If the application in the container writes or reads some files without using volume, the value of container_fs_writes_bytes_total metrics is always 0 when I upload one huge file and the application save it to one folder. And container_fs_reads_bytes_total is also always 0 when I download the uploaded file. Is there a way to get the disk IO which caused by the container.
Btw, I want to know the disk usage of each container in k8s. There is container_fs_usage_bytes metrics in cAdvisor. But the value isn't correct. If there are many container, the metrics of each container are totally same. The metrics value is the whole usage of the disk in machine. There is someone said use kubelet_vlume_stats_used_bytes, it also has the same problem.
Related
context
Our current context is the following: researchers are running HPC calculations on our Kubernetes cluster. Unfortunately, some pods cannot get scheduled because the container engine (here Docker) is not able to pull the images because the node is running out of disk space.
hypotheses
images too big
The first hypothesis is that the images are too big. This probably the case because we know that some images are bigger than 7 GB.
datasets being decompressed locally
Our second hypothesis is that some people are downloading their datasets locally (e.g. curl ...) and inflate them locally. This would generate the behavior we are observing.
Envisioned solution
I believe that this problem is a good case for a daemon set that would have access to the node's file system. Typically, this pod would calculate the total disk space used by all the pods on the node and would expose them as a Prometheus metric. From there is would be easy to set alert rules in place to check which pods have grown a lot over a short period of time.
How to calculate the total disk space used by a pod?
The question then becomes: is there a way to calculate the total disk space used by a pod?
Does anyone have any experience with this?
Kubernetes does not track overall storage available. It only knows things about emptyDir volumes and the filesystem backing those.
For calculating total disk space you can use below command
kubectl describe nodes
From the above output of the command you can grep ephemeral-storage which is the virtual disk size; this partition is also shared and consumed by Pods via emptyDir volumes, image layers,container logs and container writable layers.
Check where the process is still running and holding file descriptors and/or perhaps some space (You may have other processes and other file descriptors too not being released). Check Is that kubelet.
You can verify by running $ ps -Af | grep xxxx
With Prometheus you can calculate with the below formula
sum(node_filesystem_size_bytes)
Please go through Get total and free disk space using Prometheus for more information.
I've read through this page and I'm interested in where Kubernetes downloads an image to and how long it stores it for.
For example, let's say we have a large 3GB image. When i start up a pod will the image be downloaded to disk of the node the pod is being deployed to, and remain until that node is destroyed? If so does that mean i could allocate only 400MB of memory to a pod that is using a 3GB image?
As mentioned in comments correctly ,CRI does this rather than K8s, Assuming you are running Docker
If you want to access the image data directly, it’s usually stored in the following locations:
Linux: /var/lib/docker/
Windows: C:ProgramDataDockerDesktop
macOS: ~/Library/Containers/com.docker.docker/Data/vms/0/
If you are using containerd as runtime then the images are stored at /var/lib/containerd
it is configured in /etc/containerd/config.toml as shown below
contianerd runtime is responsible for downoading and running container using the image.
I want to learn what is the proper way to reach the processed files when restarting the spring batch application on Kubernetes. Especially if the target type is file, it is being deleted together with the pod after the job failed.
We are considering to use persistent volume or backing up the created file somewhere such as DB or sftp server by implementing a listener.
Is there anyone have the experience of persistent volume usage(nfs or other solutions) for file operations. We are concerned about the performance and unexpected problems. Do you have any suggestions?
Thank you.
You should not rely on the ephemeral file system of a Pod for files that should persist and survive a Job (Pod) failure.
You need to use a persistent volume for that, so that Spring Batch can find the (incomplete) output file in a restart scenario and resume writing where it left off.
If you want data persistence, you may begin by using hostPath volumes first. This will restrict which nodes your pods may be spawned on. But is the simplest and gives you the best performance.
https://kubernetes.io/docs/concepts/storage/volumes/#hostpath
If you want dynamic allocation, you will need to configure storage solutions such as GlusterFS, NFS, CEPH etc.
I have a python code where I process some data, write neo4j queries and then commit these queries to neo4j. When I run the code on my local machine and write the output to local neo4j it doesn't take more than 15 minutes. However, when I run my code locally and write the output to noe4j pod in k8s pod it takes double the time, and when I build my code and deploy it to k8s and run that pod and write the output to neo4j pod it takes a round 3 hours. since I'm new to k8s deployment it might something in the pod configurations or settings, so I appreciate if I can get some hints
There could be few reasons of that.
I would first check how much resources does your pod consume while you are processing data, you can do that using kubectl top pod.
Second I would check if there are any limits inside pod. You can read a great deal about them on Managing Compute Resources for Containers.
If you have a limit set then it might be too low and that's causing the extended time while processing data.
If limits are not set then it might be because of how you installed minik8s. I think as default it's being installed with 4G is memory, you can look at alternative methods of installing minik8s. With multipass you can specify more memory to allocate.
There also can be a issue with Page Cache Sizing, Heap Sizing or number of open files. Please read the Neo4j Performance Tuning.
I have an application (JHipster Gateway, UAA, Registry, 5 microservices) and each application source builds a Docker image and pushes to GitLab registry. Currently I'm running everything on Rancher using a Docker-Compose file. My volumes for Mongo databases are currently in each container.
I need advice about volume mounts. Here are my options as I see them.
Leave data in containers and monitor and backup
Use external mounts and monitor volumes on host.
If I leave Mongo data in the containers, do I just set up to just cluster and when the internal volumes fill, the database just scales? I am looking for some explanation to help my choice with Mongo database mounts, internal or external (on host)?
Thanks in advance,
David L. Whitehurst
Never store any data you care about directly in containers. There are good arguments in favor of both named volumes (native to Docker, some support in a multi-host Swarm environment, fewer host-specific dependencies) and host bind mounts (much easier to back up and maintain, possible to examine directly if needed) but use some sort of mounted storage.
The most important note here is that it's fairly routine to delete and recreate containers. If the software you're running or its underlying library stack has a security issue, you generally need to get (or build) an updated image, delete your existing container, and rebuild it against the new image. If data is stored only inside a container, then during this very routine delete-and-recreate operation, there's significant risk of losing data.
In principle, if you're really careful, and you have a replicated data store, you can roll this over without external volumes and not lose data. It's tricky, and takes a lot of patience; you'll be forced to take down one replica, wait for its data to be rebalanced across the other replicas, start up a new replica, wait for it to accept some of the data, and so on. If you can take a point release by stopping a container, deleting it, starting a new one with the same data store, and have it come up instantly with populated data, that's much easier to manage.
(The other corollary here is that you don't "back up containers", since they don't have any data you care about. You do back up the data stored on the host or in Docker named volumes, and you can always recreate the container from its image plus the external data.)