#mysql
29 September 2007
Total 14 pages. You are browsing page 14/14.
First :: Prev :: [...] [10] [11] [12] [13] [14] :: Next :: Last
22:20 <****> Because you'll awnt to do this in production as well as development, it's vital to script the data load
22:20 <****> I recommend saving a tab delimited file and using LOAD DATA; you can commit this tab delimited file into your source code control system
22:21 <****> or we can write a perl script to do it for him
22:21 <****> for a fee
22:21 * Trengo hides
22:21 <****> thoughts on this guys: http://feedblog.org/2007/09/29/using-o_direct-on-linux-and-innodb-to-fix-swap-insanity/
22:23 <****> burtonator: Interesting, I don't see the kernel version there... have you reported this as an issue to the kernel developers or discussed it on LKML?
22:23 <****> MarkR42: no..... I just figured this out now
22:23 <****> it's 2.6.18
22:23 <****> I'll update the post
22:25 <****> im on 32bits only so, no experience on it, at all
22:25 <****> I can't reproduce it, I don't have a 64bit box with 8G of ram handy
22:25 <****> it was easy to reproduce... once I figured out what was going on and could use the blackhole engine
22:25 <****> this is really interesting to know; presumably it would apply equally to myisam with a similar size keybuffer
22:26 <****> oh yeah...
22:26 <****> true
22:26 <****> if you had a HUGE
22:26 <****> db
22:26 <****> http://www.archivist.info/wench
22:26 <****> where it doesn't make sense to cache it
22:26 <****> we use boxes with that much ram at work
22:26 <****> but you still want to use a bunch of ram
22:26 <****> in fact we have one with 32G I think
22:26 <****> --memlock apparently crashes mysql on Linux
22:27 <****> you run mysql as root?
22:27 <****> no.. as 'mysql'
22:27 <****> you can't memlock on linux unless running as root
22:27 <****> (I don't think)
22:27 <****> In any case, it's a bad approach
22:28 <****> yes
22:28 <****> the admin
22:28 <****> should be able to tell the kernel
22:28 <****> how to behave
22:28 <****> this is really just a hack
22:28 <****> if I had a jave daemon
22:28 <****> I can't pass O_DIRECT
22:28 <****> and shouldn't have to
22:28 <****> take it up on LKML
22:28 <****> there may be some subtle tuning thing
22:29 <****> I thought swappiness was the only option I had
22:29 <****> but I'll bring it up on lkml
22:37 <****> how should look sql statement: i have collumens: artist, title, number, i would like to order this by number, bigger is number, first row should be that?
22:38 <****> select number, artist, title from table order by number desc;
22:38 <****> EvilDin: select * from table order by number desc;
22:38 <****> thank you
23:49 <****> hello
23:50 <****> is anybody using FileMaker with MySQL?
23:55 <****> anyone active?
23:55 <****> ?
23:55 <****> alo?
23:55 <****> yes?
23:56 <****> when using innodb, it can create internal hashes for optimization if it has extra memory i have heard
23:56 <****> does this apply only to the innodb_additional_mem_pool_size
23:56 <****> or the innodb_buffer_pool_size as well
23:57 <****> idk Mysql 4 dumbys will help ;]
23:57 <****> j/p
23:57 <****> brb though
23:58 <****> manual only states InnoDB also automatically create hash indexes in RAM if it recognizes specific common query patterns.
--- Log closed Sun Sep 30 00:00:27 2007
Total 14 pages. You are browsing page 14/14.
First :: Prev :: [...] [10] [11] [12] [13] [14] :: Next :: Last