search

cloud-bulldozer / benchmark-operator

The Chuck Norris of cloud benchmarks
Apache License 2.0
282 stars 127 forks source link

All client and Server pods are scheduled to the different nodes with 2pairs of server and client scenario #800

Closed SachinNinganure closed 1 year ago

SachinNinganure commented 1 year ago

the document says "https://github.com/cloud-bulldozer/benchmark-operator/blob/master/docs/uperf.md"

By default, the uperf server and client pods will be preferably scheduled on different nodes, thanks to a podAntiAffinity rule. In scenarios with more than 1 pair, all clients and servers will be scheduled to the same node.

But one of the uperf tests I observed : Running a scenario with 2 pairs!!! [sninganu@sninganu ~]$ oc get pods -owide NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES backpack-7d4e42be-7cmgl 1/1 Running 0 9m 10.131.0.26 sn-npt-ibm412-bcxr8-worker-1-nxqvc backpack-7d4e42be-h4zb5 1/1 Running 0 9m 10.128.2.21 sn-npt-ibm412-bcxr8-worker-2-ch7ct backpack-7d4e42be-hr7nz 1/1 Running 0 9m 10.129.0.39 sn-npt-ibm412-bcxr8-master-1 backpack-7d4e42be-ktjvb 1/1 Running 0 9m 10.128.0.53 sn-npt-ibm412-bcxr8-master-2 backpack-7d4e42be-qjjph 1/1 Running 0 9m 10.130.0.62 sn-npt-ibm412-bcxr8-master-0 backpack-7d4e42be-wgtwn 1/1 Running 0 9m 10.129.2.14 sn-npt-ibm412-bcxr8-worker-3-sw9vf

benchmark-controller-manager-7d694c6b9c-tspwh 2/2 Running 0 10m 10.131.0.25 sn-npt-ibm412-bcxr8-worker-1-nxqvc uperf-client-172.30.123.203-7d4e42be-sppdp 1/1 Running 0 7m5s 10.131.0.28 sn-npt-ibm412-bcxr8-worker-1-nxqvc uperf-client-172.30.223.179-7d4e42be-7wblr 1/1 Running 0 7m6s 10.131.0.27 sn-npt-ibm412-bcxr8-worker-1-nxqvc uperf-server-0-7d4e42be-hsc25 1/1 Running 0 7m38s 10.128.2.22 sn-npt-ibm412-bcxr8-worker-2-ch7ct uperf-server-1-7d4e42be-twkdq 1/1 Running 0 7m38s 10.128.2.23 sn-npt-ibm412-bcxr8-worker-2-ch7ct

I triggered a test for 2pairs and could see that the server and client pods are scheduled on different nodes as above; test → https://mastern-jenkins-csb-openshift-qe.apps.ocp-c1.prod.psi.redhat.com/job/scale-ci/job/e2e-benchmarking-multibranch-pipeline/job/network-perf/372/parameters/

jtaleric commented 1 year ago

In scenarios with more than 1 pair, all clients and servers will be scheduled to the same node.

All the clients will be schedule to the same node, and the servers will be scheduled on a different node.. How many nodes did your env have?

SachinNinganure commented 1 year ago

3 worker and 3 master nodes.

stale[bot] commented 1 year ago

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.