operation not permitted on entropy file
rwmaillists at googlemail.com
Mon Aug 11 16:50:23 UTC 2014
On Mon, 11 Aug 2014 17:16:53 +0200
> On Mon, 11 Aug 2014 09:52:00 -0500, Adam Vande More wrote:
> > Try fsck'ing a nearly full TB FS on a production box that has had a
> > dirty unmount and you will begin to appreciate the adventure a bit
> > more.
> I prefer appreciating my precious data. :-)
> No, honestly: On production systems, this might be an issue
> (which should be resolved by appropriate countermeasures, as
> we're probably talking about mission-critical systems), but
> for home systems, it surely isn't that bad. I'm saying this
> from my very individual experience of having fsck work on
> a nearly full 1 TB disk several times a week, in worst cases,
> few times a day. And I don't even care that it takes more
> than 10 minutes, as it makes sure my data is safe.
It depends on your hardware, it used to take several hours on my last
PC. Admittedly it was testing sequentially with a full fsck (run from a
script that mounts my geli partitions) and it was cpu limited in
I later switched to gjournal but that was very slow because it doubles
the cpu cycles in geli on writes. So I tried SU+J, but that seemed to
be more unreliable than anything, so I turned-off the "journalling" and
went back to foreground fsck. Even with faster hardware that was a
pain, it was over half an hour for my 3TB, and to speed things up I
usually did the 2TB drives in separate terminals.
I'm now back on geli+gjournal, the speed isn't so much of a problem
with AES support in the CPU. I wouldn't go back.
More information about the freebsd-questions