On Mon, 2005-09-12 at 00:12 -0400, Peter Arremann wrote:
You'll find that a lot with Bryan that he makes up his own terms or puts his opinion down as facts.
Which is why you should feel free to believe I pull everything out of my asshole -- I've said that before, and I'll say it again.
Of course, it's often my documentation that gets referenced by other people -- Google searches, print publications, etc... and "explains" things. And it's _that_repeatable_, technical information year in and year out that earns my trust. I've only been posting here 4 months. In 4 years, you might feel differently.
But seriously - 1TB is just a good round number. Unlike the US tax code that always keeps you guessing, saying "No more than 1TB for ext3" is a quick and easy rule - and considering the fsck times you can get with a whole lot of small files 1TB is a good number.
It's a signed 32-bit integer for number of sectors. That results in a 1.1TB (1TiB) limitation. Ext3 has them all over the freak'n place in its codebase, although I haven't seen it on kernel 2.6.5+. In other words, I don't like to have Ext3 data volumes over 1TB because some systems just simply can't read them.
REAL WORLD EXAMPLE:
I previously ran into the issue where I created a 2TB Ext3 volume o a SAN device, and select versions/kernels could not use it. But the second I tried to mount a sub-1TB Ext3 filesystem from the same creator on the same system, I had no issue.
I have absolutely _never_ ran into that problem with XFS -- Irix, Linux 2.4, etc...
To sum it up, 1TB is no software limit but rather a number Bryan, me and a whole bunch of others (just search google) see as the maximum filesystem size they feel comfortable with.
Whatever you think it is, go on an answer for me. I'm used to you answering for me now. ;->