Details
-
New Feature
-
Status: Open (View Workflow)
-
Minor
-
Resolution: Unresolved
-
None
-
None
Description
.state file caches and indices (MDEV-4991) were optional optimizations for Binary Logs, and (after MDEV-38908) will be optional for Relay Logs, too.
Before adding them to Relay Logs, we should evaluate how much optimization these runtime taxes really bring, and if they should be tuned or enabled separately from Binary Logs.
In particular, with @@relay_log_purge=1 (which is the default), Relay Logs may not be large enough for caches and indices to be worth (unless SQL Threads were sleeping while IO Threads were active).
Otherwise, the code for Binary Logs (e.g., gtid_index.cc) should be (needs to be checked) easily applicable for Relay Logs.
Attachments
Issue Links
- is blocked by
-
MDEV-4698 With GTID replication, relay logs cannot be relied upon while purging binary logs on master
-
- In Progress
-
- split from
-
MDEV-38909
GTID State for Relay Log
-
- Open
-
-
MDEV-38911 GTID List based Relay Log seeking
-
- Open
-