Containers, Kubernetes, OpenShift on Power

 View Only

Shared Ethernet Adapter setting crucial for etcd

  • 1.  Shared Ethernet Adapter setting crucial for etcd

    Posted Wed December 14, 2022 10:16 AM

    The etcd cluster formed during OpenShift installation requires some large messages to be exchanged.

    If the etcd logs indicate that these messages are not getting through (for e.g. this is what you would see in the logs), it might be an indication that the Shared Ethernet Adapter does not have largesend/large_receive enabled.

    2022-12-02T22:55:42.870967877+00:00 stderr F {"level":"info","ts":"2022-12-02T22:55:42.870Z","caller":"rafthttp/http.go:257","msg":"receiving database snapshot","local-member-id":"c280448f07524f27","remote-snapshot-sender-id":"16d21c31416797c6","incoming-snapshot-index":1810535,"incoming-snapshot-message-size-bytes":378299,"incoming-snapshot-message-size":"378 kB"}
    2022-12-02T22:55:48.433915582+00:00 stderr F {"level":"warn","ts":"2022-12-02T22:55:48.433Z","caller":"etcdserver/server.go:2084","msg":"failed to publish local member to cluster through raft","local-member-id":"c280448f07524f27","local-member-attributes":"{Name:p9mas2 ClientURLs:[]}","request-path":"/0/members/c280448f07524f27/attributes","publish-timeout":"7s","error":"etcdserver: request timed out, possibly due to connection lost"}

    The other indication will be that entstat/netstat output on the VIOS will show the ICMP error packets sent. counter incrementing.

    Please follow the steps in this document to modify the settings for largesend/large_receive

    Chief Engineer, OpenShift on power