| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Comments:
<spiderworm_work> i agree and have no idea why they set it up like that
<inviso> then just change the script that builds the indexes and be done with it
<spiderworm_work> i guess
<spiderworm_work> we'll give it a run... the mediawiki people did say it shouldnt cause problems
<spiderworm_work> im just worried that there would be a reason it was set up that way and i dont want to get hosed
<spiderworm_work> fear of not having spoken to The Right Guy among the mediawiki people, i guess....
<inviso> Then ask them *specifically why they did it
<inviso> If they can't tell you, stone them
<archivist> wheres domas
<spiderworm_work> lol
<spiderworm_work> is a DoS the internet equivalent to a stoning?
<domas> here!
<lakez> christ almighty why won't mysql 5.0.21 not work on freebsd 6.0 but 5.1.7 will, something is obviously broken
<domas> so, who bashes wide indexes?
<archivist> see above 500 byte length in mediawiki indexes
<domas> um, there's some issues with utf8
<domas> usually when we use binary fields
<inviso> <---- bashes wide indexes with a spiked club
<domas> inviso: innodb is wide index by itself
<domas> table
<HoloDoc> I guess there are plenty tutorials out there? Can someone suggest some tutorials with a lot of practical examples? I'm interested in complex queries in particular.
<inviso> domas: if you're indexing out past where data uniqueness is already established, you're wasting space and processing resources
<domas> inviso: you do not have to hit table again! :)
<spiderworm_work> http://pastebin.com/710261 <--- teh offending query written by those darn mediawiki people
<inviso> For example, with last names, the breakpoint comes at 3 or 4 characters depending on your data
<domas> on the other hand, how do you establish uniqueness?
<domas> by PK?
<domas> by having wide PK you put it into all indexes anyway
<inviso> no, "sufficient" uniqueness
<domas> so if you have wide index, that includes PK fields (for sorting), you do not waste space
<inviso> so the index can do it's job while being as small as possible
<domas> inviso: you can't sort based on truncated index
<inviso> entirely not true
<domas> try it
<domas> anyway, yes, we're duplicating data in indexes
<inviso> gah, I suppose the optimizer might not be making choices on the sorting, but it wouldn't have to be that way. I may have to check that for sorting
<domas> inviso: well, filesorting on semi-sorted data isn't fun
<inviso> if the cardinality is low enough, it would be fine
<domas> you have actually to do lots of small sorts
<inviso> true. Depends on when the small sorts were done
<inviso> but I guess I don't know that the optimizer p***es off the semi-sorted data or if it doesn't even try semi-sorting
<domas> spiderworm_work: if you're not really going to use unicode too much (as in rely on tailorings/collations for sorting), you may use old BINARY-based character sets code
<domas> inviso: I have some entries on that
<domas> in bugs
<spiderworm_work> ack, im getting a different error now... http://pastebin.com/710275
<spiderworm_work> wtf is "insert ignore" ?
<domas> what the heck
<inviso> ignores errors
<gleam-> what version of mysql are you using spiderworm?
<domas> spiderworm_work: which version of mysql?
<gleam-> this sounds like 3.23 to me
<will[werk]> spiderworm_work, MySQL version?
<domas> this code looks like installer to me :)
<spiderworm_work> Ver 14.7 Distrib 4.1.13a, for apple-darwin8.0 (powerpc) using EditLine wrapper
<gleam-> that's the client not the server
<spiderworm_work> oh duh
<gleam-> select version() on the server
<spiderworm_work> lol
<spiderworm_work> 3.23.58-log <-- you're right, gleam-
<domas> 500 sounds like something old to me too
<domas> !tell us about 3.23
<gleam-> are you able to upgrade?
<SQL> domas asked me to tell you this: MySQL 3.23 is a version of MySQL used by idiots or people whose systems are maintained by idiots.
<domas> \o/
<gleam-> at the very least to 4.0
<domas> had to do my daily quota
<gleam-> afaik mediawiki requirets 4.0 and up anyway
<HoloDoc> ah, another one with 3.23
<lakez> in 5.0.21 i get a floating point exception... when i enable logging, there is no exception.. wtf
<domas> spiderworm_work: we dropped 3.23 with 1.6 I guess
<spiderworm_work> ... yikes
<spiderworm_work> i better tell teh admins
<inviso> domas: Thanks for clarifying the reason for the wide indexes. That's pretty icky but it sounds unavoidable :(
<domas> inviso: there're more reasons
<HoloDoc> I told my boss that it is heavily outdated, he mailed the admin, admin said "yeah, it's so old that there are no more exploits in it, so it's mega-secure."
