| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Comments:
<0> i follwo the guidance of http://dev.mysql.com/doc/refman/5.0/en/replication-howto.html
<1> "Not I," said the fly.
<0> any idea?
<1> I'm changing the subject so I seem smarter than I really am
<2> !man DATE_FORMAT
<3> (Date and Time Functions) : http://dev.mysql.com/doc/mysql/en/Date_and_time_functions.html
<2> Hm
<1> !man CLUSTER_****
<3> Nothing found.
<4> uh, qkit
<4> http://dev.mysql.com/doc/refman/5.0/en/change-master-to.html
<5> hello, when using like or so, you can use % and _, what other signs are also used ?
<4> you're skipping the _ on like all of your variables.
<6> blehhhhh: that's it
<1> huh.. locking a table for writing locks it for reading?
<1> I mean.. I just tried it and it did
<1> but I didn't expect it to
<5> guys, if 5 connections do a select on a default myisam table at the same time, will there be a dreadlock?
<5> or will it queue ?
<7> write lock allows 1 connection to read and write, others are locked out
<1> blehhhhh, why can't they all go at the same time?
<7> there will never be deadlock on myisam
<7> read lock allows all to read, but none can write
<1> lokus, cool
<1> lokus, how do table locks work in sessions?
<7> however if 1 connection request for a write, all future read will have to wait tilkl the write completes
<0> infi: where do i put wrong?
<1> err, transactions, not sessions
<1> I guess I can try it
<7> myisam has no transaction support
<4> qkit: right where it says in the error message. "MASTER_USER", not "MASTER USER"
<4> qkit: and the rest.
<0> i see
<4> qkit: read the link I pasted.
<0> oops...never read it carefully...
<4> that much is obvious
<0> sorry, and thanks for the help infi
<4> cheers
<1> apparently table locks don't care about transactions
<1> that's actually what I want, I guess
<0> hmmm infi? even i change liao the error still appear? http://pastebin.ca/44415
<0> liao=already*
<4> wow.
<4> line 11, your new paste, skipped the CHANGE MASTER TO line, even after you added the quotes.
<4> throwing command after command at a database server, hoping one of them will work, is not really an effective way of managing a database.
<0> infi, i found the error....
<0> i missed out the , in every line...
<4> that too
<7> qkit: i guess you replication will still not work
<0> why?
<7> try to start it and see for yourself
<0> you mean start slave?
<0> i have change my conf to this
<0> http://pastebin.ca/44416
<7> this looks fine then
<0> lokus? if my processlist show like this? was it mean the database are replicating?
<0> http://pastebin.ca/44417
<7> show slave status
<7> maybe not, don;t see the slave_io
<0> sorry to be so annoying with all those , cause this is the first time i play around replication in mysql.
<7> maybe that is it there
<0> http://pastebin.ca/44418
<7> slave_io not running
<0> huh? where can you see that?
<7> show slave status\G
<0> hmmm, lokus? so that means my sql are not replicating?
<7> no
<7> it is not
<0> so how can i fixed it?
<7> troubleshoot it
<0> do i need to add something in my my.cnf?
<7> check the error log
<0> what is the used of the Slave_IO_Running: No
<0> ?
<7> !m qkit replication
<3> qkit: (Replication in MySQL) : http://dev.mysql.com/doc/mysql/en/Replication.html
<7> read that first
<8> mysql causes a lot of wait time for me while doing a large .sql import. Is that normal?
<0> ok....
<0> hmmm, i think will calll it a day for today....
<0> going to continue it on monday...after i have finish read it...
<0> and thanks SQL and lokus.
<9> anyone on ubuntu/deb? where does mysql data get stored?
<10> find /var -name mysql -type d
<10> ;-)
<7> show variables like 'datadir' will tell you
<8> I'm importing wikipedia data and it's been running about 72 hours straight importing pagelinks
<8> I'm thinking maybe I'm using my hard drives stupidly or it's some conservative default settings
<10> scarfboy: disable innodb log flushing, increase buffer pool
<8> doesn't disabling the flushing sort of destroy the point of having binlogs?
<8> Not that I need them, really.
<10> scarfboy: it's about transaction log
<10> you can reenable later if you want that consistency
<8> yeah well, it's been running three days now, I don't want to stop it
<10> scarfboy: btw, why are you importing pagelinks? why not just importing the xml dump?
<10> (our servers import full english wikipedia sql dump in ~3 hours)
<8> I was tring to make something work. Templates weren't working, turns out I needed the 1.6alpha of mediawiki
<10> the slower ones even
<8> buffer pool is 128mb. hm.
<10> well, pagelinks is obviously bigger than that
<8> I don't know if stopping it in the middle will break something. I don't even know what pagelinks is.
<10> pagelinks = 40m rows with 4gb data and 3gb indexes ;-D
<8> ouch.
<10> you can't import that with 128m buffer pool.
<10> really.
<10> it has to fit whole in buffer pool in order to make proper import :)
<8> does it?
<8> Isn't each line in the .sql a transaction of its own?
<10> well, it _may_ be slower a bit
<8> or is that unrelated?
<10> well, single line in pagelinks may hit zillions of spots in index
<10> so you would like to have those spots in memory
<10> hehehe
<10> and flushing transaction log means that every line in the sql may require flushing
<10> I don't remember if blocks are wrapped around with begin/commit :)
<10> anyway, it's rather big dataset, no wonder you're importing it 3 days
<11> http://pastebin.com/583289 <- exist() just returns the number of rows.. could that query be made into one?
<10> especially with no buffers/memory, and... single disk ;-D
<10> !m eml on duplicate key
<3> eml: (INSERT Syntax) : http://dev.mysql.com/doc/mysql/en/INSERT.html
<10> eml: INSERT ... ON DUPLICATE KEY UPDATE ..
<11> ok =)
<11> thanks domas
<8> Fair enough, I sort of suppose. Still, it'll take more than a week at this rate. Ugh.
<10> scarfboy: longer.
<10> scarfboy: indexes grow...
<10> they're slower to update this way :))
<10> and... hahaha... you've got 128mb buffer pool and sync flushing of logs
<10> and, wait, do you have binlog enabled?????
<8> yeah
<10> ;-)
<8> default install with some minor tweaking
<10> the reliability guy!
<8> why gee
<11> INSERT INTO bolted_online (usr_id, usr_lastactive) VALUES (id, timestamp()) ON DUPLICATE KEY UPDATE usr_lastactive=timestamp() WHERE usr_id=id <- like this?)
<10> I can tell you a funny secret
<8> what other reason is there to use a database?
<10> scarfboy: wikipedia mysql servers run on raid0
<8> (..other than the convenience)
<10> and have sync flushing off ;-)
<8> I don't doubt it.
<10> so if a disk on master fails
<10> site is readonly for an hour (or half an hour)
<10> or longer.
<10> but hey, new winds blowing, maybe I'll switch everything to raid 0+1 in several months! :)
<8> I'll read up on clustering and such when I have multiple servers to abuse, heh.
<10> now we run 7+8 mysql instances to keep the site live :)
<11> domas, you're an admin of wikipedia?
<10> eml: that's hobby, yes ;-)
<11> domas, cool =)
<8> Neat indeed:)
Return to
#mysql
or
Go to some related
logs:
kernel-source foresight package
#css
#linux
qtparted opening as ntfs failed operation not supported
#mysql
postfix filter outgoing mail
ubuntu binfmt_0000
#linux
gentoo Recipient addresses must be specified on the command line
debian streuner
|
|