DNS lookup fails with SocketException

[jonmill]

Target name(s)

ESP32_REV0

Firmware version

1.8.0.6

Was working before? On which version?

N/A

Device capabilities

ESP32 (ESP32-D0WDQ6 (revision 1))
Features WiFi, BT, Dual Core, 240MHz, VRef calibration in efuse, Coding Scheme None
Flash size 4MB unknown from (manufacturer 0x216 device 0x16406)
PSRAM: not available
Crystal 40MHz

Description

I'm getting a SocketException during Dns.GetHostEntry that is traced to the NativeSocket::getaddrinfo call (call stack below). The DNS entry does exist and can be retrieved using this call in a normal .NET Core app.

++++ Exception System.Net.Sockets.SocketException - CLR_E_FAIL (1) ++++
    ++++ Message: 
    ++++ System.Net.Sockets.NativeSocket::getaddrinfo [IP: 0000] ++++
    ++++ System.Net.Dns::GetHostEntry [IP: 0008] ++++
    ++++ Testing.Program::Main [IP: 013f] ++++
Exception thrown: 'System.Net.Sockets.SocketException' in System.Net.dll
An unhandled exception of type 'System.Net.Sockets.SocketException' occurred in System.Net.dll

WiFi is connected with a valid IP and DateTime and the address being looked up is infrastructure.local. A .NET Core app can lookup the address fine, but not the ESP32. There is no trace of the request hitting the DNS server

bool connected = nanoFramework.Networking.WiFiNetworkHelper.ConnectDhcp(
                WIFI_NAME,
                WIFI_PWD,
                WiFiReconnectionKind.Automatic,
                requiresDateTime: true,
                wifiAdapter: 0);
            if (connected == false)
            {
                throw new InvalidOperationException("Could not connect to WiFi");
            }
            else
            {
                Console.WriteLine($"Current Time Post-Connect: {now.Month}/{now.Day}/{now.Year} {now.Hour}:{now.Minute}:{now.Second}");
            }

How to reproduce

1. Have a DNS server (ex pihole) resolve a .local address as an A record to a local server
2. Have an ESP32 attempt DNS resolve for the .local address

Expected behaviour

The DNS entry resolves to an IP address

Screenshots

No response

Aditional information

No response

[jonmill]

Looks like this is a problem with .local domains where the mDNS lookup is not leaking out to the DNS server

[Ellerbach]

If my understanding is correct, mDNS is a bit different than DNS.
It seems to have issue in most platforms. Found an "old" discussion on Stackoverflow: https://stackoverflow.com/questions/10244117/how-can-i-find-the-ip-address-of-a-host-using-mdns/35853322#35853322
It does point out issues with ESP implementation.

And mDNS packet structure described here: https://en.wikipedia.org/wiki/Multicast_DNS

[torbacz]

Not long ago, I was working with mDNS. I've tried to set up HomeAssitant server with mDNS on board. Everything works just fine from Windows machine, but Android phone was not able to resolve address. I've switched to normal DNS and it's working since.
Maybe it's not nanoFramework issue? Maybe it's ESP issue?

BTW. I doubt that mDSN lookup will be visible in normal DNS server, from wikipedia
When an mDNS client needs to resolve a hostname, it sends an [IP multicast](https://en.wikipedia.org/wiki/IP_multicast) query message that asks the host having that name to identify itself. That target machine then multicasts a message that includes its IP address.

[Ellerbach]

Maybe it's not nanoFramework issue? Maybe it's ESP issue?

it is most likely an ESP issue regarding the comments from other sources. Still, let's keep this issue open up to the point the root cause is properly identified or that there is no interest.
And thanks for the additional info as well!

[DaveSchmid]

I've had exactly the same problem. It has nothing to do with .local. It also happens with .com. I believe it is due to a DNS request timeout that is too short. And it depends on the hardware. I have several ESP32 Devkit-C V4 boards. The same code works on some, throws an exception on others.

[josesimoes]

Currently we don't expose any setting related with the DNS execution timeout. Nor are we setting anything about this from the defaults in lwIP.

[DaveSchmid]

this idea with timeout is just a feeling of mine. A DNS query can take a certain amount of time depending on the DNS server. But the exception comes very quickly, too quickly. I would like to record the network traffic but currently it works without an error.

[josesimoes]

I can't reproduce this... always get the resolved address if it exists and an exception if it doesn't.

Worth pointing out that a SocketException it's the expected outcome when pretty much anything fails, so it may be hard to trace if this is because of a timeout or an issue with the resolution failing for a particular address. Until we have a consistent way of reproducing this it will be hard to even start looking at what could be wrong.
Considering that we have no other reports about this, I'm considering filling this under "can't reproduce: won't fix"...

https://learn.microsoft.com/en-us/dotnet/api/system.net.dns.gethostentry?view=net-7.0#system-net-dns-gethostentry(system-string)

[josesimoes]

No further details provided. Closing as resolved. Feel free to reopen if this is still happening.