On Fri, 2011-05-06 at 10:21 +0530, Rahul Sundaram wrote: > On 05/06/2011 06:08 AM, John Mellor wrote: > > > > What can I look at to determine where in the boot sequence that the > > problem bottleneck is, and how do I do it? > > > > systemd-analyze > > http://0pointer.de/blog/projects/blame-game.html Ok, now I have a couple of unsettling numbers: udev-settle.service is taking an incredible 57.670 seconds to do its probes. This is a pretty simple machine, so its somewhat unclear how it could possibly mess up that badly. 5 of the 6 SATA ports are not even connected. Is there any way to see what it thinks that it is probing that is so expensive? ntpdate.service is the next heaviest candidate, taking only 16.281 seconds. Presumably, this would be mostly due to my rogers.com uplink having a really crummy dns resolver, and most of the default 4 indirects to fedora.pool.ntp.org are to not very well known hosts (eliminating most dns cache effects). I can probably improve this a lot just by not using rogers as my dns resolver. cups.service is taking 6.373 seconds, presumably because I have not configured a printer yet on F15. Everything else is anywhere from 3.5 seconds (NetworkManager.service) to 12msec (rc.local.service), and so are probably not contributing any significant amount of extra delay to the ridiculous 1-minute udev-settle time. -- test mailing list test@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe: https://admin.fedoraproject.org/mailman/listinfo/test
