antrea icon indicating copy to clipboard operation
antrea copied to clipboard

Antrea Agent continues printing FQDN error logs like `"DNS exchange failed" err="read udp 10.176.27.243:46116->10.172.40.1:53: i/o timeout"`

Open wenyingd opened this issue 2 years ago • 1 comments

Describe the bug

Run Antrea Agent on code branch feature/externalnode to check the realization of ANP FQDN rules, and got the following error logs continously

E0613 10:38:36.460903   14114 fqdn.go:696] "DNS exchange failed" err="read udp 10.176.27.243:46116->10.172.40.1:53: i/o timeout"
E0613 10:38:36.460950   14114 fqdn.go:639] "Error syncing FQDN, retrying" err="DNS request failed for at least one of type A or AAAA queries" fqdn="www.163.com""

Having check the related code, the error is output with this logic,

	query := func(m *dns.Msg) (*dns.Msg, error) {
		r, _, err := dnsClient.ExchangeContext(ctx, m, f.dnsServerAddr)
		if err != nil {
			klog.ErrorS(err, "DNS exchange failed")
			return nil, err
		}
		return r, nil
	}

Capturing the packets on the OVS uplink and host internal port ( the port pair is added on ExternalNode feature), I got that the DNS response is received from the uplink but does not output to the host internal port, hence the dnsClient used in the fqdn module is not possible to read the response.

Checked the OpenFlow entry, it seems the packet is intercepted by this one to send to Antrea Agent:

cookie=0x1020000000000, duration=1763.772s, table=AntreaPolicyIngressRule, n_packets=266, n_bytes=64781, priority=64991,conj_id=1 actions=load:0x8->NXM_NX_REG0[13..17],controller(reason=no_match,max_len=128,id=9187)

Try to add additional action on the flow entry to output the packet to the corresponding OVS internal port, the error log is gone, and the FQDN function still works.

To Reproduce

  1. Deploy Antrea, with ANP feature enabled
  2. Run Antrea Agent with ExternalNode feature enabled
  3. Observing the output of Antrea Agent logs

Expected

The above error log is not printed.

Actual behavior

The error logs are printed continuously.

Versions:

Antrea: feature/externalnode

Additional context

wenyingd avatar Jun 13 '22 11:06 wenyingd

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment, or this will be closed in 90 days

github-actions[bot] avatar Sep 12 '22 00:09 github-actions[bot]