Hi, I have the current Multiserver Setup: Master Web,Mail Server Master/Master MySQL Replication, primary DNS Master/Master MySQL Replication, secondary DNS Slave Web All systems running Debian 6 Squeeze with MySQL 5.1.73 and ISPConfig 3.0.5p5(The first is the Master with web interface and all the others are slaves). I updated source list on all systems to: Code: deb http://packages.dotdeb.org squeeze all deb-src http://packages.dotdeb.org squeeze all deb http://packages.dotdeb.org squeeze-php54 all deb-src http://packages.dotdeb.org squeeze-php54 all I tried to update/upgrade in order to update mysql to 5.5 arrording to this-> https://www.dotdeb.org/2014/07/22/mysql-5-5-38-for-squeeze/ But only mysql-common and php updated. So I found this article -> http://www.ovaistariq.net/490/a-step-by-step-guide-to-upgrading-to-mysql-5-5/ and I'm thinking to update using that guide to the latest 5.5 version. Firstly update the Master and then the Web slave and after that stop the slaves from replicating in MASTER/MASTER and update those too one at the time. Any ideas on what I'm trying to achieve? How would you make that happen? I don't want to damage my installation. Thank you
Why dont you update to wheezy? Squeeze is quite old already and does not has ull update support anymore. Yu will have to update to wheezy sooner or later anyway, if you want to keep your system secure.
Very nice idea. I was thinking of the same thing. But isn't whezzy part of debian 7? All my systems run debian 6. I would have to reinstall the whole VPS installation on all systems, losing all settings and data, wouldn't I?
Wheezy is debian 7. Like Ubuntu, each debian version has also a codename beside of the number, so squueze = 6, wheezy = 7 Debian systems can be updated to the next distribution version, you dont have to reinstall the systems. I have some servers here that were updated from Debian 5 to 6 and now 7.
This seems decent https://www.howtoforge.com/how-to-upgrade-debian-squeeze-to-wheezy Do you recommend any other way to update for cluster systems? What about MySQL Master/Master replication? Stop the slaves and update or just update without bothering about replication?
I updated to wheezy but i got an error regarding php mysql and dovecot could not be started, maybe because i previously tried to update mysql to 5.5. Mail account could not connect. So I completed removed dovecot php and mysql services and reinstall. MySQL works fine under 5.5 on Web and mail servers and php under 5.4 Dovecot starts fine but all mails cannot connect using imap pop or webmail. MySQL Master/Master cannot connect/start after reboot Help please??
Check your logs regarding mail. For mysql use show slave stats \G in mysql. mysql 5.5 uses a different definition for the master (moved from my.cnf to mysql-console). I upgraded some servers from lenny to wheezy a few days ago - and this was not much fun.
Mail logs point mainly to a database connectivity issue: Code: amavis[7140]: (07140-01) (!!)TROUBLE in process_request: connect_to_sql: unable to connect to any dataset at (eval 116) line 247. postfix/qmgr[3998]: error: unsupported dictionary type: mysql postfix/smtpd[11805]: fatal: no SASL authentication mechanisms postfix/smtpd[3173]: lost connection after CONNECT from localhost.localdomain[127.0.0.1] What do you mean by that? I got that so I'm currently backing everything up to start clean and setup the all the systems all over again. That was what i was trying to avoid.
There should be mysql errors in the /var/log/syslog file that show which config settings in the mysql my.cnf cause the start problem.
With mysql 5.5 use Code: CHANGE MASTER TO MASTER_HOST="host.example.com", MASTER_USER='user', MASTER_PASSWORD='password' to set the the master (and remove the settings from my.cnf. Are you sure, that the latest apt-get upgrade installed and configured all packges?
Well the error was because of what florian030 mentioned. Basically from 5.5 and forward properties of master-host, master-user, master-password have been deprecated and not allowed in my.cnf. Yes but now i started all over again anyway. I made it work without my.cnf on a clean installation(at least i think i did. I haven't had the chance to test yet but "Slave_IO_Running" and "Slave_SQL_Running" respond to "YES" on both MySQL servers). From what i understand MySQL 5.5+ takes care of the replication without my.cnf master-* properties. You should set the server-id in my.cnf Create user in all servers and grant slave privileges Change master to master_host, master_user, master_password, master_log_file, master_log_pos on all servers accordingly Start slaves on all servers I don't know if the collision issue that was bypassed (by increasing auto_increment_increment and auto_increment_offset based on number of servers) has been fixed in MySQL 5.5+ If someone know about that, is worth mentioning. I will test in near future with and without auto_increment_increment properties if anyone wants to know.
You don´t need to adjust the log-positions. Just remove the client-settings for the master from my.cnf and set them in the mysql-console. You will get an inconsistence replication if you changed the log-pos to a higher value. You can dump the databases on one master including --master-data and import this dump on the slave (but do not include any databases that you excluded from your replication ). Maybe you find some more here or here.
Well it was a clean installation without any databases. So i reset the masters on both servers and set the correct master_log_file, master_log_pos on both(it is a Master/Master Replication). It seems that is replicating just fine an example_db. Does 5.5+ needs me to specify databases that i don't want to be replicated?(binlog_ignore_db?? Replicate_Ignore_DB??)