#mysql
20 October 2007
Total 24 pages. You are browsing page 24/24.
First :: Prev :: [...] [20] [21] [22] [23] [24] :: Next :: Last
23:52 <****> sivik: needs to be mysql:mysql
23:52 <****> chadmaynard, ok
23:52 <****> sivik: chown -R /path/to/datadir
23:52 <****> whoops
23:52 <****> chadmaynard, i don't want to do that to the /usr file
23:52 <****> beebeee: depends on the engine, but typically, the OS disc cache, and the mysql key buffer
23:52 <****> krow: I think it might be better to do a thread per connection
23:52 <****> sivik: chown mysql:mysql /path/to/datadir
23:52 <****> sivik: the datadir isn't /usr
23:52 <****> fuse: Probably... depends on the application.
23:53 <****> the datadir says /var/lib/mysql
23:53 <****> fuse: We have never created an async protocol yet, or it would be a lot easier to do this sort of thing.
23:53 <****> sivik: good. chown -R mysql:mysql /var/lib/mysql/
23:53 <****> already owned by mysql:mysql
23:53 <****> drwxr-xr-x
23:53 <****> It's cool, It should only take me a couple hours to get it re-written.
23:53 <****> fuse: But, you aren't after speed or responsiveness, and you just want to save resources... then go for a singlle thread.
23:54 <****> sivik: if you pastebin it'd help. does ibdata exist? and who owns it?
23:54 <****> fuse: Keep in mind that each prep statement takes up memory on the server. If you aren't careful you can run the server out of memory.
23:54 <****> mysql:mysql
23:54 <****> chadmaynard, mysql:mysql
23:54 <****> sivik: ibdata does exist then?
23:54 <****> krow: I am going for throughput
23:54 <****> chadmaynard, yes
23:54 <****> ibdata1
23:55 <****> sivik: then rerun mysqld --user=root and pastebin the output this time
23:55 <****> fuse: Probably a worker set of SQL threads then.
23:55 <****> i have to wait for kde to finish to install before i can pastebin
23:55 <****> i'm only in cli right now
23:55 <****> fuse: It might be a good idea to create all the prepared statements you need immediately after connect and use them thereafter; if you need to reconnect for any reason (e.g. server kicks you off), then you will have to prepare them again.
23:55 <****> sivik: that's why god created SSH
23:55 <****> MarkR42: ok
23:56 <****> chadmaynard, thats try, didn't think about that
23:56 <****> dee de dee
23:56 <****> A robust, long-lived application must handle reconnection gracefully
23:56 <****> I have been putting this off, and haven't needed yet, but now I am gonna have to create a struct for each thread context.
23:58 <****> I'm looking for a way to create a user with *all* the same permissions as another user in my MySQL database. I have two users I didn't create, Root and WP-ADMIN who I can access from pages run either on the same server as the database or on a remote page accessing the database, respectively, and I'd like to create a third username with access that they each have from their respective hosts, however, I cannot find how to do this.
23:58 <****> If you are not concerned with being able to to concurrent acces, you could wrap a fat mutex/monitor around all your mysql access functions.
23:58 <****> Icehawk78: no formatting
23:58 <****> here the error: http://pastebin.com/m6bfc2d94
23:58 <****> Sorry about that.
23:58 <****> Of course in such a program every function must handle errors absolutely correctly
23:58 <****> MarkR42: I'd prefer having one per thread.
23:58 <****> If an error occurs in a function it *must* clean up all resources and exit cleanly, otherwise it would die
23:58 <****> sivik: i said mysql --user=mysql didn't i?
23:58 <****> chad: do I need to resend the message, or were you just telling me for future usage?
23:58 <****> i don't have that set as the user anymore
23:59 <****> Using C for this is just asking for trouble, especially with your current error handling policy
23:59 <****> Icehawk78: future
23:59 <****> yes i did, same error
23:59 <****> There are many, many places where errors could occur. Any time you call a mysql server-side thing you could get an error which requires you to reconnect and restart the txn
23:59 <****> chadmaynard, http://pastebin.com/m6791b9f1
23:59 <****> There are a lot of possibilities for error in a long-running daemon written in C
--- Log closed Sun Oct 21 00:00:01 2007
Total 24 pages. You are browsing page 24/24.
First :: Prev :: [...] [20] [21] [22] [23] [24] :: Next :: Last