Re: [vchkpw] vpopmail with drbd / intermezzo

[Nick Harring]

Erik Bourget wrote: TuxRelated <[EMAIL PROTECTED]> writes: Hi List, our current qmail/vpopmail server gets a little unstable due to some raid problems, so i want to setup a 2-node failover cluster with qmail, vpopmail and imap. the plan is to use drbd to mirror /home via tcp/ip to the second node and to use heartbeat to do a failover if the primary fails. the second node runs the same daemons as node-1, so users still have full access to their imap-accounts. soooo :) does anybody have any experience with this setup, has done it before or has any hints? im also thinking about intermezzo as distributed FS. any eperiences with that? many thanks for your opinions, You're over-engineering. DRDB and Intermezzo are not ready for this. Try to imagine recovering from a failure in either of these. I whole heartedly agree. Intermezzo looks and feels like a research FS, which its pretty darn good for. Its not a production FS at this point. I suggest simply having every mail deliver to both locations, and hacking a pop/imap server that deletes from both. Personally, I don't go that far; I just deliver to both and have rsync clean up the second member of the storage pair when a delete happens. This is a little nappy if you're going to try and "cluster" these, as you'll get mail showing up and disappearing based on where the user hits until the rsync finishes. I don't know about anybody else, but I also get nervous cron'ing an rsync which removes files. Consider separating your MXs away from your mail storage boxes. Yes. Double Yes. Really, though, all you need is one stable box. Don't buy scummy IDE drives. You'll have a better time with one really good machine than with two crashy ones, and spend less. If people are paying to access their mailbox, I disagree 100%. 3 or 4 9s of availability isn't that hard to achieve with a well engineered solution. However, be ready to spend the right amount of money. There are several ways to solve this: Shared scsi storage with active/passive failover using something in the realm of Linux-HA or Veritas Cluster Services (opposite ends of the price spectrum. Also opposite ends of the ability to sleep at night spectrum). Simply hang an external scsi storage array (or fibre channel) off of two boxes on its two channels. Then have a cluster service mediating who has access to the boxes. This means one box is always going to waste, something which I'm not real keen on. You can expand this by having two external boxes and criss-crossing, with the ability for each machine to take control of both raid's, and then offer services from each with virtual IPs and fail over in each direction. The other solution is to build a real cluster, with redundant storage and front end machines delivering mail via NFS. Then you can abstract out as many or as few functions to their own clusters, and just access them via virtual IPs. This is the setup I'm migrating towards, piece by piece, for my hosted email platform. Right now I've got four front end smtp/pop/imap servers delivering to an active/passive veritas cluster on mirrored external raid (two external raid boxes mirrored in software by veritas). I'm moving this to a pair of NetApp filer heads, each offering different services and able to take over for each other (I feel this is less wasted hardware).  I'm also running a pair of fail-over capable MySQL servers on my NFS servers, though those'll soon be moved off to a different pair of servers with more horsepower. I'm using Linux-HA tools to cluster MySQL, Veritas Foundation Suite and Cluster Server to cluster NFS, and a Foundry load balancer to cluster/balance my front end servers. Its a rather complex setup, but so far I've got 5 9s of availability, which means I get to sleep through the night, every night. Hope that provides some suggestions. Obviously a lot of this can be done on a more modest, or more extravagant, scale as the needs of your platform are very definitely going to be different. You can get external storage really cheap these days using internal IDE and external scsi interfaces (I bought a 1.2TB raw unit for under $7K USD recently). You can build an LVS load balanced cluster with pretty low end hardware that'll keep up with full 100Mbs line speed. Nick Harring Webley Systems