Open maxisam opened 3 years ago
Is this a BUG REPORT?
I used default Rancher 2.5.7 setting for weave with 1376 MTU
My network is 10GBit/s
Pod to Pod should at least get 80% speed.
========================================================= Benchmark Results ========================================================= Name : knb-68458 Date : 2021-04-29 17:21:42 UTC Generator : knb Version : 1.5.0 Server : ck8sdl03 Client : ck8sdl02 UDP Socket size : auto ========================================================= Discovered CPU : Intel(R) Xeon(R) Gold 6150 CPU @ 2.70GHz Discovered Kernel : 5.4.0-65-generic Discovered k8s version : v1.19.9 Discovered MTU : 1376 Idle : bandwidth = 0 Mbit/s client cpu = total 5.80% (user 3.37%, nice 0.00%, system 1.68%, iowait 0.75%, steal 0.00%) server cpu = total 3.45% (user 2.19%, nice 0.00%, system 0.88%, iowait 0.38%, steal 0.00%) client ram = 1304 MB server ram = 1352 MB Pod to pod : TCP : bandwidth = 3438 Mbit/s client cpu = total 18.24% (user 2.37%, nice 0.00%, system 15.36%, iowait 0.51%, steal 0.00%) server cpu = total 27.32% (user 1.16%, nice 0.00%, system 25.88%, iowait 0.28%, steal 0.00%) client ram = 1308 MB server ram = 1351 MB UDP : bandwidth = 1820 Mbit/s client cpu = total 27.57% (user 3.36%, nice 0.00%, system 23.64%, iowait 0.57%, steal 0.00%) server cpu = total 31.07% (user 3.51%, nice 0.00%, system 27.18%, iowait 0.38%, steal 0.00%) client ram = 1308 MB server ram = 1352 MB Pod to Service : TCP : bandwidth = 3548 Mbit/s client cpu = total 19.13% (user 2.41%, nice 0.00%, system 16.29%, iowait 0.43%, steal 0.00%) server cpu = total 26.05% (user 1.58%, nice 0.00%, system 24.22%, iowait 0.25%, steal 0.00%) client ram = 1315 MB server ram = 1353 MB UDP : bandwidth = 1716 Mbit/s client cpu = total 26.88% (user 3.29%, nice 0.00%, system 23.21%, iowait 0.38%, steal 0.00%) server cpu = total 27.00% (user 3.30%, nice 0.00%, system 23.40%, iowait 0.30%, steal 0.00%) client ram = 1317 MB server ram = 1353 MB =========================================================
Rancher 2.5.7 Weave 2.7.0 K8S 1.19.9 Ubuntu 20.04.2 LTS Docker 20.10.3
Test it with 2.8.1.
Is this a BUG REPORT?
I used default Rancher 2.5.7 setting for weave with 1376 MTU
My network is 10GBit/s
What you expected to happen?
Pod to Pod should at least get 80% speed.
What happened?
How to reproduce it?
Anything else we need to know?
Versions:
Rancher 2.5.7 Weave 2.7.0 K8S 1.19.9 Ubuntu 20.04.2 LTS Docker 20.10.3