[ale] MySQL slave startup from crash

[Chris Fowler]

I'm doing some testing in a container on my workstation where the container is a MySQL 8 slave and the maser is a MySQL 8 master at Digital Ocean.
I'm not using any container management packages.  I'm only using unshare to create namespaces for each container.   SSHD is pid 1 of each container.   Stopping these containers simply terminates SSHD of each and all pids of those namespaces are terminated as if by signal 9.   They simply vanish immediately.

Does MySQL 8 slave replay its binlog after crash?  When I start up the container, the slave will stop replication with duplicate key errors.  As if it did not retain its log position from the previous run and simply replayed the binlog.

In production, slaves will not terminate in this manner.  I'm now just focusing on this behavior for instances where the system the slave runs on crashes.
Is there a way to start the slave back up without reinitializing off the mater?

Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.ale.org/pipermail/ale/attachments/20211222/489e5bf9/attachment.htm>