[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Disk Mirroring in Sarge howto




On Tue, 14 Jun 2005, Siju George wrote:

> Yes its partition 2 . what would be an appropriate location???
> I just followed the BSD way where swap (b) comes immediately after / (a).
> What is appropriate for linux??

swap can be anywhere on the disks

swap is supposedly never used ...

why would you put swap in between ( / ) files you need hundreds of times
per hour  and /var files and /usr files ... 
	the head has to skip over swap space
	-- too much moving of the heads ...

- ask 100 people the same questions .. you will get 100 different replies
  for where to put swap and why ... ( experimental data vs "i heard in
  the grapevine" )
 
> so I hope 500 MB is o.k??? or is it more??? :-)

zero or all of the disk for "user"

the system should be able to clean itself
	- log rotates in pariticular

> 
> last time I did't check but what would you suggest?? but I've noticed
> increase upto 100 MB sometimes in a matter of two days.

100MB/day  --> and you want 5GB ---> that last you about days
assuming there is no log rotation and compression 

/var/log is temporary data

/var/cache or /var/apt is temporary or not .. depending on what you do
with the files

> Could you please explain this a little more to me?

nah ...
  
> Do you suggets I create a seperate /opt partition??? if so how much size??

i always use /opt ... and its "all the disks space" that the system doesnt
use
  
> I 've only used Ext3 till now because I heard data recover is
> difficult in all others except ext2.

you should NEVER have to recover data in the first place ..
	- something else is wrong if you do

	- if you like fiddling with inodes and meta-data, that'd be fun

> and also I understand that ReiserFS is better than XFS if the system
> goes down suddenly due to power failure. Don't know much about JFS :-(

"better" is all relative to what and where's the imperical data or source
code is what is say

"power failure problems" is pretty much the same problems for all fs

	- its a matter of how fast can you write your metadata 
	without corrupting the rest of the fs that was good prior to
	the power failing and you have say 2ms to fix it all up and clean
	up and exit before the disk write head goes bonkers

	- writing during a power failure is crazy, even if its to flush
	the cache

c ya
alvin



Reply to: