On Fri, Apr 4, 2014 at 6:16 PM, Stefan Seyfried <stefan.seyfried@googlemail.com> wrote:
That's completely unworkable for rotating media (they're *orders of magnitude* slower for random access).
Never ever put the persistant journal onto rotating rust. Just use syslog. Due to the genius design, the files fragment heavily and thrash the disk. Even plain "systemctl status foo.service" which looks into journal for logs will take 30+ seconds with journal on non-SSD.
Can you test that on rotating media please? A great deal of us are still using spinning rust for our storage needs, testing on SSD and calling it representative is probably as wrong as you can get.
Putting the journal on anything but SSD or RAM is as wrong as you can get :-)
Why not preallocate the space? That would get rid of fragmentation, and it should be a mostly-oneliner patch (call to posix_fallocate). For reading, posix_fadvise (or madvise) could read it whole (to avoid reading it in random fashion). Has this been tried? -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org