<inviso> domas: fire away :)
<domas> inviso: really, when you're doing various scans, hitting PK+row data may be to expensive
<HoloDoc> I fear going into a discussion about that :)
<domas> I already told how innodb holds data
<domas> so the price is same :)
<domas> inviso: we want all our data to be clustered properly
<domas> inviso: so no random hits would be done
<inviso> domas: meaning innodb effectively stores the data in the index?
<domas> inviso: yes
<inviso> And not in the table?
<domas> inviso: right
<domas> table _is_ index
<inviso> interesting
<domas> it's very powerful weapon, once you know that :)
<inviso> /nod
<domas> anyway, if you find something weird in mediawiki code
<domas> it _has_ to be so ;-)
<archivist> hmm heard of it but didnt know innodb did that
<inviso> domas: hehe
<domas> well, it's not much sense, when your PK is ID BIGINT AUTO_INCREMENT
<inviso> domas: In which case, I'll come hunting you for an explanation ;)
<domas> because if you do not define PK, InnoDB creates BIGINT there
<MrBIOS> hi there, is there a way to do an SQL-level search-and-replace?
<MrBIOS> on a field in a table?
<domas> MrBIOS: REPLACE()
<domas> MrBIOS: and LIKE :)
<domas> inviso: our database run at ~30k sql queries/s at peaks :)
<booga> sorry about the same paste
<booga> http://rafb.net/paste/results/IDSey964.html
<angela24> hi does anybody know about mysql control center?
<domas> angela24: it's deprecated
<booga> this is what is on the logs for workx.err
<angela24> what?
<dan__t> unf.
<dan__t> The only references that I can find to FreeBSD's kern.maxdsiz and friends are listed, well, in the MySQL manual for FreeBSD notes.
<inviso> domas: including or excluding writes?
<domas> inviso: with writes
<inviso> domas: what's the ratio on reads to writes?
<inviso> approx
<dan__t> Anyone here do much with MySQL on FreeBSD?
<domas> inviso: um, 1:20 maybe
<angela24> does anybody know about mysql control center?
<domas> on master
<domas> *shrug*
<domas> not that many writes
<inviso> domas: not too shabby :)
<domas> maybe some 200-300 writes/s
<MrBIOS> domas, thanks, what I need to replace is in a multi-line MEMO field, is there anything special I need to do?
<domas> MrBIOS: no
<angela24> :( pls
<HarrisonF> angela24: what about mysql control center?
<domas> angela24: it was deprecated, use mysql administrator!
<angela24> um what is deprecated?
<angela24> I was creating a database but the server doesn't connect
<angela24> HarrisonF?
<angela24> domas?
<booga> Starting mysqld daemon with databases from /usr/mysql/mysql/data
<booga> then it just stops there
<booga> doesn't move to the next prompt
<archivist> angela24, deprecated == no longer supported or not recommended
<angela24> thanks you archivist
<angela24> where can I get mysql administrator?
<archivist> mysql.com and dont ask that in ##php
<domas> ;-)
<swarmz> hi. i'd like to select the structure of a table. i have tried http://dev.mysql.com/doc/refman/4.1/en/show-create-table.html but it does not return valid SQL. is there a way to select a valid SQL create structure for a table?
<mercestes> are you trying to duplicate a table with no data? or are you trying to specifically output the SQL statement to "create table"
<swarmz> i am trying to duplicate a table with no data
<swarmz> is there a more direct query possible?
<mercestes> well, ppl yell at me for suggesting this...I don't remember why...
<mercestes> but create table <table2> select * from <table1> where 1 = 0 always works for me.
<mercestes> that duplicates your table with no data....
<Woosta> Is there any way to do a multiple update? Like a multiple insert is like: VALUES (1,2,3,4), (5,6,7,8) ..
<mercestes> There is some rare instance that that doesn't work tho....I think...atleast....that's what ppl tell me when I suggest that.
<swarmz> hmm thanks i'll check it out
<mercestes> update by definition is multiple....update table set col1 = value1, col2 = value2 where <condition>
Return to
#mysql
or
Go to some related
logs:
suse 10.1 wine.conf
#web
gentoo docache
opensuse low ram install 10.1
broadcom 7021 driver
unmasking ipw3945 amd64
wingrub ubuntu
md0p1 node
gien.com whois
#perl
|
|
