Bug #106770 | Document MySQL replication protocol change 8.0.27 -> 8.0.28 | ||
---|---|---|---|
Submitted: | 18 Mar 2022 8:46 | Modified: | 24 Mar 2022 9:59 |
Reporter: | Simon Mudd (OCA) | Email Updates: | |
Status: | Not a Bug | Impact on me: | |
Category: | MySQL Server: Documentation | Severity: | S4 (Feature request) |
Version: | 8.0.28 | OS: | Any |
Assigned to: | Jon Stephens | CPU Architecture: | Any |
Tags: | 8.0.27, 8.0.28, casting, RBR, replication, virtual column |
[18 Mar 2022 8:46]
Simon Mudd
[18 Mar 2022 9:48]
MySQL Verification Team
Testcase from https://bugs.mysql.com/bug.php?id=106437 can be used. You'll see due to invalid CAST in 8.0.28 column @3 has NULL in the binary log.
[18 Mar 2022 13:35]
Jon Stephens
I've been following this issue, assigning bug report to myself.
[18 Mar 2022 19:37]
Jon Stephens
Possibly of interest for developer docs, but does not seem relevant to end user documentation.
[24 Mar 2022 9:59]
Simon Mudd
> Possibly of interest for developer docs, but does not seem relevant to end user documentation. Indeed, from the normal user perspective this should be invisible and transparent. However, these days with CDC type imports often happening from binlogs describing the binglog flows and content and behaviour becomes quite important. A lot of software is often not updated and may not support the latest changes in the binlog streams and I think that may partly be because until you read the source code you can't tell what's changed or when it changed so writing code to read binlogs for processing outside of the traditional master -> replica setup may be more common. I have seen such issues with compressed binlogs not being supported even though it's a feature released some time ago. For any developer of such CDC type software in the end they ask the DBAs to turn off the feature as they simply can not handle it which completely defeats the purpose and reason it was enabled on the servers. So while this specific request is for add a comment to the release notes (technical developer facing) generally having more specific documentation on the binlog format and how it should be consumed would be very interesting.