https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# rke2
a
This message was deleted.
p
Yeah it sounds like your coredns is dying from so much dns requests. This could be because your underlying storage may be slow, or your server cpu may be a bit weak in mono-thread performance.
n
We are using SSDs as storage. And also scaled up the coredns replicas from 3 to 7. But none of them helped me. can you please give me some more info about how to find the CPU mono thread performance?
p
Do you know on what physical hosts your nodes run?
n
Yes on Bare Metal VMWare Hosts
p
Yeah but do you know what cpus thoses hosts have?
n
no I don't have that information
p
RIP then 😞
n
😁
I can get that information. Can you please give me some information on what to check after getting the info?
p
No i don't think it's worth it
n
Host CPU Details: 64 CPUs x 2.2 GHz Intel(R) Xeon(R) Gold 6338N CPU 2 sockets 32 cores per socket 128 Logical Processors Active Hyperthreading Are these the details? or you require any specific details?
p
Oh yeah no those are beasts
Definitely not the issue
n
any idea, what might be causing this issue?
I feel like too much load on coredns pods, when creating the dynamic pods
p
The fact that this issue seems to only occur on high load could be also the result of an underlying race condition, overall if you can make do with creating 5 pods at a time, i think it wouldn't be worth it to spend more time on this
Just record the issue on github
n
https://github.com/rancher/rke2/discussions/5879 Already Created. Can you please take a look?
👍 1
p
lgtm. Good luck !
2 Views
Open in Slack
PreviousNext
Powered by