Remove experimental label of innodb_locking_fake_changes

Bug #1154956 reported by Stewart Smith
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Percona Server moved to https://jira.percona.com/projects/PS
Status tracked in 5.7
5.1
Won't Fix
Undecided
Unassigned
5.5
Won't Fix
Low
Unassigned
5.6
Won't Fix
Low
Unassigned
5.7
Invalid
Undecided
Unassigned

Bug Description

Introduced in 5.5.28-29.2, this option to fake changes makes the fake transactions take no locks.

   When this variable is set to ``FALSE``, it makes fake transactions not to take any row locks. This feature was implemented because, although fake change transactions downgrade the requested exclusive (X) row locks to shared (S) locks, these S locks prevent X locks from being taken and block the real changes. However, this option is not safe to set to FALSE by default, because the fake changes implementation is not ready for lock-less operation for all workloads. Namely, if a real transaction will remove a row that a fake transaction is doing a secondary index maintenance for, the latter will fail. This option is considered experimental and might be removed in the future if lockless operation mode fixes are implemented.

This bug is to track removing the experimental label, calling the feature fully baked, or to remove the feature.

Revision history for this message
Laurynas Biveinis (laurynas-biveinis) wrote :

I think we should keep the experimental label. Fake changes should work with the least required locking automatically, and once this is implemented we should remove the option, which is a band-aid until then.

Revision history for this message
Stewart Smith (stewart) wrote : Re: [Bug 1154956] Re: Remove experimental label of innodb_locking_fake_changes

Laurynas Biveinis <email address hidden> writes:
> I think we should keep the experimental label. Fake changes should work
> with the least required locking automatically, and once this is
> implemented we should remove the option, which is a band-aid until
> then.

I agree that for the time being we should keep it. Let's keep this bug
open to track it though.

--
Stewart Smith

tags: added: fake-changes
tags: added: xtradb
Revision history for this message
Shahriyar Rzayev (rzayev-sehriyar) wrote :

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-2910

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.