Date: Tue, 01 Oct 2002 20:05:31 -0600 From: Ralph Forsythe <rf-list@centerone.com> To: freebsd-isp@freebsd.org Subject: mysql and LDAP in an ISP - redundancy Message-ID: <5.1.0.14.2.20021001195042.033c9370@mail.centerone.com>
next in thread | raw e-mail | index | archive | help
My databases currently function off mysql and openldap2 on a single server, which is backed up nightly. This is adequate for our current business but I'd like to move to a redundant scenario, and I'm envisioning this as follows: DB server 1 acts as the normal master for both databases. Using VRRP, assumes primary role for a shared IP address. DB server 2 is backup. It's also configured as a master server, but replicates off DB1. Should VRRP initiate a failover, DB2 then seems to be the exact same server to all clients. Once DB1 is restored, it replicates off DB2 for anything it missed, and regains control. So, what am I missing? Can this even be done? I know having two servers as master can be bad if the same tables are updated at the same time, but in this scenario that doesn't appear to be possible. Will manual intervention be required at some point? My base requirement is a transparent failover (within 3-9 seconds) to the other server without any reconfig of client systems. It's ok if restoration to primary service needs some manual control of some kind, but the hot-fail needs to happen. These servers will be in a semi-remote facility where access can be difficult at times, so I'm going for a solution that will stay running except in complete network meltdown at least in backup mode until we can fix it. Any DB gurus care to take a stab at this one? I haven't found much if anything on the net or in books on my design, leading me to believe I'm one of the early ones to try it. I can't be the first I'm sure however, so someone out there has to have an idea on how it works, if it works, or if I'm just smoking crack over here. :) For my next trick - load balancing (probably requiring some mad coding skillz) between multiple servers, which AFAIK is not something that can currently be done with mysql and openldap... Also - how resilient are current IDE drives to constant r/w activity? Will they wear out a lot faster than if the reads/writes are more sporadic? I know back in the day (i.e. MFM and such - all you people who even remember those are cringing right now) that was a concern, but I'm not sure if it's as such anymore. I ask because I'd like to crank up replication to a fairly constant process, and am curious what blasting the drives will do to MTBF. Thanks, - Ralph To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5.1.0.14.2.20021001195042.033c9370>