[Tickets #12476] Re: Database running mad after update to 5.1.1
noreply at bugs.horde.org
noreply at bugs.horde.org
Mon Jul 22 08:47:37 UTC 2013
DO NOT REPLY TO THIS MESSAGE. THIS EMAIL ADDRESS IS NOT MONITORED.
Ticket URL: http://bugs.horde.org/ticket/12476
------------------------------------------------------------------------------
Ticket | 12476
Updated By | marth at tsvschlieben.de
Summary | Database running mad after update to 5.1.1
Queue | Horde Framework Packages
Version | Git master
Type | Bug
State | Resolved
Priority | 2. Medium
Milestone |
Patch |
Owners | Michael Rubinsky
------------------------------------------------------------------------------
marth at tsvschlieben.de (2013-07-22 08:47) wrote:
Thanks for the feedback - here is what I found out. I suggest to
reopen the ticket. (see below)
> Since my history data set, even on production is not anywhere near
> as large as your data set, can you see if adding the following index
> improves this for you? EXPLAIN now shows that it is using the index
> so it should improve...
>
> ALTER TABLE horde_histories ADD INDEX (history_modseq, object_uid);
It's now using the index - but the query itself becomes even slower:
Before:
1.13 sec
After:
2.30 sec
(Indeed EXPLAIN shows that the index history_modseq is used)
> If not, the next attempt would be to use something like:
>
> SELECT history_modseq FROM horde_histories WHERE object_uid LIKE
> 'turba:%' ORDER BY history_modseq DESC LIMIT 1;
0.00 sec (!!!)
So this solves the speed issue.
> but I'd be wary of doing this since I doubt it would perform the
> same from one RDBMS to another.
It think it's worth a test because the performance raise in MySQL is
DRAMATICALLY. The key length (according MySQL explain) that is used is
now 4 instead of 767. Maybe this could be the reason for the gain.
More information about the bugs
mailing list