Set compression algorithm for replication, in case of intermediate primary.

[masa213f]

What

In wide-area inter-cluster replication, compression is effective because the communication bandwidth is narrow.

So in the case of intermediate primary, set compression algorithm to the CHANGE MASTER TO statement.
https://dev.mysql.com/doc/refman/8.0/en/change-master-to.html

How

Check the following issues.
If they looks okay, set the compression algorithm.

Related issues

There are some issues with the compression algorithm.

• MySQL Bugs: #103595: using zstd and zlib compression algorithm,master dump thread always be killed https://bugs.mysql.com/bug.php?id=103595

  • Semisync + compressionでプチプチ切れるケース

• MySQL Bugs: #102561: mysqld crashes when using replication with compression https://bugs.mysql.com/bug.php?id=102561

  • アクセスできません。8.0.23でレポートされています

  Server crashes with the following backtrace when attempting to replicate
  from a primary with compression enabled during connection retries. The
  server_extension structure for tracking compression state is stored in
  the Master_info, but it is only initialized once when the I/O thread is
  started. On disconnects/retries, the mysql structure is
  freed/re-initialized, but the server_extension structure is never set
  again.

• MySQL Bugs: #99372: MySQL8.0.19 crashes when slave tries to reconnect to master after master restart https://bugs.mysql.com/99372

  • アクセスできません、8.0.19ですが #102561 と同件っぽく見えますが、 "Fixed in 8.0.20" とされているのでリグレッションなのか微妙に違うのか…

  Ran into an interesting behavior where the replica crashes after it
  tries to re-connect to a master, after a master gets restarted (dump
  thread being killed on master). Because we have master/master setup,
  this ends up crashing the whole cluster

  ---

  How to repeat:

  1. Have a master and a replica, lets say "MasterA" and "ReplicaB"
  2. Stop MySQL in "MasterA"

     systemctl stop mysql.service

  3 On the "ReplicaB" in the log message observe that that MasterA is
  down, and you can see that it tries to connect a couple of times, but
  cannot because MasterA is down
  3. Start MasterA, and observer the crash when ReplicaB (slave threads
  tries to reconnect)
  You can see in the log file below, which is the continuation of the
  log file above

• MySQL Bugs: #93957: slave_compressed_protocol doesn't work with semi-sync replication in MySQL-5.7 https://bugs.mysql.com/bug.php?id=93957

  • 5.7.24ですが大絶賛Verifiedで放置中です
    • MySQL Bugs: #86230: slave_compressed_protocol doesn't work with semi-sync replication in MySQL-5.7 https://bugs.mysql.com/bug.php?id=86230
    • これを5.7.21で直したと言っているのに再発のようです

Checklist

• Finish implementation of the issue
• Test all functions
• Have enough logs to trace activities
• Notify developers of necessary actions