-
Notifications
You must be signed in to change notification settings - Fork 36
Conversation
I'm on vacation next two weeks, so just to keep priorities straight - either this PR, or #547, need to be applied, so that we're 100% sure we don't have autocheckpoint enabled on connections, and only control the checkpoints ourselves. More importantly, in the fiber that performs the checkpoint, we need to make sure that it's performed from a connection that has our custom WAL methods registered, so that we call the sqld/sqld/src/replication/primary/logger.rs Line 171 in bb27f19
|
a20241f
to
5ae39a2
Compare
afc814a
to
eac46d2
Compare
eac46d2
to
495d77c
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks like #547 is applied and turns off autocheckpointing where applicable
495d77c
to
c9126e1
Compare
Build on top of #547
This PR introduces async DB file snapshotting. Basically when checkpoint happens, we trigger asynchronous DB file backup. If the next checkpoint happens, it first waits until previous snapshot upload completes before continuing.