-
Task
-
Resolution: Unresolved
-
Major
-
None
-
None
-
None
-
Quality / Stability / Reliability
-
13
-
False
-
-
False
-
ToDo
-
-
-
-
Very Likely
-
0
-
None
-
Unset
-
Unknown
-
None
During OADP tests, we saw the 'dateuploads' rate on MINIO (pod) is ~1.5GBps.
The NICs and Switches are configured to 25Gb (~3GBps).
MINIO BM:
Kambiz's mail:
[kni@f01-h07-000-r640 kambiz]$ kubectl exec -it iperf3-deployment-7f4dfdd86-pkrgk – iperf3 -c 10.129.1.69 -p 5201 -i 1 -b 20000M
Connecting to host 10.129.1.69, port 5201
[ 5] local 10.130.0.22 port 50364 connected to 10.129.1.69 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 1.77 GBytes 15.2 Gbits/sec 1288 436 KBytes
[ 5] 1.00-2.00 sec 1.72 GBytes 14.8 Gbits/sec 1135 567 KBytes
[ 5] 2.00-3.00 sec 1.87 GBytes 16.1 Gbits/sec 1163 471 KBytes
[ 5] 3.00-4.00 sec 1.78 GBytes 15.2 Gbits/sec 1093 444 KBytes
[ 5] 4.00-5.00 sec 1.86 GBytes 16.0 Gbits/sec 1341 507 KBytes
[ 5] 5.00-6.00 sec 1.79 GBytes 15.4 Gbits/sec 1598 495 KBytes
[ 5] 6.00-7.00 sec 1.84 GBytes 15.8 Gbits/sec 1597 436 KBytes
[ 5] 7.00-8.00 sec 1.88 GBytes 16.1 Gbits/sec 668 519 KBytes
[ 5] 8.00-9.00 sec 1.84 GBytes 15.8 Gbits/sec 1380 363 KBytes
[ 5] 9.00-10.00 sec 1.80 GBytes 15.5 Gbits/sec 1127 569 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 18.2 GBytes 15.6 Gbits/sec 12390 sender
[ 5] 0.00-10.00 sec 18.1 GBytes 15.6 Gbits/sec receiver
iperf Done.
I tried various bandwidth limitations and noticed that I could not push greater than 15gb/s though if I'm not mistaken, the link speed theoretical max should be 25gb/s. The measurement of 15 gb/s is still significant, though I would have expected the measurement to be closer to 25 gb/s, which is the link speed. This might be due to other activity in your openshift environment that are limiting how much bandwidth the pods can use, or just a setting within openshift to ensure no pods will saturate the worker links.
I thought I would compare the network traffic between two pods (running on different worker nodes as above), to bare metal hosts installed as vanilla RHEL systems in scalelab.