Forum Discussion
MJ_1024
Altocumulus
AltocumulusErrors in setup of BIGIP-NEXT CM VE on KVM
I've been hammering at trying to get a stable BIGIP-NEXT CM VE running in a lab for a while now and keep running into issues. This would be the new and improved v20.2.0 in KVM (qcow image).
It seems to never fully install properly, or the services fail to run entirely.
I've gotten the web interface to load a few times, but it crashes before doing anything useful.
I do have one error that keeps coming up is during initial setup, I see a warning when installing "mbiq" pod.
Something about "path cannot be used with pathType Prefix". Searching around, I see some threads for Kubernetes ingress controller issues, but that's about it.
Not sure if how this may affect the installation, but it seems to hang after that.
If I kill the setup, or let it eventually fail on its own, the second attempt to run setup appears to work, but watching "kubectl get pods" many of the containers fail to initialize, which may explain why the WebUI crashes and is otherwise not functional.
The main resource I've been using has limited troubleshooting details:
https://clouddocs.f5.com/bigip-next/latest/install/cm_install_setup_mbip_kvm.html
I haven't seen any known bugs with these indicators.
Curious if anyone has thoughts on this / ideas?
That does sound frustrating. A few things to try:
1- Make sure you have EXACTLY one network interface on your guest
2- Make sure you have 8 cores and 16GB
3- Check that the hypervisor isn't oversubscribed (CPU or memory): On slow systems, there can be timeout problems. With low memory the pods will constantly crash.
4- On the crashed pods, try checking out the logs:
https://kubernetes.io/docs/reference/kubectl/generated/kubectl_logs/One other thing is that the setup command for both Central Manager and the BIP Next Instances does produce many warnings that can be ignored, and the message you see may be normal.
- Lucas_Thompson
Employee
That does sound frustrating. A few things to try:
1- Make sure you have EXACTLY one network interface on your guest
2- Make sure you have 8 cores and 16GB
3- Check that the hypervisor isn't oversubscribed (CPU or memory): On slow systems, there can be timeout problems. With low memory the pods will constantly crash.
4- On the crashed pods, try checking out the logs:
https://kubernetes.io/docs/reference/kubectl/generated/kubectl_logs/One other thing is that the setup command for both Central Manager and the BIP Next Instances does produce many warnings that can be ignored, and the message you see may be normal.
MJ_1024Yes, from what I could tell several services (pods) were failing to start, although it did often look like 1/2 of them would.
There were enough crashed (or failed to start nodes) that logs were difficult to parse, but I certainly could try collecting them all.
As for performance, the Cores and CPU are matched, but the disk on that system was pretty slow (hdd).
I will try moving it over to faster NVME storage and try some more, here is hoping!- MJ_1024
So, sure enough, when using NVME (fast) storage instead of HDD (slow) no issues observed on install or run.
So the warning about slow systems clearly applies to storage as well.