Richard Fish, mused, then expounded: > > Also, I didn't mean to deride xfs or xfs_fsr/fsr_xfs, so please don't > take it personally.
It's just a problem with perception. Say "defragger" and everyone thinks that Norton Utilites' disk utils have been cloned for Linux. I didn't take it personally, but didn't want others to assume that space compaction would be part of xfs_fsr. > It is the *only* filesystem that offers a online > defrag tool for linux today, and that is a big bonus. My complaint is > more of a wish list than anything else. But it isn't a big enough > wish for me to spend time working on it myself! > Well, it does also offer a fairly painless partition copy/restore, which actually works very well, if there is an extra drive around. > > But actually, it is with the handling of multi-gigabyte files that I > find it lacking. My VMWare virtual disk images are 10-20G. So if I > don't have at least 20G of free space on a filesystem, I cannot > defragment those. Moreover, since xfs_fsr doesn't consolidate free > space, it is probable that no improvements could be made to those > files even if I have 50G free, since there are likely to be some files > spread out over the disk if the filesystem has been in use for any > length of time. > True enough. Though the original customers that xfs_fsr was written for were running multi-terabyte arrays where a 20 GB file was a small single texture. Even those running HDTV editing on a 1P box were working with 300 GB of video (15 minutes). Thus they typically had the need for 3x the space during editing, but had plenty of free space for defragging the file. Thus the target audience were those with attached raid/jbod arrays. But again, I was just attempting to clarify vs. being offended. Apologies if it came across that way. Bob - -- [email protected] mailing list
