"Mustafa A. Hashmi" <mahashmi@gmail.com> writes:> All: > > Our test cluster is approximately 82GB. Current disk usage shows up as: > > Used Avail > 82G 1.3G 76G 2% /lustre > > That''s a little odd right there... 1.3GB in use and 76GB available?+5% reserved for root? MfG Goswin
All:
Our test cluster is approximately 82GB. Current disk usage shows up as:
Used Avail
82G 1.3G 76G 2% /lustre
That''s a little odd right there... 1.3GB in use and 76GB available?
If I however run a du -Hs /lustre, it shows only 459 MB.
There is a single oss which serves two osts (both on the same server)
with stripe_per_obj set to 1.
Both osts are approx 40 GB. Failover configuration applies with
replication in place via drbd.
Everything is running extremely smoothly, I am just curious as to how
usage is being calculated.
If someone could please shed some light on this.
Please note: Lustre reports everything healthy and nothing seems out
of place here. If it helps I can post our script which generates the
xml configuration file.
Thank you,
--
Mustafa A. Hashmi
mahashmi@gmail.com
mh@stderr.net
On Jun 15, 2006 12:15 +0200, Goswin von Brederlow wrote:> "Mustafa A. Hashmi" <mahashmi@gmail.com> writes: > > > All: > > > > Our test cluster is approximately 82GB. Current disk usage shows up as: > > > > Used Avail > > 82G 1.3G 76G 2% /lustre > > > > That''s a little odd right there... 1.3GB in use and 76GB available? > > +5% reserved for root?And a 400MB journal for each OST. This is in the FAQ or Knowledge Base I believe. Cheers, Andreas -- Andreas Dilger Principal Software Engineer Cluster File Systems, Inc.
Andreas Dilger <adilger@clusterfs.com> writes:> On Jun 15, 2006 12:15 +0200, Goswin von Brederlow wrote: >> "Mustafa A. Hashmi" <mahashmi@gmail.com> writes: >> >> > All: >> > >> > Our test cluster is approximately 82GB. Current disk usage shows up as: >> > >> > Used Avail >> > 82G 1.3G 76G 2% /lustre >> > >> > That''s a little odd right there... 1.3GB in use and 76GB available? >> >> +5% reserved for root? > > And a 400MB journal for each OST. This is in the FAQ or Knowledge Base > I believe. > > Cheers, AndreasDoesn''t the journal show up as used? Maybe I rember that wrong from when jorunals weren''t even hidden proberly by ext2/3 but df on a new fs used to show the journal as used. MfG Goswin
On 6/15/06, Andreas Dilger <adilger@clusterfs.com> wrote:> On Jun 15, 2006 12:15 +0200, Goswin von Brederlow wrote: > > "Mustafa A. Hashmi" <mahashmi@gmail.com> writes: > > > > > All: > > > > > > Our test cluster is approximately 82GB. Current disk usage shows up as: > > > > > > Used Avail > > > 82G 1.3G 76G 2% /lustre > > > > > > That''s a little odd right there... 1.3GB in use and 76GB available? > > > > +5% reserved for root? > > And a 400MB journal for each OST. This is in the FAQ or Knowledge Base > I believe.I see. Thank you all. Regards, -- Mustafa A. Hashmi mahashmi@gmail.com mh@stderr.net