1700309 - Firefox can’t establish a connection to the server at wss://analysis.chess.com/.

Status: RESOLVED FIXED

(Core :: Networking: WebSockets, defect, P1)

Description

[Gian-Carlo Pascutto [:gcp]]

Attachment: log.txt-main.431134.moz_log

STR:

1. Play a game on chess.com
2. Run the "Game Report" feature (this may require a paid account)

Expected:

Game analysis runs and appears at the end.

Actual:

Intermittently, you get the message:

"We are having some trouble connecting to the server right now... Try again when you have a better connection."

Devtools shows:

Firefox can’t establish a connection to the server at wss://analysis.chess.com/. analysis.client.757676fd.js:2:5303396
Socket is closed. Reconnect will be attempted in NaN seconds. analysis.client.757676fd.js:2:5315492
Uncaught (in promise) undefined

Networking log is attached.

The problem is intermittent on different machines on different operating systems. I tried switching profiles, bisecting, Fission and some experimental prefs on/off but it occurs with all configurations, and intermittently enough that bisecting won't work.

Comment 1

[Kershaw Chang [:kershaw]]

I took a quick look at the log, but the log doesn't have the information about websocket.
So, could you try to capture the log again? This time, please add nsWebSocket:5 to MOZ_LOG module. Thanks.

Comment 2

[Gian-Carlo Pascutto [:gcp]]

Attachment: log.txt-main.14820.moz_log.gz

Comment 3

[Kershaw Chang [:kershaw]]

Thanks for the log. It's really helpful.
So the problem is that we failed to use the IP hint address because nsIDNSService::RESOLVE_BYPASS_CACHE is set. At here, AsyncResolveNative returns an error immediately, so the connection is failed. I think we should retry when RESOLVE_IP_HINT is set.

2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp Creating DnsAndConnectSocket [this=1fc44834480 trans=1fc4ab56800 ent=analysis.chess.com key=.S......[tlsflags0x00000000]analysis.chess.com:443 <ROUTE-via analysis.chess.com.cdn.cloudflare.net:443> {NPN-TOKEN h2}^partitionKey=%28https%2Cchess.com%29]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp DnsAndConnectSocket::SetupDnsFlags [this=1fc44834480] 
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver Resolving host [analysis.chess.com.cdn.cloudflare.net]<^partitionKey=%28https%2Cchess.com%29> type 0. [this=1fc32072820]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver   Offline request for host [analysis.chess.com.cdn.cloudflare.net]; ignoring.
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp Setting Socket to use IP hint address
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp DnsAndConnectSocket::SetupDnsFlags flags=24577 flagsBackup=24609 [this=1fc44834480]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp DnsAndConnectSocket::SetupEvent state=0 event=0
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: V/nsHttp DnsAndConnectSocket::TransportSetup::ResolveHost [this=1fc448344e0 analysis.chess.com.cdn.cloudflare.net bypass cache]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: E/nsHttp nsHttpTransaction::OnSocketStatus [this=1fc4ab56800 status=804b0003 progress=0]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver Resolving host [analysis.chess.com.cdn.cloudflare.net]<^partitionKey=%28https%2Cchess.com%29> - bypassing cache type 0. [this=1fc32072820]
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver   No usable record in cache for host [analysis.chess.com.cdn.cloudflare.net] type 0.
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver NameLookup host:analysis.chess.com.cdn.cloudflare.net af:0
2021-03-25 11:57:11.381000 UTC - [Parent 14820: Socket Thread]: D/nsHostResolver Skip lookup if RES_IP_HINT is set

Comment 4

[Kershaw Chang [:kershaw]]

Attachment: Bug 1700309 - Retry DNS query when RESOLVE_IP_HINT is set

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/cd8358587fa4
Retry DNS query when RESOLVE_IP_HINT is set r=necko-reviewers,dragana

Comment 7

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/cd8358587fa4

Comment 8

[Gian-Carlo Pascutto [:gcp]]

Confirming this is fixed for me.