Virtual disk allocation advice requested
I'm sick at home and in consolation am building myself a Xen VM test-bed
server - but have not used Xen before, nor LVM. I'll appreciate competent advice about organising and allocating disk space, please. Using a single large HDD initially, I thought: a) primary partition, <1GB, ext3, /boot b) LVM logical volume(s) for Xen's Dom0 c) multiple logical volumes, one for each Xen DomU d) NFS shared space (both for network and between DomUs) e) Samba shared space (somewhat optional because have other such spaces) - that latter revealing that I have an heterogeneous network Queries: 1 would it make better sense to put all of the Dom0 onto the primary partition or even a separate extended partition? 2 for Dom0 I thought something like: 2GB, swap, /swap 5GB, reiser, /var 2GB, reiser, /root (rootuser) 10GB, ext3, / but is it as necessary to break things out in a Dom0 situation? 3 I thought to use LVM logical volumes for the DomUs for the inherent extensibility, eg one domain will be used to build an email server, and if the project is successful I would add another HDD; but would it be somehow better to use extended partitions? 4 I haven't figured out where/when/how one allocates space within a DomU. Does/should one separate out space allocated to logs, the root user, etc, as we were advised in 'the good old days' or has such gone by the board/become irrelevant in the brave new virtual world? 5 will NFS 'play nicely' in an LVM logical volume or should it be an extended partition? 6 similarly will Samba go into an LVM logical volume or should it be an extended partition? 7 if I set up a DomU temporarily, eg to test some software/project, and subsequently discontinue it, should I also remove the partition/virtual disk allocated within LVM and then build space anew before creating some future DomU, or can/should one simply recycle the nominated logical volume? Any and all advice (or reference to reading materials) will be much appreciated! Regards, =dn |
Don't know much about Xen, but I can help with some of the issues.
1: No clue, but I avoid LVM, as it messes up fs journaling, which can lead to all sorts of fun issues such as data corruption and file loss. Here is some info on it. Plus LVM has a performance hit, I've not benchmarked it, but you can "feel" it during disk intensive operations. 2: Don't know, how you partition your fs all depends on your usage. I usually do a 100MB /boot, 2x the RAM for swap, 10GB-25GB for / and the rest for /home. 3: See #1 5/6: These are network protocols are are file system agnostic. They don't care how your disk is partitioned as long as the files are accessible. You could do an NFS share of a SMB share from a Windows machine mounted on your system. Hope that helps a bit |
Sadly not much help: Xen's disk virtualisation involves something of a performance hit, and sparse files are NOT the answer. However I'll be watching the performance issues for the same reasons as stated!
Having so many partitions, as advised, multipled across so many VMs, seems to me that we run out of partitions before too long, hence the thoughts about LVM... I have since noted that paravirtualisation (which I didn't include earlier) apparently contra-indicates the use of journalled file systems in certain situations (SuSE documentation). Regards, =dn |
An extended partition allows for either 127 or 255 logical disks inside it, can't remember the exact number, so you should be okay with that.
|
All times are GMT -5. The time now is 12:48 PM. |