Topic: RAID rebuild options - some input needed

[SoftDux]

Hi all

Before you tell me todo a search, or to learn the inner workings of SME, let me shed some background on what I need.

I currently have two identical PIV machines with 2x 160GB SATAII HDD's, setup with software RAID on Suse 9.3. Using Heartbeat the two machines act as a failover cluster. Furthermore, I use RSYNC to sync a PHP based app between the two and run MySQL-Max 5.0.15 to get master-master database replication.


This work(ed) well for quite some time, untill one of the HDD's in one of server went corrupt, and took down the whole server.   Luckely, there is two server, so no downtime.

This whole setup took me about 2 days to get working, since none of the stuff is built in to Suse, so it's trail & error all the way. RAID in itself was rather trail & error, since Grub doesn't boot properly from the second HDD.

SME caters for RAID, and can setup RAID on the fly, which I like but I have some very important questions to ask, and would like some input. Searching the forums, and trying to read throug aprox 2317 threads isn't really getting me far, as I don't have that much time on my hands.


So, here goes:
1. What's the most reliable setup? I want to use software RAID, as the client can't afford hardware RAID. Would a 4xHDD config as RAID 5 be good enough? Budget is limited and RAID 1 didn't work well.

2. In the event of a harddrive failure, how would I be notified of it? How easy is it to replace the HDD? The client in question is about 700Km's away, so I can't just pop-in and out as I like to fix problems. In this case I'd Fedex the new HDD(s) to the client, guide him over the phone on howto fit the new HDD, and then try and rebuild the RAID array via the internet.

3. What filesystem is being used? Is it easy to upgrade the HDD space?

I'v read through the manual, and as many posts on the forum as I can, but can't find defenate answers to my questions.

[Confucius]

From my point of view I would go for the most suitable solution. RAID5 is possible by software and has a better appreciation as all predecessors. SO that would be RAID5.

Rebuilding an array is as far as I can see (after being busy last weekend on USB + Raid) done by making sure the disk you would ship contains at least a descriptor code "FD" = "Linux Raid Autodetect" (done with FDisk).

Filesystem EXT3 is the most preferred choice since that can deal with bigger files as the EXT2.

Don't narrow yourself down to this forum only to search about this subject. SME is Linux and therefore you can find more information as only here ofcourse. I did find the info also everywhere.

Harro

[SoftDux]

So far SME seems to be the best option available for my needs.

The process, so far it seems, could be simple:

1. Install SME on server
2. Restore Backups
3. Configure MySQL cluster


Much, much easier than what I had

[idp_qbn]

Try using AFFA to rsync between two servers as well.
http://forums.contribs.org/index.php?topic=36499.0
(I am not an affa guru but the forums can help)

I lknow the RAID5 option is good but it still leaves you with a single server which won't help if there is a hardware failure other thank HDD.

You could use two SME servers, with two disks each as RAID1 and 'affa' to keep them in sync. This gives you a second chance when a server fails.

SME handles failure of a disk well, especially the RAID1 - I have had that at home and had to replace a disk which took very little time and very little effort on the keyboard: logon as root in a console, run "console" program and choose "Manage Disk Redundancy" and follow it from there.....should be easy (ha!) to do over the phone. And your user could be up and running quickly, while the RAID array synchronises in the background.

Cheers

Ian

[SoftDux]

I had a look @ AFFA, and it could do very well for what I need. I just need to try and figure out how to manage MySQL replication with little problems