net: Go's DNS resolver retry AAAA request to next nameserver even A response has records #64783

WeiZhixiong · 2023-12-18T15:15:20Z

Go version

go version go1.21.5 linux/amd64

What operating system and processor architecture are you using (`go env`)?

GO111MODULE=''
GOARCH='amd64'
GOBIN=''
GOCACHE='/root/.cache/go-build'
GOENV='/root/.config/go/env'
GOEXE=''
GOEXPERIMENT=''
GOFLAGS=''
GOHOSTARCH='amd64'
GOHOSTOS='linux'
GOINSECURE=''
GOMODCACHE='/root/go/pkg/mod'
GONOPROXY=''
GONOSUMDB=''
GOOS='linux'
GOPATH='/root/go'
GOPRIVATE=''
GOPROXY='https://proxy.golang.org,direct'
GOROOT='/opt/go'
GOSUMDB='sum.golang.org'
GOTMPDIR=''
GOTOOLCHAIN='auto'
GOTOOLDIR='/opt/go/pkg/tool/linux_amd64'
GOVCS=''
GOVERSION='go1.21.5'
GCCGO='gccgo'
GOAMD64='v1'
AR='ar'
CC='gcc'
CXX='g++'
CGO_ENABLED='1'
GOMOD='/dev/null'
GOWORK=''
CGO_CFLAGS='-O2 -g'
CGO_CPPFLAGS=''
CGO_CXXFLAGS='-O2 -g'
CGO_FFLAGS='-O2 -g'
CGO_LDFLAGS='-O2 -g'
PKG_CONFIG='pkg-config'
GOGCCFLAGS='-fPIC -m64 -pthread -Wl,--no-gc-sections -fmessage-length=0 -fdebug-prefix-map=/tmp/go-build403659500=/tmp/go-build -gno-record-gcc-switches'

What did you do?

Problem

here is /etc/resolv.conf

nameserver 10.154.200.27
nameserver 10.66.200.202
options timeout:1 attempts:1

we have some private domain, it's only has A records, no AAAA, I use net.Dial("tcp", "domain:80") start a connect, the Go's DNS resolver retry AAAA request to next nameserver even A response has records.

if use cgo, no retry. like go build -tags 'netcgo' same program, the result is:

Reason

we debug find hit the errLameReferral error. but in glibc, it's different.

func checkHeader(p *dnsmessage.Parser, h dnsmessage.Header) error {
    ...
    // libresolv continues to the next server when it receives
    // an invalid referral response. See golang.org/issue/15434.
    if h.RCode == dnsmessage.RCodeSuccess && !h.Authoritative && !h.RecursionAvailable && err == dnsmessage.ErrSectionDone {
        return errLameReferral
    }
    ....
}

func (r *Resolver) tryOneName(ctx context.Context, cfg *dnsConfig, name string, qtype dnsmessage.Type) (dnsmessage.Parser, string, error) {
    ...
    for i := 0; i < cfg.attempts; i++ {
        for j := uint32(0); j < sLen; j++ {
            server := cfg.servers[(serverOffset+j)%sLen]
 
            p, h, err := r.exchange(ctx, server, q, cfg.timeout, cfg.useTCP, cfg.trustAD)
            ...
            if err := checkHeader(&p, h); err != nil {
                dnsErr := &DNSError{
                    Err:    err.Error(),
                    Name:   name,
                    Server: server,
                }
            ...
            }
    ...
}

in glibc A or AAAA anyone has records, no retry.

next_ns:
    if (recvresp1 || (buf2 != NULL && recvresp2)) {
      *resplen2 = 0;
      return resplen;
    }
 
...
 
if (anhp->rcode == NOERROR && anhp->ancount == 0
    && anhp->aa == 0 && anhp->ra == 0 && anhp->arcount == 0) {
    goto next_ns;
}

What did you expect to see?

if domain A type response has records, no AAAA retry, if domain AAAA type response has records, no A retry.

What did you see instead?

if domain A type response has records, AAAA has no record, AAAA request retry to next nameserver.

The text was updated successfully, but these errors were encountered:

mateusz834 · 2023-12-18T15:33:32Z

Does CL 550435 help with this?
Can you share the output of these commands?

dig @10.154.200.27 abtest.bilibili.co A
dig @10.154.200.27 abtest.bilibili.co AAAA
dig @10.66.200.202 abtest.bilibili.co A
dig @10.66.200.202 abtest.bilibili.co AAAA

WeiZhixiong · 2023-12-18T15:50:19Z

Does CL 550435 help with this?

Yes, but there has two problem, CL 550435 only fix one.

dig @10.154.200.27 abtest.bilibili.co A

; <<>> DiG 9.10.3-P4-Debian <<>> @10.154.200.27 abtest.bilibili.co A
; (1 server found)
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 27383
;; flags: qr rd; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1
;; WARNING: recursion requested but not available

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;abtest.bilibili.co.		IN	A

;; ANSWER SECTION:
abtest.bilibili.co.	80	IN	A	10.154.200.12

;; Query time: 0 msec
;; SERVER: 10.154.200.27#53(10.154.200.27)
;; WHEN: Mon Dec 18 23:42:46 HKT 2023
;; MSG SIZE  rcvd: 63

dig @10.154.200.27 abtest.bilibili.co AAAA

; <<>> DiG 9.10.3-P4-Debian <<>> @10.154.200.27 abtest.bilibili.co AAAA
; (1 server found)
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 64271
;; flags: qr rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1
;; WARNING: recursion requested but not available

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;abtest.bilibili.co.		IN	AAAA

;; AUTHORITY SECTION:
bilibili.co.		542	IN	SOA	shylf-dns-03. hostmaster.bilibili.co. 2023121803 16384 900 1048576 2560

;; Query time: 0 msec
;; SERVER: 10.154.200.27#53(10.154.200.27)
;; WHEN: Mon Dec 18 23:45:05 HKT 2023
;; MSG SIZE  rcvd: 106

dig @10.66.200.202 abtest.bilibili.co A

; <<>> DiG 9.10.3-P4-Debian <<>> @10.66.200.202 abtest.bilibili.co A
; (1 server found)
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 50743
;; flags: qr rd; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1
;; WARNING: recursion requested but not available

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;abtest.bilibili.co.		IN	A

;; ANSWER SECTION:
abtest.bilibili.co.	120	IN	A	10.154.200.12

;; Query time: 1 msec
;; SERVER: 10.66.200.202#53(10.66.200.202)
;; WHEN: Mon Dec 18 23:45:36 HKT 2023
;; MSG SIZE  rcvd: 63

dig @10.66.200.202 abtest.bilibili.co AAAA

; <<>> DiG 9.10.3-P4-Debian <<>> @10.66.200.202 abtest.bilibili.co AAAA
; (1 server found)
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 62038
;; flags: qr rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1
;; WARNING: recursion requested but not available

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;abtest.bilibili.co.		IN	AAAA

;; AUTHORITY SECTION:
bilibili.co.		600	IN	SOA	shylf-dns-03. hostmaster.bilibili.co. 2023121803 16384 900 1048576 2560

;; Query time: 1 msec
;; SERVER: 10.66.200.202#53(10.66.200.202)
;; WHEN: Mon Dec 18 23:46:10 HKT 2023
;; MSG SIZE  rcvd: 106

mateusz834 · 2023-12-18T17:36:03Z

First of all your resolvers are not recursive. All resolvers in /etc/resolv.conf should be recursive.

I suspect that CL 550435 helps, because the responses have an EDNS(0) resource (arcount != 0), so that the errLameReferral is not returned.

Yes, but there has two problem, CL 550435 only fix one.

What do you mean that there are two problems? What it fixes? What it did not fix?

thanm · 2023-12-18T18:52:17Z

@ianlancetaylor @neild per owners

seankhliao · 2023-12-18T23:05:50Z

Duplicate of #57697

gopherbot · 2024-01-12T07:40:18Z

Change https://go.dev/cl/550435 mentions this issue: net: Prevent unintended retries upon receiving an empty answer response from the DNS server.

…se from the DNS server. CL https://golang.org/cl/37879 migrates DNS message parsing to the golang.org/x/net/dns/dnsmessage package. However, during the modification of the "lame referral" error check introduced by CL https://golang.org/cl/22428, a condition was overlooked. This omission results in unexpected retries when a DNS server returns an empty response (not an invalid response, but one that includes an additional section). Fixes #57697 Fixes #64783 Change-Id: I203896aa2902c305569005c1712fd2f9f13a9b6b Reviewed-on: https://go-review.googlesource.com/c/go/+/550435 LUCI-TryBot-Result: Go LUCI <[email protected]> Commit-Queue: Ian Lance Taylor <[email protected]> Auto-Submit: Ian Lance Taylor <[email protected]> Reviewed-by: Damien Neil <[email protected]> Reviewed-by: Bryan Mills <[email protected]> Reviewed-by: Dexter Ouyang <[email protected]>

gopherbot · 2024-02-20T03:19:39Z

Change https://go.dev/cl/565295 mentions this issue: net: prevent unintended retries upon receiving an empty answer response from the DNS server.

gopherbot · 2024-02-20T03:22:53Z

Change https://go.dev/cl/565296 mentions this issue: net: prevent unintended retries upon receiving an empty answer response from the DNS server.

thanm added the NeedsInvestigation Someone must examine and confirm this is a valid issue and not a duplicate of an existing one. label Dec 18, 2023

seankhliao marked this as a duplicate of #57697 Dec 18, 2023

seankhliao closed this as not planned Won't fix, can't repro, duplicate, stale Dec 18, 2023

mateusz834 mentioned this issue Dec 19, 2023

net/http: Client.Do blocks on DNS AAAA record until timeout even after A record succeeds #57697

Closed

haunt98 mentioned this issue Jul 7, 2024

Draft make-go-great/dnsurprise-go#1

Open

golang locked and limited conversation to collaborators Feb 19, 2025

gopherbot added the FrozenDueToAge label Feb 19, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

net: Go's DNS resolver retry AAAA request to next nameserver even A response has records #64783

net: Go's DNS resolver retry AAAA request to next nameserver even A response has records #64783

WeiZhixiong commented Dec 18, 2023

mateusz834 commented Dec 18, 2023

WeiZhixiong commented Dec 18, 2023

mateusz834 commented Dec 18, 2023

thanm commented Dec 18, 2023

seankhliao commented Dec 18, 2023

gopherbot commented Jan 12, 2024

gopherbot commented Feb 20, 2024

gopherbot commented Feb 20, 2024

net: Go's DNS resolver retry AAAA request to next nameserver even A response has records #64783

net: Go's DNS resolver retry AAAA request to next nameserver even A response has records #64783

Comments

WeiZhixiong commented Dec 18, 2023

Go version

What operating system and processor architecture are you using (go env)?

What did you do?

Problem

Reason

What did you expect to see?

What did you see instead?

mateusz834 commented Dec 18, 2023

WeiZhixiong commented Dec 18, 2023

mateusz834 commented Dec 18, 2023

thanm commented Dec 18, 2023

seankhliao commented Dec 18, 2023

gopherbot commented Jan 12, 2024

gopherbot commented Feb 20, 2024

gopherbot commented Feb 20, 2024

What operating system and processor architecture are you using (`go env`)?