[00:48] <magcius> mbiebl, is there no other easier way to get the jobs?
[21:03] <Keybuk> Writing /var/lib/ureadahead/sda1.pack
[21:03] <Keybuk> 508 files, 678 blocks (8425472 kB)
[21:03] <Keybuk> quest src% sudo ./ureadahead --debug                            
[21:03] <Keybuk> /var/lib/ureadahead/sda1.pack: created Fri, 23 Oct 2009 20:02:59 +0000 for hdd 8:1
[21:03] <Keybuk> \o/
[23:51] <wasabi__> Yeah Keybuk, how does readahead find the blocks that are most important?
[23:51] <wasabi__> s/yeah/hey?
[23:52] <Keybuk> mincore()
[23:53] <wasabi__> hmm
[23:54] <wasabi__> i'm thinking about making something similar, and running it all the time.
[23:54] <wasabi__> to identify most important blocks on LVM volumes.
[23:54] <wasabi__> and move those to a faster PV (SSD)
[23:54] <Keybuk> it's kinda hard to do it that way
[23:54] <wasabi__> It is, but the tools are present.
[23:55] <wasabi__> If I can just find blocks and map to PEs
[23:56] <wasabi__> Would love to add a 10GB SSD to my 500GB disk array, and automatically have the 10GB of data I use most on it
[23:57] <Keybuk> /sbin/sreadahead --no-fork  0.00s user 0.14s system 0% cpu 36.128 total
[23:58] <Keybuk> /sbin/ureadahead --verbose  0.01s user 0.17s system 2% cpu 7.718 total
[23:58] <Keybuk> ...
[23:58] <Keybuk> that, is, what, we, call, a, BIG, difference