I've had some brief discussion about this with other people on other lists and have decided to move the conversation here, since there are far wiser MySQL gurus here than I. Here's the scenario:
I want to have two (or more, but for now let's say two) db servers running. These servers will have MySQL and OpenLDAP2 on them and some other minor processes we won't worry about for the time being. Even OpenLDAP can be left out of the conversation since it's unrelated to this list. Now, for now we don't care about any freaky load balancing or anything (not that it could be done anyway), the load isn't that high - but it is operationally critical, and having the db server fail would mean bad things until it could be restored. If performance on the one server becomes an issue we'll just put in a faster machine / faster drives, or whatever. Oh, did I mention these will be running at a colocatoin facility where I'm not at most of the time? :) My idea is as follows: Both DB servers think they are masters. Normally this induces bad dreams of simultaneous db updates and such, but that isn't a concern here. Both servers are configured on an IP address using VRRP - the primary server is the only one taking client queries normally. The second server either replicates on a different IP alias, or a second interface completely (isolated vlan) on a fairly constant basis, to maintain a complete dataset of the first db. In the event that server 1 commits seppuku, server 2's VRRP process assumes the shared IP and MAC address from the first machine, and except for a few seconds in between client systems should never notice the difference. This is why server 2 needs to be a master normally, because at this point it will be updating the tables just like the first one. Ok, so there is failover step #1. Now for restoring the primary... The primary server is down, and will be kept that way (i.e. would require a manual switch back to primary) since we want intervention to know why it broke. So I'm assuming we can manually script it to re-replicate from server 2 to get a current set of information, then after that is accomplished assume the primary role again. So, is this even possible? It was suggested to me that 3.23.x cannot do this, but that MySQL 4 *might* be able to. What sayeth the gurus? Am I just too hopeful for an HA solution, is there a better/different way to accomplish the same thing, or what? Thanks!! - Ralph Forsythe Aspiring ISP Ninja --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
