On Fri, Nov 21, 2003 at 07:33:31AM +0100, Spider wrote: > > > What is the justification for this putdown of XFS? I communicate with > > other groups, and almost universally people sing the praises of XFS > > (including use for critical servers). Is this just an old wives tale > > from the early days of XFS? Did someone put this in and it was never > > reviewed? Almost anyone I talk to maintains that XFS recovers from > > power failures just as well as any other journaling fs. > > No, The problem is in how XFS caches content and restores metadata but > fills content with null (^@ ) whenever it goes down. > > We had XFS on our main distribution server and it fucked us over deeply > and badly due to this, by chewing in most of the data at a point. This > was bad enough to warrant the change. Before that XFS was recommended > above others for its performance. > > //Spider > wizend old and with bad memory.
that was well over a year ago wasn't it? what about todays XFS implementations in 2.4.22+patches or 2.6.0-test? This is why filesystem discussions are so bad. I know people who avoid reiserfs because it ate their disk in 1998. reiserfs has been stable for 3 years now. What should really be discussed is the particular implementation (fs driver version & kernel version), not the filesystem itself (which is really just the on disk data format). I've used XFS on my laptop for 18 months without problems because it was recommended at install time and I wanted to try something I had never used on linux before but my old laptop doesn't count as a high volume heavy load server. It has survived all battery failures or buggy driver induced crashes i've put it through without a single hiccup. -g -- [EMAIL PROTECTED] mailing list
