc/presto/conf/config.properties
sudo restart presto-server
and similarly for /etc/presto/jvm.conf. The only caveats are that one needs to include the logic in the bootstrap action to execute only after Presto has been installed, and that the server on the coordinating node needs to be restarted last (and possibly with different settings if the master node's instance type is different than the core nodes).
You might also need to change resources.reserved-system-memory from the default by specifying a value for it in config.properties. By default, this value is .4*(Xmx value), which is how much memory is claimed by Presto for the system pool. In my case, I was able to safely decrease this value and give more memory to each node for executing the query.
As a matter of fact, there are configuration classifications available for Presto in EMR. However, please note that these may vary depending on the EMR release version. For a complete list of the available configuration classifications per release version, please visit 1 (make sure to switch between the different tabs according to your desired release version). Specifically regarding to jvm.config properties, you will see in 2 that these are not currently configurable via configuration classifications. That being said, you can always edit the jvm.config file manually per your needs.
Amazon EMR 5.x Release Versions
1
Considerations with Presto on Amazon EMR - Some Presto Deployment Properties not Configurable:
2
When you pick a more performant node, say a r3.xlarge vs m3.xlarge, will Spark automatically utilize the additional resources? Or is this something you need to manually configure and tune?
As far as configurations go, which are the most configuration values to tune to get the most out of your cluster?
It will try..
AWS has a setting you can enable in your EMR cluster configuration that will attempt to do this. It is called spark.dynamicAllocation.enabled. In the past there were issues with this setting where it would give too many resources to Spark. In newer releases they have lowered the amount they are giving to spark. However, if you are using Pyspark they will not take python's resource requirements into account.
I typically disable dynamicAllocation and set the appropriate memory and cores settings dynamically from my own code based upon what instance type is selected.
This page discusses what defaults they will select for you:
http://docs.aws.amazon.com/ElasticMapReduce/latest/ReleaseGuide/emr-spark-configure.html
If you do it manually, at a minimum you will want to set:
spark.executor.memory
spark.executor.cores
Also, you may need to adjust the yarn container size limits with:
yarn.scheduler.maximum-allocation-mb
yarn.scheduler.minimum-allocation-mb
yarn.nodemanager.resource.memory-mb
Make sure you leave a core and some RAM for the OS, and RAM for python if you are using Pyspark.
We use Jenkins for our CI build system. We also use 'concurrent builds' so that Jenkins will build each change independently. This means we often have 5 or 6 builds of the same job running simultaneously. To accommodate this, we have 4 slaves each with 12 executors.
The problem is that Jenkins doesn't really 'load balance' among its slaves. It tries to build a job on the same slave that it previously built on (presumably to reduce the time syncing from source control). This is a problem because Jenkins will build all 6 instances of our build on the same slave (or more likely between 2 slaves). One build machine gets bogged down and runs very slowly while the rest of them sit idle.
How do I configure the load balancing behavior of Jenkins, and how it controls its slaves?
We were facing a similar issue. So I've put together a plugin that changes the Load Balancer in Jenkins to select a node that currently has the least load - https://plugins.jenkins.io/leastload/
Any feedback is appreciated.
If you do not find a plugin that does it automatically, here's an idea of what you can do:
Install Node Label Parameter plugin
Add SLAVE parameter to your jobs
Restrict jobs to run on ${SLAVE}
Add a trigger job that will do the following:
Analyze load distribution via a System Groovy Script and decide on which node to start next build.
Dispatch the build on that node with Parameterized Trigger
plugin
by assigning appropriate value to SLAVE parameter.
In order to analyze load distribution you need to install Groovy plugin and familiarize yourself with Jenkins Main Module API. Here are some useful initial pointers.
If your build machines cannot comfortably handle more than 1 build, why configure them with 12 executors? If that is indeed the case, you should reduce the number of executors to 1. My Jenkins has 30 slaves, each with 1 executor.
You may also use the Throttle Concurrent Builds plugin to restrict how many instances of a job can run in parallel on the same node
I have two labels -- one for small tasks and one for big tasks. I have one executor for the big task and 4 for the small tasks. This does balance things a little.