On Mon, 3 Dec 2001, Ken Ambrose wrote: > My company's RAID is running on a 2.2-kernel RFS for over a year, now, and > has worked fine with one (1) glitch in all that time: it got foobarred > when someone tried to create a larger-than-four-GB-file.
The only time ReiserFS has given me problems is when I tried to do an online resize of a filesystem beyond the end of the device, and it listened to me. The kernel was actually fairly willing to believe the disk array was four times larger than it was, but I imagine I had still created a fairly accurate simulation of /dev/null. reiserfsck saw it as a truncated filesystem and said I was screwed; resize_reiserfs saw it as a truncateed filesystem and refused to run. This is why backups are a good idea. :-/ > Try getting FS support like that out of M$ for under $200/hr! Try getting support like that out of MS, period. > The userland tools have improved in the past six months or so; Oh boy. I just started exploring the userland tools in the past six months or so! :-) > But ext2/3 have some inherent limitations that are going to become more > evident with time ... The only ones I am aware of are that it handles very large files and large directories poorly. Are there others? If not, I suspect ext2/3 will be around for a long time, since they meet the needs of "most people". > whereas most of ReiserFS's limitations deal with optimization, which (we > are assured) ReiserFS 4.x (due out late next year) are supposed to fix. I've said it before, and I'll say it again: Until I can actually buy (or download) a product, it is all vaporware to me. :-) > In the meantime, I'll take a 2.4 JFS any day. You can keep 2.4. I'll take the JFS, though. ;-) > ... with a regular, non-journaling filesystem, if you're writing a file > when your system crashes, *by definition* you have a corrupted file. It is worth stressing that (AFAIK) the same applies to ReiserFS. ReiserFS only journals metadata, not the file contents. The reiserfs-list gets messages now and again with someone whose file suddenly has blocks of garbage in it after a system crash. What ReiserFS *does* protect you from is having your entire filesystem, or a portion of it, scrambled due to a crash. Directory structure and such are intact; at most, files that were being written are damaged (and, as has been noted, they will likely be damaged anyway). With ext2, it is quite possible to lose an entire chunk of your filesystem if the crash occurs just as the kernel is (say) updating an allocation bitmap. -- Ben Scott <[EMAIL PROTECTED]> | The opinions expressed in this message are those of the author and do not | | necessarily represent the views or policy of any other person, entity or | | organization. All information is provided without warranty of any kind. | ***************************************************************** To unsubscribe from this list, send mail to [EMAIL PROTECTED] with the text 'unsubscribe gnhlug' in the message body. *****************************************************************