Dear all,

I'm trying to setup a development server which hosts a number of virtual hosts.
and i need your advice with the following:

1. What's the best Filesystem to be used with an apache server + mysql?

Relevant info:

i. Mysql DB size' 1 GB
ii. Code directory size' 8 GB (Max file size is 5 KB)

2. What's the best file structure to maximize performance ?

Relevant info:

i. 2 X 1 TB SATAII 64MB blk Edition Int HDD

ii. 1 X 320GB


General info:

* I have an additional directory which hosts all my code under /Development . this directory has a size of almost 300 GB.
* 14 Virtual Hosts would be created on the server under /etc/httpd/VHosts/
* All 14 Developer connect to this server and work simultaneously each on his branch.




I was thinking of the following structure:

1. Drive A (320 GB): "/" + "/boot"
2. Drive B (1 TB): "/var" + "/home"
3. Drive C (1 TB): /development + /opt

Either that or going with the default setup of all HDD being set as one LVM group


Would my custom setup work? i followed no guidelines with the above setup, it's just using my own logic. so i'd appreciate an expert advice with this.

Thanks,

--Roland

I would set up the two 1TB disks in raid1 + ext[3|4] and forget about the 320G disk unless you just want it for things like downloaded packages or tmp space.

Putting 14 developers time investment on a single spindle is pretty scary.

Thanks for the advice.
can you please elaborate on what you found as scary ?

i've changed my setup to the following:

320 GB:
/boot
lvm: /

1TB:
/var #mysql and such
/development #Code worked on by developers
/opt #files needed by developers are stored here

1TB:
Raid1 of /var, /development,/opt

and rsync cronjob that syncs every relevant directory to another physical server.

As described in the first post, a single drive failure would lose data permanently and if it happened to be the drive with the repo you would be sunk. With your rsync at least you would only lose recent commits if that is acceptable.

excuse me for being slow to understand your point.
i'm going to state what i've understood.
you pointed out that loosign the 320 GB would make me loose my work.
though there's nothign on it except the OS, + Virtual hosts file .conf files.
so if that drive died, a simple live cd would solve the problem untill issue is solved and if that didnt work there's always the other phsyical server.
note that all code needed is under /var and /shared which are both on the 1TB which is raided as raid1 to the other 1 TB.


if one of the 1TB, went down, the raided disk would take over.

is there something i'm missing ?

No, but your first post said nothing about raid

failure on drive C?

oh ok, well i took your advice as i previously posted: