Re: my.cnf file
michel wrote: I set up mysql and can't start it because I need to hard code the IP address parameter (bind-address) into my.cnf ... but I have three of them in different sub directories of /mysql/mysql-test/suite Should there not be one basic one? http://dev.mysql.com/doc/refman/5.1/en/option-files.html It searches in order of locations, /etc/my.cnf being the first. -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Mysql upgrade from 4.1 to 5.0 to 5.1
A DB politics between mysql oracle again Thats quite ironic given recent events :) -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Default my.cnf for (very) high performance servers....
Hi All, We're setting up a group of servers using MySQL Enterprise 5.1 - Rather than starting with a blank canvas I wondered if there was a suitable my.cnf that is tuned to the kind of environment I'm running where I can tweak it from there. We're running on RHEL, on Sunfire X4140's - 8 disks, 16G RAM, 2 x dual core 3000mhz 64bit... which is reasonably beefy. Environment is more read than write, but write speed is important. Anyone know where I can look? Cheers Craig -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Default my.cnf for (very) high performance servers....
Craig Dunn wrote: Hi All, We're setting up a group of servers using MySQL Enterprise 5.1 - Rather than starting with a blank canvas I wondered if there was a suitable my.cnf that is tuned to the kind of environment I'm running where I can tweak it from there. We're running on RHEL, on Sunfire X4140's - 8 disks, 16G RAM, 2 x dual core 3000mhz 64bit... which is reasonably beefy. Environment is more read than write, but write speed is important. Anyone know where I can look? Cheers Craig I should add, I wanted something a bit more up to date than my-huge.cnf, which seems to think a huge server is a system with memory of 1G-2G -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Default my.cnf for (very) high performance servers....
Andrew Braithwaite wrote: There's no such thing as a generic my.cnf for high performance MySQL servers, you will need to provide more information.. Well, I was more after something a bit more up to date than my-huge.cnf that I could use as a starting point, I see a few example ones posted to Mysql Forge, but they are very innodb orientated. Some questions: Are you going to run InnoDB or MyISAM or both (if both, what's the split?) Both, 90% MyISAM Is there anything else running on that server? i.e. how much of the 16GB is available for MySQL to use? It's a dedicated MySQL box Can you partition your disks as you wish? (How much data do you need host?) About 50G of databases - I've currently got 6 disks with RAID 10 running soley /var/lib/mysql (datadir) on an LVM with the binlogs being written to the other 2 disks (which has the OS on them too) Will this server be a master or slave or standalone? (Do we need to deal with binlogs here?) There are 3 in total, 1 master and 2 slaves (one of which is capable of being failed over to as a master) The current MySQL 4.1 servers that they are replacing have at any one time on average about 1000 open tables, about double the number of selects than inserts, between 2000 and 5000 qps - if thats any use. Cheers Craig -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Default my.cnf for (very) high performance servers....
Andrew Braithwaite wrote: Your disk config is good and you'll need all the nessesary my.cnf entries to point all the logs and data to the correct place. Slaves should have the relay-logs going to the OS disk too. I assume you've set up the master slave config in the my.cnf too. Yeah the replication and file location stuff is fine, I was after a rough idea of buffer sizes...etc, thanks for all that, most helpful.. Cheers Craig -- Linux web infrastructure consulting, cr...@codenation.net Free live poker tournament listings, http://www.g5poker.com -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
INNER JOIN order issues.
Hi, I'm trying to migrate an application from 4.1 to 5.1, theres a bunch of queries that seem to be failing and it looks like the order of INNER JOIN's... for example... SELECTetc INNER JOIN tablex AS x ON y.foo = a.bar INNER JOIN tablea AS a ON y.foo = b.bar ... works in 4.1, but in 5.1 I get unknown column a.bar, but if I switch the order of these joins around it works fine. Is this a known compatibility issue between the major versions? and is there a work around apart from changing the SQL (which we're trying to avoid where possible!) Cheers Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Master-master replication configuration...how?
blue.trapez...@gmail.com wrote: Hi I have 2 servers and I am trying to set up (for testing purposes) a master-master replication environment. I read the instructions in the manual for master-slave, but am not able to find any information on how to set up a master-master system. Can someone on this list point me to any documentation on this topic? What you are probably trying to achieve is bi-directional replication (arguably different from multi master but as close as you can get) - take a look at auto_increment_increment and auto_increment_offset, the idea is that all your primary keys use auto_increment integers and one master will be offset from the other (ie: server A uses 1,3,5,7,9...etc and server B uses 2,4,6,8...etc) so they never clash Also, I am new to replication. Could someone briefly tell me typical problems with this kind of setup, or why it is/is not advisable? What's your primary reason for wanting multi master? It's worth noting that this approach shouldnt be used if your primary concern is master load... since using traditional statement based replication, each master is going to process all write queries. We're currently implementing a passive/active multi master set up using MMM (http://code.google.com/p/mysql-master-master/) to handle failover, it's a lot safer. Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Upgrade large databases from 4.1 to 5.1
Hi All, I need to migrate a large (30G) database from 4.1 to 5.1 on a live system that cannot afford a large amount of downtime. The official method (copy files, run mysql_upgrade...etc) is looking like it will take forever, particularly since I need to move it 5.0 before 5.1. How do people normally manage this in a high availability environment? One idea being floated is to set up slave running 5.1 to replicate off 4.1 and then cut it over to being the master when we're ready to migrate... is this feasable or dangerous? Anyone else who's dealt with this kind of migration before have any other ideas? Thanks in advance. Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Upgrade large databases from 4.1 to 5.1
Baron Schwartz wrote: If you can't take downtime, I'd go the slave route. You should certainly test your application to make sure 5.1's differences (data types, syntax, etc) don't cause problems. Otherwise you're risking getting badly stuck and having to downgrade to 4.1 again in a crisis. If you dump and reload, you don't need to go to 5.0 first. That is only for in-place upgrades with mysql_upgrade, which I would not do anyway because of the file format changes. I would dump and reload. Sorry I wasn't very clear there - testing will all be done in a QA environment before anything is cut over, what I'm after is a way of switching from 4.1 to 5.1 as quickly as possible when we come to do the live stuff. Looks like replication may work from what you are saying. Cheers Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Operating Systems.
I'm currently spec'ing out a design plan for a large scale MySQL infrastructure to support a high-read large scale web environment for a client. I've got the overall MySQL set up planned out (which I'll post here later to get peoples inputs/advice) but one issue which is currently undecided is what operating system to support. The boxes will be Sun AMD64 servers with 8 local disks each, there are numerous people arguing for Solaris and equal numbers arguing for Linux. So, leaving advocacy at the door, what is the best OS to use for a MySQL set up, Solaris or Linux? And specifically, the reasons why one is better or worse than the other. Also, what filesystem type to use for datadir would be recommended for best performance, this question obviously depends on what OS we're running. The applications running on it will be mostly using MyISAM. Thanks in advance, Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/mysql?unsub=arch...@jab.org
Re: Comparing Data
Darvin Denmian wrote: Hello, i need to know : there is a way to compare data between Mysql Master and Mysql Slave. We had some problems with replication and do not know if the data is intact. Thanks. Have a look at http://www.maatkit.org - I think they have a useful tool for this. Regards Craig -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
