I've read the recommendations in the HOWTO, but none of them mention a proper
chunk/stride size for an SQL server with RAID5...
This machine will also be doing web serving (db intensive web app).
I will do some tests with mysqldump and other things...
Also, if I have a chunk size of 128k but many files under 128k, will they
all end up on one disk?
It says the chunk size is the amount of consecutive data on one disk, so I'm
not sure about this.
Most of our web files are under 100k, but the database is considerably
larger than that.
I just want to make sure I'm not doing it so wrong that it is slower than a
single drive (reading only of course).
And just to clarify (this might be useful in the options for mke2fs section
in the faq):
stride = chunksize/blocksize, correct?
Sorry to bother everyone again.
BTW, the problems I had with mkraid cleared up after I redid the kernel with
raid5 compiled into the kernel (not as a module).
Chris
