Any advantage to using the other mysql tabletypes? (bdb, innodb) with RT?

RT Users
1

Just curious. I’m seeing a lot of “Copying tmp table to disk” that hold up my
RT install for 2-3 minutes at a time. WHile I know specifically
what causes it and can train my users to not run that set of queries, I"m
just wondering if using different table types and transaction stuff might
let mysql run the other pending queries from other RT users…

2

“JM” == Jaye Mathisen mrcpu@internetcds.com writes:

JM> just wondering if using different table types and transaction stuff might
JM> let mysql run the other pending queries from other RT users…

Not that this solves your problem, but my experience tells me that
this is where Postgres really shines – the concurrency model with
transactions is really smart – it does row level locking rather than
table-level, and only locks the rows you’re gonna update (if you tell
it).

I switched an app from mysql to postgres and tweaked it to give
postgres the necessary hints and the concurrency went way up.

Vivek Khera, Ph.D. Khera Communications, Inc.
Internet: khera@kciLink.com Rockville, MD +1-240-453-8497
AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/

3

I have the same problem (although i’ve given mysql enough memory so
it doesn’t hit the disk. and trust me, You cannot train
users not to run certain queries :slight_smile: If the option is there they
will use it, I had to take it away :slight_smile:

Mat.

4

Vivek Khera khera@kcilink.com writes:

“JM” == Jaye Mathisen mrcpu@internetcds.com writes:

JM> just wondering if using different table types and transaction stuff might
JM> let mysql run the other pending queries from other RT users…

Not that this solves your problem, but my experience tells me that
this is where Postgres really shines – the concurrency model with
transactions is really smart – it does row level locking rather than
table-level, and only locks the rows you’re gonna update (if you tell
it).

PostgreSQL does not implement row-level locking, but multi-version
concurrency control, right?

(The effect is about the same, though: parallel read/write access is
more or less independent.)

Florian Weimer Weimer@CERT.Uni-Stuttgart.DE
University of Stuttgart http://CERT.Uni-Stuttgart.DE/people/fw/
RUS-CERT fax +49-711-685-5898

5

“FW” == Florian Weimer Weimer@CERT.Uni-Stuttgart.DE writes:

FW> PostgreSQL does not implement row-level locking, but multi-version
FW> concurrency control, right?

Nope. You can get row-level locking when you ask for it.

FW> (The effect is about the same, though: parallel read/write access is
FW> more or less independent.)

That too.

Vivek Khera, Ph.D. Khera Communications, Inc.
Internet: khera@kciLink.com Rockville, MD +1-240-453-8497
AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/