-
Notifications
You must be signed in to change notification settings - Fork 469
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
kube-router not working as expected #93
Comments
maybe related to #57? the service was really not created:
Oh, I'm using latest version of kube-router |
@felipejfc sorry about the issues you are running into. So far me and @bzub have been running Kubernetes e2e tests to catch the nagging bugs. Honestly we did not run the tests at the scale of yours. We would definitely attempt to run tests at scale ASAP to address any instability. We have been focussing on testing/CD to bring the project production grade reliability and stability before moving Beta release. Coming to this bug, do you happen to see any errors in the kube-router logs, related service not getting created? I will attempt to reporduce the bug and share my findings. |
@felipejfc quick update so wrote small script to test deploying large number of services. I am able to repro the problem Network service controller thread in Kube-router dies due to fatal failure in nlgo library used by Kube-router. We have an issue #78 reported for this. We recently updated dependent libraries, which forced us to update nlgo to latest version. I am able to repro the issue at just 10 services. So basically network service controller stops functioning due to nlgo's error I will investigate further and run some scale tests and get back to you. Redeploying the pod fixes the problem but need to make code resilient to nlgo failures. |
Nice @murali-reddy, if we can repro it then we can fix it :) Thanks!! |
@felipejfc If possible, please test the |
I'm curious since I'm running into the same problem w/ libipvs in a separate project. Did you figure out the actual issue in libipvs? |
@jsravn We hit this issue long time ago when kube-router was using https://github.com/mqliang/libipvs for IPVS, then we switched to docker/libnetwork. We never ran into this issue with docker/libnetwork. |
Can you give me a version of kube-router that had fixed this problem? (Including changed to libnetwork) |
I have a scenario where I have 19 nodes, running like 900 pods + 900 services, a service per pod, I'm tying kube-router and I've realised that some of the services (type nodeport) are not working, I can telnet to pods directly but not through the service:
As a side-effect I cant access the service through the nodeports.
This is not the only case that is not working as expected, in another cases, I can access a svc type NodePort from some nodes but in others the port is closed...
I really would like to use kube-router in production as it fits my need for performance and low overhead, but unfortunately, with this kind of instability it's impossible.
Any hints @murali-reddy ?
Thanks
The text was updated successfully, but these errors were encountered: