Messages in this thread | | | Date | Fri, 11 Dec 1998 19:25:46 +0000 (GMT) | From | Riley Williams <> | Subject | Re: atomicity |
| |
Hi Ted.
>>> With ext2fs you should never need a defragmenter
>> What does this mean? Is 22.4 % insignificant or should I be >> worried?
>> /dev/md1: 2407/78624 files (22.4% non-contiguous), 185079/313260 >> blocks
> A high non-contiguous percentage doesn't necessary mean that your > filesystem is highly fragmented. Files which are larger than the > maximum number of data blocks available in a block group are > guaranteed to be non-contiguous, by the very nature of the ext2 > filesystem.
> Hence, I was very hesitant when I included that metric in e2fsck, > since it could be very easily misused and misunderstood. The > problem is that doing a better metric is rather difficult.
> So for example, in the above example, there are 2407 files taking > up 185 megabytes, for an average size of 77k a file. Given that > there is usually a large number of small files (i.e., less than > 10k), there is likely a goodly number of large files which are much > more likely to overflow the confines of the block group and thus > get counted as a "non-contiguous" file. But more simply, there's a > big difference between a file which has the following allocation > pattern:
> Blocks 10-8191, 8202-6000
Is that latter group valid?
> and ....
> Blocks 10-20, 30-56, 60, 64-66, 100-140, 142-143, 150-200, ....
> Both of the above two files are counted as a single > "non-contiguous" file. But obviously, there's a very big difference > as to the fragmentation of that particular file.
Perhaps a better metrics would be: Number of files containing multiple partial block groups.
Best wishes from Riley.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |