On below dashboard, yellow line represents memory limit, red line represents memory requests and green memory usage. Why is memory consumption still being reported by Prometheus after job completion? I checked job logs and job completed same time when memory request and limits went to 0. Job TTL is set to 60 sec so I think it's not related.
prometheus grafana metrics
Factually a completed job means the process is no longer running let a lone consuming any resources. So, what you see in your logs is probably due to a delay in the metrics refresh period.
Keep in mind that K8s-related metrics like resource requests are reported by reaping information from the K8s API-Server, whereas actual resource consumption is reported by different infrastructure components, e.g. the Metrics Server. Those systems will likely have different refresh periods, which explains the discrepancy when you aggregate them on the same chart.
Related
I am running load test of 24 hours using Jmeter in Azure Kubernetes service. I am using Throughput shaping timer in my jmx file. No listener is added as part of jmx file.
My test stopped abruptly after 6 or 7 hrs.
jmeter-server.log file under Jmeter slave pod is giving warning --> WARN k.a.j.t.VariableThroughputTimer: No free threads left in worker pool.
Below is snapshot from jmeter-server.log file.
Using Jmeter version - 5.2.1 and Kubernetes version - 1.19.6
I checked, Jmeter pods for master and slaves are continously running(no restart happened) in AKS.
I provided 2GB memory to Jmeter slave pod still load test is stopped abruptly.
I am using log analytics workspace for logging. Checked ContainerLog table not getting error.
Snapshot of JMX file.
Using following elements -> Thread Group, Throughput Controller, Http request Sampler and Throughput Shaping Timer
Please suggest for same.
It looks like your Schedule Feedback Function configuration is wrong in its last parameter
The warning means that the Throughput Shaping Timer attempts to increase the number of threads to reach/maintain the desired concurrency but it doesn't have enough threads in order to do this.
So either increase this Spare threads ration to be closer to 1 if you're using a float value for percentage or increment the absolute value in order to match the number of threads.
Quote from documentation:
Example function call: ${__tstFeedback(tst-name,1,100,10)} , where "tst-name" is name of Throughput Shaping Timer to integrate with, 1 and 100 are starting threads and max allowed threads, 10 is how many spare threads to keep in thread pool. If spare threads parameter is a float value <1, then it is interpreted as a ratio relative to the current estimate of threads needed. If above 1, spare threads is interpreted as an absolute count.
More information: Using JMeter’s Throughput Shaping Timer Plugin
However it doesn't explain the premature termination of the test so ensure that there are no errors in jmeter/k8s logs, one of the possible reasons is that JMeter process is being terminated by OOMKiller
Suppose I have an API that allows users to create jobs (V1Jobs) in Kubernetes. What I would like is for the user to be able to submit as many jobs as they want without a failed response from the API but to have Kubernetes queue/throttle the jobs until there are enough available resources in the given namespace. For example, suppose I create a resource quota and specify a limit of 1cpu and 1Gi memory. Then suppose a user submits 100 1cpu/1Gi jobs. I'd like Kubernetes to process one at a time until they are complete. In other words running 100 jobs one at a time. Is creating the resource quota and letting the job-controller/scheduler handle the throttling the right way to go or would there be benefits to handle tracking the cluster usage externally (in an application) and only submit/create the V1Jobs to the API once there is capacity in the namespace?
ResourceQuotas are a good start, limiting the amount of resources a that may be used within a namespace, or by resources matching an expression.
It would indeed prevent the scheduler from creating Pods that would exceed your quota limitations. The API would still accept new Job objects posted by clients. If you have N Jobs requesting 1 CPU/1G RAM, while your quota only allows for less than 2CPU/2G RAM to be used, you should see those jobs running sequentially.
Though it could still make sense to track how many pending/running jobs you have in your namespace, as this could show there are currently too many jobs to run with your current quota configuration. The kube-state-metrics exporter from Prometheus would gather the metrics you need for this, you'ld find sample dashboards in Grafana, alerting rules over there.
If there's a risk some containers would start without passing proper cpu or memory resources requests / limits, you could also look into LimitRanges, forcing some defaults.
Resource quotas are applied at the namespace level. For containers resourceRequests and resourceLimits are available to specify the min and Max cpu and memory respectively. Setting resourceRequests and resourceLimits, allows the job to use only the specified cpu and memory range. This way we can schedule more than one job in a namespace with both resourceQuotas at namespace level and resourceRequests and resourceLimits at jobs.
Also, there is no harm in letting the scheduler takes care of not scheduling jobs when there is no resource.
According to both of these Link1 and Link2, my Airflow DAG run is returning the error INFO - Task exited with return code -9 due to an out-of-memory issue. My DAG run has 10 tasks/operators, and each task simply:
makes a query to get one of my BigQuery tables, and
writes the results to a collection in my Mongo database.
The size of the 10 BigQuery tables range from 1MB to 400MB, and the total size of all 10 tables is ~1GB. My docker container has default 2GB of memory and I've increased this to 4GB, however I am still receiving this error from a few of the tasks. I am confused about this, as 4GB should be plenty of memory for this. I am also concerned because, in the future, these tables may become larger (a single table query could be 1-2GB), and I'd like to avoid these return code -9 errors at that time.
I'm not quite sure how to handle this issue, since the point of the DAG is to transfer data from BigQuery to Mongo daily, and the queries / data in-memory for the DAG's tasks is necessarily fairly large then, based on the size of the tables.
As you said, the error message you get corresponds to an out of memory issue.
Referring to the official documentation:
DAG execution is RAM limited. Each task execution starts with two
Airflow processes: task execution and monitoring. Currently, each node
can take up to 6 concurrent tasks. More memory can be consumed,
depending on the size of the DAG.
High memory pressure in any of the GKE nodes will lead the Kubernetes scheduler to evict pods from nodes in an attempt to relieve that pressure. While many different Airflow components are running within GKE, most don't tend to use much memory, so the case that happens most frequently is that a user uploaded a resource-intensive DAG. The Airflow workers run those DAGs, run out of resources, and then get evicted.
You can check it with following steps:
In the Cloud Console, navigate to Kubernetes Engine -> Workloads
Click on airflow-worker, and look under Managed pods
If there are pods that show Evicted, click each evicted pod and look for the The node was low on resource: memory message at the top of the window.
What are the possible ways to fix OOM issue?
Create a new Cloud Composer environment with a larger machine type than the current machine type.
Ensure that the tasks in the DAG are idempotent, which means that the result of running the same DAG run multiple times should be the same as the result of running it once.
Configure task retries by setting the number of retries on the task - this way when your task gets -9'ed by the scheduler it will go to up_for_retry instead of failed
Additionally you can check the behavior of CPU:
In the Cloud Console, navigate to Kubernetes Engine -> Clusters
Locate Node Pools at the bottom of the page, and expand the default-pool section
Click the link listed under Instance groups
Switch to the Monitoring tab, where you can find CPU utilization
Ideally, the GCE instances shouldn't be running over 70% CPU at all times, or the Composer environment may become unstable during resource usage.
I hope you find the above pieces of information useful.
I am going to chunk the data so that less is loaded into any 1 task at any given time. I'm not sure yet whether I will need to use GCS/S3 for intermediary storage.
I have a bit of a unique use-case where I want to run a large number (thousands to tens of thousands) of Kubernetes Jobs at once. Each job consists of a single container, Parallelism 1 and Completions 1, with no side-car or agent. My cluster has plenty of capacity for the resources I'm requesting.
My problem is that the Job status is not transitioning to Complete for a significant period of time when I run many jobs concurrently.
My application submits Jobs and has a watcher on the namespace - as soon as a Job's status transitions to 'succeeded 1', we delete the Job and send information back to the application. The application needs this to happen as soon as possible in order to define and submit subsequent Jobs.
I'm able to submit new Job requests as fast as I want, and Pod scheduling happens without delay, but beyond about one or two hundred concurrent Jobs I get significant delay between a Job's Pod completing and the Job's status updating to Complete. At only around 1,000 jobs in the cluster, it can easily take 5-10 minutes for a Job status to update.
This tells me there is some process in the Kubernetes Control Plane that needs more resources to process Pod completion events more rapidly, or a configuration option that enables it to process more tasks in parallel. However, my system monitoring tools have not yet been able to identify any Control Plane services that are maxing out their available resources while the cluster processes the backlog, and all other operations on the cluster appear to be normal.
My question is - where should I look for system resource or configuration bottlenecks? I don't know enough about Kubernetes to know exactly what components are responsible for updating a Job's status.
I have sidekiq custom metrics coming from prometheus adapter. Using thoes queue metrics from prometheus i have setup HPA. When jobs in queue in sidekiq goes above say 1000 jobs HPA triggers 10 new pods. Then each pod will execute 100 jobs in queue. When jobs are reduced to say 400. HPA will scale-down. But when scale-down happens, hpa kills pods say 4 pods are killed. Thoes 4 pods were still running jobs say each pod was running 30-50 jobs. Now when hpa deletes these 4 pods, jobs running on them are also terminated. And thoes jobs are marked as failed in sidekiq.
So what i want to achieve is stop hpa from deleting pods which are executing the jobs. Moreover i want hpa to not scale-down even after load is reduced to minimum, instead delete pods when jobs in queue in sidekiq metrics is 0.
Is there any way to achieve this?
Weird usage, honestly: you're wasting resources even your traffic is on the cool-down phase but since you didn't provide further details, here it is.
Actually, it's not possible to achieve what you desire since the common behavior is to support a growing load against your workload. The unique wait to achieve this (and this is not recommended) is to change the horizontal-pod-autoscaler-downscale-stabilization Kubernetes Controller Manager's flag to a higher value.
JFI, the doc warns you:
Note: When tuning these parameter values, a cluster operator should be aware of the possible consequences. If the delay (cooldown) value is set too long, there could be complaints that the Horizontal Pod Autoscaler is not responsive to workload changes. However, if the delay value is set too short, the scale of the replicas set may keep thrashing as usual.
As per the discussion and the work done by #Hb_1993 it can be done with a pre-stop hook to delay the eviction, where the delay is based on operation time or some logic to know if the procession is done or not.
A pre-stop hook is a lifecycle method which is invoked before a pod is evicted, and we can then attach to this event and perform some logic like performing ping check to make sure that our pod has completed the processing of current request.
PS- Use this solution with a pinch of salt as this might not work in all the cases or produce unintended results.
To do this, we introduce asleep in the preStop hook that delays the
shutdown sequence.
More details can be found in this article.
https://blog.gruntwork.io/delaying-shutdown-to-wait-for-pod-deletion-propagation-445f779a8304