[TriLUG] Failure Copying Large Amounts of Data

boyle at laue.chem.ncsu.edu boyle at laue.chem.ncsu.edu
Fri Apr 16 13:51:14 EDT 2010



On 16 Apr, Glenn Starling wrote:
> Do you have sufficient swap file space for the copy program to use as it
> does it's work?

The system has 1GB of main memory and 2GB of swap, so I don't think that
is the problem.

To respond to Steve Litt:  only the backup drive is a USB drive the
other is IDE.

In my second post in this thread, I noted  that the 'df -i' command
indicates there were 0 available inodes. 

 Since that post, I ran
dumpe2fs on filesystem "A" (/dev/sdb1) and "C" (/dev/sdc2).
Excerpt from the output for "A" and "C" side by side:
		A		C
Inode count:		19546112		238592
Block count:		39072080		61059048
Free blocks:		14014352		44077885
Free inodes:              19202016				0
First block:		0		0
Block size:		4096		4096
Fragment size:		4096		4096
Blocks per group:		32768		32768
Fragments per group: 	32768			32768
Inodes per group:		16384		128
Inode blocks per group:   512				8

The places where C falls short are on "Inode count" (only 238K !),
"Inodes per group" ( 128 vs 16384) and "Inode blocks per group" (8 vs.
512).

The question is:  how to remake the file system so that I get a higher
number of inodes and inodes per group as well as inode blocks per group?


I am studying the mke2fs man page now and trying to do some web
research.  I remember reading several years ago (or more) about the
relationships between number of inodes and block size and other
filesystem parameters.

Thanks for the suggestions.  Filesystem configuration really isn't my
strong point, so I'll happily accept any other advice or tips.

Paul


-- 
Paul D. Boyle			    |	boyle at laue.chem.ncsu.edu
Director, X-ray Structural Facility |	phone: (919) 515-7362
Department of Chemistry - Box 8204  |	FAX:   (919) 515-8909
North Carolina State University     |	http://www.xray.ncsu.edu  
Raleigh, NC, 27695-8204		    |
PGP Key Fingerprint = D23A 0077 A1BF 277F 8140  FFD2 E4FA 4F99 677D 863E



More information about the TriLUG mailing list