Details
-
Task
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
None
Description
oltp_update_index.lua benchmark commits mini transaction 5 times. This causes log_sys->mutex contention.
First two mini transactions (write of the first undo log record and the creation of the undo log header) can be combined.
This change was verified to give ~10% better throughput.
Attachments
Issue Links
- causes
-
MDEV-22770 trx_undo_report_rename() fails to release page latches
-
- Closed
-
-
-
- Closed
-
Activity
| Field | Original Value | New Value |
|---|---|---|
| Epic Link | MDEV-14442 [ 64369 ] |
| Status | Open [ 1 ] | Confirmed [ 10101 ] |
| Status | Confirmed [ 10101 ] | In Progress [ 3 ] |
| Summary | Combine the write of the first undo log record and the creation of the undo log header into a single mini-transaction | Reduce the overhead of writing undo log records |
| issue.field.resolutiondate | 2018-01-29 18:29:25.0 | 2018-01-29 18:29:25.085 |
| Fix Version/s | 10.3.5 [ 22905 ] | |
| Fix Version/s | 10.3 [ 22126 ] | |
| Resolution | Fixed [ 1 ] | |
| Status | In Progress [ 3 ] | Closed [ 6 ] |
| Link |
This issue causes |
| Link |
This issue causes |
| Workflow | MariaDB v3 [ 85163 ] | MariaDB v4 [ 133454 ] |
I had some doubts whether this is safe to do in all cases or if some deadlocks could be caused due to making changes to some pages after a page allocation. (There is a rule about that in operations that affect the change buffer; see the comment in the function btr_compress() which BTW is involved in the deadlock that has been reported in
MDEV-14637.)It turns out that whenever there is not enough space for the undo log record in the undo log page, InnoDB will commit the mini-transaction and start a new one. It also turns out that the undo log record write would be attempted on the same page that was already written to within trx_undo_assign_undo(). So, merging the mini-transactions should be completely safe.