-
Notifications
You must be signed in to change notification settings - Fork 102
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
IO Engine Cannot set affinity #1458
Comments
It looks like it doesn't respect kubelet cpu static policy. |
hmm I'm not too falimiar with cpu policies but seems this may be true. @Abhinandan-Purkait ? |
Hi, I am getting the same error on 2 servers, while the third one managed to start the pod:
I'm using microk8s and installed mayastor via add-on. Kubernetes version is 1.27 with mayastor 2.0.0.
As you can see 2 If not the core count, could that be the CPU frequency too low? The server that managed to start mayastor-io-engine runs at 3.0 Ghz, while the 2 servers that failed have a lower spec CPU running at 1.7 Ghz. I would not want to change the CPUs right now, so is there another way? |
How many cpu cores on these 2 servers? |
I have allocated 8 cores, 16 GB of RAM, and 64 GB of space, on all 3 servers. I will try to add more cores - 32, and will get back with the results. Update Added 32 to cores to LXC container running microk8s. Rebooted the container and added
On the other server, that still has 8 cores, I get slightly different output
|
@tiagolobocastro Any ideas? |
Is there some kind of limit to your lxc container to run on a subset of your cpus? |
I tried to install v2.3.0 from a chart and it did not help. There are no limits for LXC container. I decided to upgrade CPUs, and if it helps I will post an update. |
If it doesn't help, would you be able to change io-engine container image to something else that would allow you to run this from the container:
Also, do you have a cpu manager policy of static? |
I've tested this with lxd, and when we limit lxc containers to cpu, indeed I start to see the cpu allowed list being setup by lxc, example:
In this case to get io-engine to run I had to change the cpu-list to those... |
@tiagolobocastro, sorry I forgot to update. I have replaced the CPUs, but that did not resolve the issue. I think one of the issues I have experienced with k8s in LXC and various storage solutions, including ebs, ceph(csi driver), and others, was the inability to mount new drive inside the lxc container(even though it was privileged). Can't remember exactly why, but it seems like a limitation of LXD all together. I did find a post regarding this... I ultimately just installed k8s bare bones on the server and most of those issues disappeared. I'm sure if I would try to run open ebs, it would work. So the issue is most likely related to running Kubernetes inside LXC. |
Describe the bug
First install of mayastor, I'm getting a "Cannot set affinity" error
To Reproduce
Steps to reproduce the behavior:
Expected behavior
** OS info (please complete the following information):**
Additional context
One is fine.
I also tried giving mayastor dedicated cpus and running helm upgrade. This lead to an etcd issue though.
The text was updated successfully, but these errors were encountered: