nfs-kernel-server vs nfs-user-server
I recently upgraded my machine, replaced 4×60gb drives with 4×320GB drives, the 60gb drives were RAID'ed together using the onboard HPT370 chipset, while it was a rather nice arrangement with the 60gb drives, they were slow, old, small and two of them were predicting SMART failure.
setting up the new drives was non-trivial. The HPT370 bios does not support LBA48, so I was tried to use the debian sarge cd and RAID (tried LVM too) but neither would work (they failed on formatting the drive, assuming that i used anything more than a single partition, if i used a single partition they would fail installing grub or lilo. It was quite frustrating. I ended up just formatting all four drives seperately, since I'm too cheap to use RAID5 this way if one drive has an unexpected demise I only loose one drive of data, not all four. In the end, this means I have about 2200GB (2.2TB) of space on that system now, not too shabby. But, back to the impetus for this post.
I was having an issue with NFS, the drives are mounted something like this:
/home /home/files/treeX
at the moment there is tree1-tree8. This is good because of the way that the internal script manages moving files around, everything is nicely organized by a set of scripts and a MySQL backend.
From this, /home/movies is mounted to a 2nd machine.
For some reason I had the issue where nfs utterly refused to see content in directories that were different mounted directories.
So, on the 2nd machine (or even if i mounted it on my laptop) cd to /home/movies/tree2 and the directory is empty.
Not quite sure why this is, but I recalled having this issue when I first setup the system some years back.
Reading the debian package listing for nfs-kernel-server showed me this unhelpful piece of info:
The user-mode NFS server in the "nfs-user-server" package is slower but more featureful and easier to debug than the kernel-mode server.
the nfs-user-server had nothing useful on it, I figured I would try switching from the kernel-mode to the user-mode server.
Well, after a few issues (don't try switching from kernel-mode to user-mode or the reverse while remote directories are mounted :P) it worked. I was now able to properly browse all subdirectories, which is great, because it means that what would have been a list of 8 mounts is now reduced to one, which, well, must be good for some reason, at least I hope so or else this entire entry is entirely in vain.
Just a small tidbit, when I purchased the 60GB drives that were in use in this system several years ago, they cost me 300$ per drive. that's 5$ per gigabyte. The 4 320GB drives I purchased cost 122$ each, that's about 38 cents per gigabyte. Granted, at the time 60GB drives were the largest avaliable, whereas 500GB drives take that spot now and they cost considerably more than 122$ (357.98$ to be exact) but at 71 cents per gigabyte they are almost twice the per-gigabyte price of the 320GB drives. I suppose that the next step is a 9550SX-16ML.
This is all running on a venerable KT7A-RAID which, at the time was an absolutely amazing motherboard (can I hear four IDE channels).
No Comments »
No comments yet.
RSS feed for comments on this post. TrackBack URI
Leave a comment
Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>