tree b2cdba07375cb30aea1762d5eff5c79f76ce85f9
parent c1788c3db37ca4d2ff2d55e3d3ca6b667575a95b
author Guohao Li <gehafearless@apache.org> 1710300751 +0800
committer GitHub <noreply@github.com> 1710300751 +0800
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJl8R5PCRC1aQ7uu5UhlAAANCkQAKgXb47pldntAdUvI1sSYgC9
 aAZln/b0VmpFSi2rnBSSpncUvoDRnPP0cDU2hkw0zXiPqVjlZ8ds+AQWNjYeBTnx
 hyD4szDfk7cOd+CrrqfPJMd/kSSy3cpggPyYeuHEH+EsNxXxp47VRv54ym3veKbY
 2RrFDZgCMM88oW6tw8FeWOGfWhF1AhBSIIR7GEGF9ta5jtq45aQTaKpCffvgSFay
 KzhN2zVdLo8qs91p03KUxOmige65HWlOMPb3ym+MAld01D5wETTZS9+ZRuPySxiS
 Bcg7rg5OdMuXNJELAkSCul3/0a/eaLz3t2QRFOAsASMQ+U1cNlQvHQ13r+mGp9T9
 Lf6lNS3vdwBaHwwPawuZeRLj3/P9tWSsQ0CgYAdD432P3B4spNOqPW85moly2K8U
 g1H5Jx9khk5YLqWNseLur0glBiAFnxmtG5kwWWMRWw1Aw9GE7qWvp7xDiiLuIQiJ
 hTG6UcWVEkBRh0UbNbT3yxMXwxWUEdhvGVaAkJukKHeRqob9v4C6m0Rph3tWhr9z
 9zmGow4YPU8dEnu8HGf1EPzW95sKCFbkT/1V4N9RYlyv+XM5HkOwdfhweaU9IZ7g
 L1XPtHF1BpfoM8uEtiKPCrC2uMCO418meConHoZ74zBCr5px6cNOTvZGwAd3yx6r
 U3mzJERL1Cy7n4jDA+Ts
 =B2e8
 -----END PGP SIGNATURE-----
 

feat(FQDN): replace rpc_address to host_prot for function & add host_port on thrift struct (#1658)

https://github.com/apache/incubator-pegasus/issues/1659

1. Add a new host_port type field if rpc_address field exists in IDL structures.
2. Use both rpc_address and the optional host_port fields in C++ code.
3. Use host_port to communicate with remote nodes firstly.

TODO:
- Make sure everything works well after upgrading from old versions without any config changes
- Make sure old version client is able to access the new version cluster
- Make sure new version client is able to access the old version cluster

NOTE: 
It's not able to upgrade Pegasus cluster and clients smoothly to the new version with FQDN enabled, it's needed to follow the steps if you want to enable the FQDN feature:
1. Stop the cluster and clients
2. Set the new configs (using hostnames instead of IPs)
4. Rebuild the metadata of the cluster (See https://pegasus.apache.org/zh/administration/meta-recovery)
5. Start the whole cluster and application clients

NOTE: Do not support rolling restart servers in the cluster with FQDN related configs changed