On Mon, Jul 6, 2009 at 10:48 AM, Nagananda<naganan...@gmail.com> wrote: > 1. Where do we find the mysql 5.0.51b [Mysql version of current db] > for linux installation. [any guides on linux installations – please suggest]
Naganand, I also recommend Ubuntu Linux 8.04 Server: http://www.ubuntu.com/getubuntu/download-server We've run this successfully, as does another big customer, Enda. It's what we have the most experience with. Be sure to get 8.04 - it's the long-term supported version. > 2. [...] Hence a full backup takes 5-6 > hrs and restoration takes atleast 14 hrs. We cannot afford this kind of > downtime in production as the only day available for the same is Sunday > only. where as hard copy of datafile can be done in 2 – 3 hrs. Yes, it's good to avoid long downtime. I have a few suggestions for options; I recommend the first option as it is the easiest in my opinion. A. Use a two-step restore using incremental binary log data. This is probably the easiest way. 1. Stop the production Windows Mifos and MySQL servers. 2. Take a full backup (at a time I will call T1). 3. Start the Windows Mifos and MySQL server again, this time starting MySQL with --log-bin option. (See http://dev.mysql.com/doc/refman/5.1/en/backup.html) 4. Restore the backup from time T1 onto the Linux server - this takes a long time but you are running on the Windows server, so that is ok. 5. At a later time that is convenient (time T2) stop both Windows Mifos and MySQL servers. 6. Rotate the Windows MySQL binary log by using FLUSH LOGS (see the link above). 7. Copy the binary log from the Windows server to the Linux server. 8. Use point-in-time recovery to apply the binary log data (everything that happened from time T1 up to time T2) to the Linux MySQL. (See http://dev.mysql.com/doc/refman/5.1/en/point-in-time-recovery.html) 9. Start the Linux MySQL server and point the Windows Mifos server at it. 10. You are in production! Steps 6-9 will be fast - on the order of a few hours - since you will not be taking and restoring a full backup. . . . . B. Set up Master-Slave replication to a Linux MySQL database Second easiest. This might be useful if you are going to set up replication anyway. Once you do it once, it will be easy to do it again. 1. Set up master-slave replication between Windows and Linux MySQL servers (Windows: Master; Linux: slave). This can be done with only a small amount of downtime for the production server. 2. Wait for the replication to complete 3. Stop both MySQL servers. 4. Make the Linux server the primary server (either disable the replication or make the Linux the master and Windows the slave) 5. Point the Windows Mifos server at the new Linux MySQL server. 6. You are in production! . . . . C. Use a multi-threaded restore program. I've never done this, so there is some uncertainty with with it. I'd recommend testing this first to see how fast it is and whether it works well. 1. Get the Maatkit MySQL tools (http://www.maatkit.org/tools.html) 2. Test thoroughly on test servers before doing this! 3. Stop the Windows Mifos server. 4. Use mk-parallel-dump to make a dump of the Windows MySQL database. 5. Use mk-parallel-restore to restore the database dump taken in step 3 to the Linux server. 6. Point the Windows Mifos server at the new Linux MySQL server and start it. 7. You are in production! > 4. Also, the replication server will be also our reporting server > which will be on windows. So can we do replication of mysql database from > linux to windows. Yes, this should be ok as long as both MySQL servers are the same version. We are not storing binary data in the database as far as I know - all the data is standard MySQL datatypes, which should be fine. -adam -- Adam Feuer <adamf at pobox dot com> ------------------------------------------------------------------------------ _______________________________________________ Mifos-users mailing list Mifos-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mifos-users
