CoreDNS issue under load

[mahdiadnan]

Hi,
I'm using CoreDNS with alternate plugin as high available DNS alongside Unbound DNS. the server is getting around 60K/s
My config is:

.:53 {
    debug
    loop
    bind 10.x.x.1
    bind 10.x.x.2
    bind 37.x.x.3
    bind 127.0.0.1

    forward . 127.0.0.1:9053

    alternate original SERVFAIL,REFUSED . 37.x.x.4 37.x.x.5

    reload 30s

    cache 300 {
        success 50000
        denial 25000
    }

    prometheus localhost:9253
}

If I kill the local Unbound instance :9053 CoreDNS displays messages like the following:

[DEBUG] Wrong reply for id: 42177, img.like.video. 1
debug: 000000 a4 c1 01 00 00 01 00 00 00 00 00 01 03 69 6d 67
debug: 000010 04 6c 69 6b 65 05 76 69 64 65 6f 00 00 01 00 01
debug: 000020 00 00 29 08 00 00 00 80 00 00 00
debug: 00002b
[DEBUG] Wrong reply for id: 52794, e1329.g.akamaiedge.net. 1
debug: 000000 ce 3a 01 00 00 01 00 00 00 00 00 01 05 65 31 33
debug: 000010 32 39 01 67 0a 61 6b 61 6d 61 69 65 64 67 65 03
debug: 000020 6e 65 74 00 00 01 00 01 00 00 29 08 00 00 00 80
debug: 000030 00 00 00
debug: 000033
[DEBUG] Wrong reply for id: 24443, me-du.shadow.igamecj.com. 1
debug: 000000 5f 7b 01 00 00 01 00 00 00 00 00 01 05 6d 65 2d
debug: 000010 64 75 06 73 68 61 64 6f 77 07 69 67 61 6d 65 63

Then the DNS queries start to get dropped, and after a minute or so, the container of CoreDNS exits.
The logs I shared were the last messages from CoreDNS before it crash.

Any hints on this?

EDIT: CoreDNS version is 1.8.6 and 1.8.7 same issue.

[zouyee]

any progress？