Maxim Wexler schrieb: > On 5/28/09, Volker Armin Hemmann <volkerar...@googlemail.com> wrote: >> On Donnerstag 28 Mai 2009, Florian Philipp wrote: >>> Maxim Wexler schrieb: >>>> Hi group, >>>> >>>> For a netbook 4G SSD. Attempting to install mozilla-firefox. jdk >>>> fails: No space left on device. >>>> >>>> df -i reveals no more inodes. I reboot thinking this will help. Wrong. >>>> Lots of 'No space left on device messages' with reference to >>>> /var/lib/iinit.d/* in the boot console. And this gem: '*ERROR: local >>>> is already starting'. And: '*ERROR: netmount is already starting'. >>>> [...] >>>> >>>> I know 4G is pretty small by today's standards but apart from xorg and >>>> firefox everything else on this unit is command-line type utilities >>>> and such. That can't account for 4G already. >>>> >>>> Maxim >>> That you run out of inodes doesn't mean that you run out of physical (or >>> logical) space on your disk. It just means that you run out of what you >>> could call file descriptors. >>> >>> There is exactly one inode per file which stores meta information about >>> this file. Ext2-4 have a fixed amount of inodes set when you format the >>> partition. Reiserfs and JFS create them on the fly and therefore don't >>> have problems with running out of inodes or wasting space on unused ones. >>> >>> Most likely you have a bunch of very small files on our disk, for >>> example the portage tree. These don't consume much space but a lot of >>> inodes. >>> >>> My advice: Save everything to another disk and then reformat the >>> partition with a higher amount of inodes. If you use ext2, format it with >>> >>> mke2fs -N 732960 /dev/sda2 >>> >>> This will create a file system with three times as many indoes as you >>> had before. >>> >>> Hope this helps. >> or don't use extX. >> >> > Ok, thanks everybody, getting ready to dive in and fix this thing. Two > more questions please: > [...] > > What's the best fs for a 4G SSD? I picked ext3 because of another eee > forum post. > > Maxim >
I just want to point to three blog posts from Theodore Ts'o: Partioning scheme and formatting tricks for optimal performance: http://thunk.org/tytso/blog/2009/02/20/aligning-filesystems-to-an-ssds-erase-block-size/ Talk about some general issues (ATA TRIM, mostly): http://thunk.org/tytso/blog/2009/02/22/should-filesystems-be-optimized-for-ssds/ Making an argument for using journalling filesystems: http://thunk.org/tytso/blog/2009/03/01/ssds-journaling-and-noatimerelatime/ Of course, T'so talks about an Intel X25-M which is a completely different beast from those cheap SSDs you find in netbooks. Delaying commits with ext4 and/or laptop-mode will reduce the wear-down of your SSD but it might as well freeze your system when the actual commit takes place because these things tend to have a terribly low write performance. In the end it will be a matter of playing with parameters.
