I'm running a new instance of the Apache Ignite 3 beta on Windows and am hoping someone might recognize the error reading the MAGIC_BYTES I'm seeing trying to run the examples.
The cluster starts successfully and I can connect via the CLI; e.g.: 'node status' shows [name: defaultNode, state: started]
However, when I attempt to run any of the examples, such as SqlJdbcExample, it fails in ClientMessageDecoder.readMagic(). In there it is attempting to read the 4 MAGIC_BYTES (representing the 4 ASCII characters 'INGI').
What I see instead are the bytes 0x01, 0x00, 0x03, 0x00
This ultimately results in an IgniteException:
IGN-CMN-65535 TraceId:xxxx Invalid magic header in thin client connection. Expected 'IGNI', but was '▯▯'.
(Note: in debug, if I read more bytes out of the buffer I can see those 4 bytes are followed by an ASCII newline character (0x12) then the ASCII 'defaultNode'.)
In the SqlJdbcExample when it is initializing the JDBC connection, I can see that is has called socket.send() with those MAGIC_BYTES IN TcpClientChannel.handshakeReq().
I am running the examples with no change to any configuration files and have set up the environment as per the documentation.
Set up Apache Ignite 3 beta and ran samples. They failed with
IgniteException: IGN-CMN-65535 TraceId:xxxx Invalid magic header in thin client connection. Expected 'IGNI', but was '▯▯'.
Verified as best I can that everything is configured correctly, but can't determine why I'm not picking up these bytes.
Ignite examples use port 10800 by default. Looks like something else is using that port on your machine, so Ignite server is on a different port.
Look into the log file in ignite3-db-3.0.0-beta1 directory (ignite3db-0.log file name), and search for Thin client protocol started successfully[port=10800] line. The port is likely different. Copy the port number and update the connection string in the example accordingly