-
Notifications
You must be signed in to change notification settings - Fork 2.8k
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
cilium_bpf_map_pressure show incorrect map utilization #32311
Comments
Thanks for this issue @spal-x, but unfortunately without some extra information like the zip file produced by the process at https://docs.cilium.io/en/stable/operations/troubleshooting/#automatic-log-state-collection, it will be difficult to answer your question. Could you upload a sysdump and I will mark this for the correct team's attention? |
Unfortunately I can only provide cilium bugtool output. It`s enough? |
Anything you can give will help, yes. And yes, I believe your understanding of the metric is correct. |
cilium-bugtool-20240503-091354.767+0000-UTC-1826744592.zip I delete some large table dumps (like connection tracking or SNAT). |
Any info? |
I'd suggest reproducing on a current release. Even for v1.13 you're already missing a full year of fixes... |
Sorry for the long response. I observe same behavior on 1.15.2 version in our test enviroment. |
Is there an existing issue for this?
What happened?
We encountered errors when trying to create a service:
2024-04-24T18:48:56.681579908+03:00 level=error msg="Error while inserting service in LB map" error="Unable to update service entry [10.237.56.27:4699](http://10.237.56.27:4699/) => 225157 0 (19535) [0x0 0x0]: Unable to update element for LB bpf map: You can resize it with the flag \"--bpf-lb-map-max\". The resizing might break existing connections to services" k8sNamespace=dev-webinar-bl k8sSvcName=webinar-bl subsys=k8s-watcher
According to documentation
we check metric
cilium_bpf_map_pressure
with labelmap_name="lb4_services_v2"
and saw its value is equal0.420758
Configured max size is 65536.
When we superficially check real map size with bpftool on nodes, we made sure that the map is full and has 65536 entries.
The question arises: why do we get the wrong metric value?
Cilium Version
Client: 1.13.2 8cb94c7 2023-04-17T23:19:21+02:00 go version go1.19.8 linux/amd64
Daemon: 1.13.2 8cb94c7 2023-04-17T23:19:21+02:00 go version go1.19.8 linux/amd64
Kernel Version
Linux 5.14.0-168.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Fri Sep 23 11:43:25 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
Kubernetes Version
1.27.1
Regression
No response
Sysdump
No response
Relevant log output
No response
Anything else?
No response
Cilium Users Document
Code of Conduct
The text was updated successfully, but these errors were encountered: